Language Mind Culture Societyfiles.clickweb.home.pl/34/60/346047fd-5467-4c0e-a260-17ef4bfa99ea.pdfsical layer of the aria Vissi d’Artez from Puccini’s Tosca. Her study serves as

LanguageMindCultureSociety

nr 3 2019ISSN 2544-6886

Editor-in-Chief / Redaktor NaczelnyAgnieszka Libura, University of Wrocław / Uniwersytet Wrocławski

Editorial Board / Komitet RedakcyjnyMałgorzata Fabiszak, Adam Mickiewicz University in Poznań / Uniwersytet im. Adama Mickiewicza w PoznaniuBarbara Lewandowska-Tomaszczyk, State University of Applied Sciences in Konin / Państwowa Wyższa Szkoła Zawodowa w KoninieAgnieszka Mikołajczuk, University of Warsaw / Uniwersytet WarszawskiMichał Szawerna, University of Wrocław / Uniwersytet WrocławskiElżbieta Tabakowska, Jagiellonian University in Kraków / Uniwersytet Jagielloński

Editorial Assistants / Sekretarze RedakcjiAgnieszka Mierzwińska-Hajnos, Maria Curie-Skłodowska University in Lublin / Uniwersytet Marii Curie-Skłodowskiej w LublinieMałgorzata Paprota, Maria Curie-Skłodowska University in Lublin / Uniwersytet Marii Curie-Skłodowskiej w LublinieJanusz Badio, University of Lodz/ Uniwersytet ŁódzkiReview Editors / Dział RecenzjiAdam Głaz, Maria Curie-Skłodowska University in Lublin / Uniwersytet Marii Curie-Skłodowskiej w LublinieKarolina Krawczak, Adam Mickiewicz University in Poznań / Uniwersytet im. Adama Mickiewicza w Poznaniu

Advisory Board / Rada NaukowaEnrique Bernardez, Complutense University of MadridAlan Cienki, Vrije Universiteit AmsterdamElżbieta Górska, Uniwersytet WarszawskiMartin Hilpert, University of NeuchâtelLaura A. Janda, University of TromsøMarlene Johansson-Falck, Umeå UniversityZoltán Kövecses, Eötvös Loránd UniversityGitte Kristiansen, Complutense University of MadridTorsten Leuschner, Ghent UniversityBrigitte Nerlich, University of NottinghamAnatol Stefanowitsch, Free University of BerlinHeli Tissari, Stockholm UniversityMark Turner, Case Western Reserve UniversityJordan Zlatev, Lund University

Cover design and layout / Projekt szaty graficznejMichał Wolski

Composed using the Alegreya and Alegreya Sans typefaces designed by Juan Pablo del Peral

Journal website / Witryna internetowa czasopismalamicus.eu

Paper mail / Adres redakcjiLamicusInstytut Filologii Polskiej UWrPlac Nankiera 1550-140 WrocławPoland

e-mail: [email protected]

© Copyright by Polish Cognitive Linguistics Association (Polskie Towarzystwo Językoznawstwa Kognitywnego), 2019

ISSN 2544-6886

48

12

40

60

84

108

128

136

142

148

158

166

Table of contents

EditorialOd Redakcji

Articles / ArtykułyHubert Kowalewski

Małgorzata Misiak

Weronika Drosdowska

Agnieszka Mierzwińska-Hajnos

Francesca Carbone

DebateKarolina Krawczak

– Małgorzata Fabiszak

Barbara Dancygier

Johanna Kissler

Dylan Glynn

Karolina Krawczak – Małgorzata Fabiszak

Reviews Renata Przybylska,

spistreści

Structural empiricist Cognitive Grammar

Co to znaczy być Łemkiem? W kręgu pytań tożsamościowych. Ujęcie etnolingwistyczneMultimodal construction of fear in a TV se-ries – American Horror Story: Cult – Verbal and visual modeWhere music and text meet: A multimodal analysis of Vissi d’Arte ariaIdentification of emotions by prosody and se-mantics in French spontaneous speech: A pilot study

Introduction / Wstęp

The status of multimodal constructions in de-veloping models and empirical testing – a multimodal semiotic perspectiveThe status of multimodal constructions in de-veloping models and empirical testing – a neuro-cognitive perspectiveThe status of multimodal constructions in de-veloping models and empirical testing – a methodological perspectiveOverview / Podsumowanie

Krystyna Waszakowa 2017: Kognitywno-komu-nikacyjne aspekty słowotwórstwa. Wybrane

Andreas Baumann

Martin Hilpert

Adam Głaz

zagadnienia opisu derywacji w języku polskim. Warszawa: Wydział Polonistyki Uniwersytetu Warszawskiego. 230 s. ISBN: 978-83-65667-44-1William H. Edmondson 2017: The Sequential Imperative. General Cognitive Principles and the Structure of Behaviour. (Value Inquiry book series) Leiden, Boston: Brill Rodopi. ISBN: 978-90-04-34289-7 (Pb), 978-90-04-34299-6 (E-book)Goldberg, Adele E. 2019: Explain Me This. Creativity, Competition, and the Partial Productivity of Con-structions. Princeton: Princeton University Press. 195 pp. ISBN-13: 978-0-691-17426-6Bert Peeters (ed.) 2019: Heart- and Soul-Like Con-structs across Languages, Cultures, and Epochs. (Routledge Focus: Routledge Studies in Linguistics.) New York and London: Routledge. viii, 148 pp. ISBN: 978-138-74530-8 (Hb), 978-1-315-18067-0 (E-book)

171

177

185

4 L aMiCuS 2019 no. 3

Editorial

Volume 3 of LaMiCuS continues the research line outlined in the previous two volumes. It contains articles that follow the cognitive linguistic tradi-tion as originally developed in the United States, as well as those that rely on the Lublin School of Ethnolinguistics. Two contributions offer a cogni-tive approach to multimodality, one draws on psycholinguistic approach to communicating and understanding emotions.

Hubert Kowalewski analyses the premises and practice of Langacker’s (1987, 2008, 2016, 2017) cognitive grammar, which he approaches from the perspective of the philosophy of science, in particular, structural empiricism (Van Fraassen 1980, 2002, 2008). Kowalewski shows that, within this perspec-tive, theoretical concepts should be empirically adequate, but do not need to be directly observable in empirical data. Graphic representations enjoy a similar status, as they constitute mathematically formalized theoretical claims. The use of invented examples in theory modelling is justified as em-pirical testing is the next step of theory development.

The article by Małgorzata Misiak follows the Lublin School of Ethnolin-guistics (cf. Bartmiński 2012, Niewiara 2009, Tambor 2008) and focuses in Lemko identity. The study is based on an analysis of ethnographic interviews, which shows that Lemko self-image can be described in terms of four main profiles: Lemko – a victim of history, Lemko – an exile, an ethnic and cultural Lemko, and Lemko – a peasant. These profiles emerge in the context of the multidimensional opposition SELF – OTHER – ALIEN and the binary opposition US – THEM.

Weronika Drosdowska and Agnieszka Mierzwińska-Hajnos continue the focus on multimodality, the key theme of LaMiCuS , volume 2. Drosdows-ka’s study combines the theory of metaphor and metonymy originating with Lakoff and Johnson (2003) with social semiotics as proposed by Kress (2010) to analyze the still under-investigated genre of TV series. The starting point of her project is the list of metaphors and metonymies of fear as proposed

5Editorial

by Kövecses (1990, 2004, 2008). Drosdowska shows that there are some dif-ferences in how these metaphors are realized in the verbal and visual mode. For example, verbal metaphors allow for a more abstract understanding of fear (fear is currency), while in the visual mode they seem more concrete (fear is a trap). This does not mean, however, that the filmic mode does not allow for the construction of the sense of uncertainty, which illustrates diffused, unidentified sources of fear (e.g. filming an escape with a hand held camera as a realization of fear is a tormentor metaphor).

In her article, Agnieszka Mierzwińska-Hajnos approaches a challeng-ing task of analyzing selected modalities (verbal and musical) of an operatic piece (cf. Kuhl 2007), within the framework of conceptual integration theory as proposed by Brand and Brand (2005). She focuses on the textual and mu-sical layer of the aria Vissi d’Artez from Puccini’s Tosca. Her study serves as an example of how conceptual integration theory allows researchers to analyze each mode separately to finally combine them in the blend.

The study by Francesca Carbone has been conducted within the experi-mental psycholinguistics paradigm and, similarly Drosdowska’s contribution, concerns emotions. It does not, however, explore the image of a particular emotion in a work of fiction, but rather the perception and recognition of emotions in spontaneous speech in experimental conditions. The stimulis came from emotionally loaded TV interviews from the French EmoTV corpus. The results show that emotions are recognized more accurately and faster in the case of congruent multimodal stimuli (cf. Collignon et al. 2008; Pell 2005; De Gelder & Vroomen 2000; Massaro & Egan 1996), i.e., when the listeners have access to both semantic and prosodic information.

In this 3rd volume of LaMiCuS we propose a new form of scholarly ex-change: the Debate. This section of our journal will give f loor to discussants at PCLA conferences, where a discussion panel on specific themes will from now on be, hopefully, regular practice. Out first debaters are Barbara Dan-cygier, Johanna Kißler, and Dylan Glynn : they discuss the nature of mul-timodal communication and approaches to analyzing it. We hope that this section, in a lighter form, will enhance the impact of cognitive linguistics on students and young researchers, who can then pursue specific issues in full-f ledged academic discourse.

The Reviews section contains one review in Polish and three in English. First, Renata Przybylska discusses Krystyna Waszakowa’s monograph Kog-nitywno-komunikacyjne aspekty słowotwórstwa, whose author approaches Pol-ish word-formation from the perspective of Langacker’s Cognitive Grammar (and, to an extent, Fauconnier and Turner’s Mental Spaces Theory), capitaliz-ing at the same time on the findings of structural linguistics. Then Andreas Baumann reviews William Edmondson’s proposal in the latter’s The Sequen-tial Imperative, in which Edmondson seeks correlations between grammar,


linguistic aspects of cognition, and the functional specialization of the brain – and all in the context of evolution. Next, Martin Hilpert offers his assess-ment of Adele Goldberg’s intriguing Explain Me This, a monograph devoted to construction grammar approach to grammaticality judgments. Finally, Adam Głaz looks at a publication with a significant cultural component and couched within Anna Wierzbicka’s NSM framework: a concise volume edited by Bert Peeters, titled Heart- and Soul-Like Constructs across Languages, Cultures, and Epochs. We hope that the reviews will throw useful light on the content and level of the broad range of publications being discussed.

Małgorzata Fabiszak, Karolina Krawczak, Adam GłazVolume editors

ReferencesBartmiński, Jerzy 2012: Zmiany obrazu Polaka we wspołczesnym dyskursie

publicznym. In: Jezykowe podstawy obrazu swiata. Lublin: Wydawnictwo Uniwersytetu Marii Curie Skłodowskiej, 208-225.

Brandt, Line, Per Aage Brandt 2005: Making sense of a blend: A cognitive-se-miotic approach to metaphor. Annual Review of Cognitive Linguistics 3, 216-249.

Collignon, Oliver, Frederic Gosselin, Sanjoy Roy, Dave Saint-Amour, Maryse Lassonde, Franco Lepore 2008: Audio-visual integration of emotion expression. Brain Research 1242, 126-135.

De Gelder, Beatrice, Jean Vroomen 2000: The perception of emotions by ear and by eye. Cognition & Emotion 14, 3, 289-311.

Gorska, Elzbieta 2018: From music to language and back. Language, Culture, Mind and Society 2, 82-103.

Kovecses, Zoltan 1990: Emotion Concepts. New York: Springer-Verlag. Kovecses, Zoltan 2004: Metaphor and Emotion: Language, Culture, and Body in

Human Feeling. Cambridge: Cambridge University Press.Kovecses, Zoltan 2008: On metaphors of emotion: A reply to Ayako Omori.

Metaphor and Symbol 23(3), 200-203.Kress, Gunther R. 2010: Multimodality: A Social Semiotic Approach to Contempo-

rary Communication. London: Routledge.Kuhl, Ole 2007: Musical Semantics: Cognitive Musicology and the Challenge of

Musical Meaning. Bern: Peter Lang.Langacker, Ronald W. 1987: Foundations of Cognitive Grammar. Stanford: Stan-

ford University Press.Langacker, Ronald W. 2008: Cognitive Grammar. A Basic Introduction. New

York: Oxford University Press.

7Editorial

Langacker, Ronald W. 2016: Nominal Structure in Cognitive Grammar: The Lublin Lectures. Adam Głaz, Hubert Kowalewski, Przemysław Łozowski (ed.). Lublin: Maria Curie-Skłodowska University Press.

Langacker, Ronald W. 2017: The Function of Trees. In: Przemysław Łozowski, Adam Głaz (eds.) 2017: Route 66: From Deep Structures to Surface Meanings: A Festschrif t for Henryk Kardela on His 66th Birthday. Lublin: Wydawnictwo Uniwersytetu Marii Curie-Skłodowskiej, 73-91.

Lakoff, George, Mark Johnson. 2003. Metaphors We Live by. London: The Uni-versity of Chicago Press.

Massaro, Dominic W., Peter B. Egan 1996: Perceiving affect from the voice and the face. Psychonomic Bulletin & Review 3, 2, 215-221.

Niewiara, Aleksandra 2009: Kształty polskiej tozsamosci. Katowice: Wydawnictwo Uniwersytetu Slaskiego.

Pell, Marc David 2005: Prosody–face interactions in emotional processing as revealed by the facial af fect decision task. Journal of Nonverbal Behavior 29, 4, 193-215.

Tambor, Jolanta 2008: Mowa Górnoslazaków oraz ich swiadomosc jezykowa i et-niczna, 2nd ed. Katowice: Wydawnictwo Uniwersytetu Slaskiego.

Van Fraassen, Bas C. 1980: The Scientific Image. Oxford: Clarendon Press. Van Fraassen, Bas C. 2002: The Empirical Stance. New Haven: Yale University

Press.Van Fraassen, Bas C. 2008: Scientific Representation: Paradoxes of Perspective.

Oxford-New York: Oxford University Press.


Od Redakcji

Oddajemy w Państwa ręce trzeci numer LaMiCuSa, który kontynuuje proble-matykę podjętą we wcześniejszych rocznikach. W tym numerze publikujemy zatem artykuły należące zarówno do nurtu językoznawstwa kognitywnego mającego swoje korzenie w tradycji amerykańskiej, jak i odwołujące się do Lubelskiej Szkoły Etnolingwistyki czy też łączące podejście kognitywne i badania nad multimodalnością oraz wywodzące się metodologicznie z psy-cholingwistycznych badań nad przekazywaniem i odczytywaniem emocji.

Artykuł Huberta Kowalewskiego analizuje założenia i praktykę grama-tyki kognitywnej Langackera (1987, 2008, 2016, 2017) z perspektywy filozofii nauki, a konkretnie empiryzmu strukturalnego (Van Fraassen 1980, 2002, 2008) i wskazuje, że w ramach tego podejścia pojęcia teoretyczne powinny wykazywać się adekwatnością empiryczną, nie są natomiast bytami po-strzegalnymi bezpośrednio w danych empirycznych. Podobny status mają reprezentacje graficzne, które stanowią matematyczne formalizacje twier-dzeń teoretycznych. W prezentowaniu teorii, jak podkreśla Kowalewski, uzasadnione jest również używanie przykładów skonstruowanych przez autora, gdyż badania empiryczne to dopiero kolejny krok w testowaniu za-łożeń teoretycznych.

Kolejny artykuł, utrzymany w duchu etnolingwistyki lubelskiej (por. Bartmiński 2012, Niewiara 2009, Tambor 2008),to praca Małgorzaty Misiak na temat tożsamości Łemków. W swoich badaniach opartych na danych z wywiadów etnograficznych Autorka pokazuje, że obraz własny Łemków układa się w 4 główne profile: Łemko – ofiara historii, Łemko wygnaniec, Łemko etniczno-kulturowy i Łemko – chłop. Profile te kształtują się w kontekście wielokierunkowej opozycji SWÓJ – INNY – OBCY oraz opozycji binarnej MY – ONI.

Artykuły Weroniki Drosdowskiej i Agnieszki Mierzwińskiej-Hajnos kontynuują tematykę badań nad multimodalnością, której poświęcony był numer drugi LaMiCuSa. Praca Drosdowskiej łączy teorie metafory i meto-

9Od Redakcji

nimii konceptualnej zapoczątkowanej przez Lakoffa i Johnsona (2003) oraz semiotyki społecznej Kressa (2010) do analizy słabo dotychczas zbadanego materiału z serialu telewizyjnego. Wychodząc od listy metafor i metoni-mii strachu zaproponowanej przez Kövecsesa (1990, 2004, 2008), Autorka wykazuje, w jaki sposób realizacja niektórych z nich różni się w warstwie werbalnej i wizualnej. Na przykład metafory werbalne pozwalają na nieco bardziej abstrakcyjne rozumienie strachu (strach to waluta), natomiast w warstwie wizualnej są one bardziej ukonkretnione (strach to pułapka). Nie oznacza to jednak, że środki filmowe nie pozwalają na wytworzenie at-mosfery niepewności, ilustrując rozproszone, nie do końca rozpoznawalne źródła lęku (np. filmowanie ucieczki z ręki jako realizacja metafory strach to dręczyciel).

Artykuł Agnieszki Mierzwińskiej-Hajnos stawia czoła wyzwaniu, jakim jest analiza wybranych modalności (werbalnej i muzycznej) dzieła operowego (por. Kuhl 2007). W tym celu Autorka stosuje model integracji pojęciowej za-proponowany przez Brandt i Brandta (2005). Skupia się na warstwie teksto-wej i muzycznej arii Vissi d’Artez opery Tosca Pucciniego i na tym przykładzie pokazuje, w jaki sposób badania w ramach teorii amalgamatów pojęciowych pozwalają najpierw na przeanalizowanie każdej z modalności, a następnie na zintegrowanie ich strategii tworzenia znaczeń w przestrzeni amalgamatu.

Ostatnia w dziale artykułów naukowych praca Francesci Carbone wy-wodzi się z nurtu psycholingwistycznych badań eksperymentalnych i – po-dobnie jak artykuł Weroniki Drosdowskiej – poświęcona jest kwestii emocji. Nie dotyczy jednak ich obrazowania w utworze fikcyjnym, jakim jest serial telewizyjny, ale ich percepcji i rozpoznawaniu w warunkach eksperymen-talnych. Za bodźce posłużyły wycinki z wywiadów telewizyjnych na tematy o silnym ładunku emocjonalnym (tzw. francuski korpus EmoTV). Wyniki badań wskazują, że emocje rozpoznawane są dokładniej i szybciej w przy-padku zintegrowanych bodźców zgodnych (por. Collignon et al. 2008; Pell 2005; De Gelder & Vroomen 2000; Massaro & Egan 1996), tzn. – w przypadku tego eksperymentu – kiedy słuchaczom dostępna jest zarówno informacja semantyczna, jak i prozodyczna.

W tym numerze czasopisma wprowadzamy również nowy dział pod na-zwą „Debata”, w którym publikowane będą wypowiedzi gości zaproszonych na konferencje organizowane pod egidą PTJK. W pierwszej debacie biorą udział Barbara Dancygier, Johanna Kißler i Dylan Glynn, którzy zastanawiają się, czym są zjawiska komunikacji multimodalnej i w jaki sposób należy je ba-dać. Mamy nadzieję, że ta forma popularyzacji zagadnień naukowych, która pozwala przekazać istotne treści w nieco lżejszej formie, ułatwi studentom i początkującym naukowcom wejście w świat dyskursu ściśle akademickiego.

Jak zwykle bogaty jest dział recenzji – Czytelnik otrzymuje jedną recen-zję w języku polskim i trzy w języku angielskim. Najpierw Renata Przybylska


omawia monografię Krystyny Waszakowej Kognitywno-komunikacyjne aspekty słowotwórstwa, której Autorka wykorzystuje inspiracje płynące z gramatyki kognitywnej Ronalda Langackera (i częściowo z teorii przestrzeni mental-nych Fauconniera i Turnera), nie rezygnując wszakże z osiągnięć struktu-ralizmu. Następnie Andreas Baumann dokonuje oceny propozycji Williama Edmondsona zawartej w książce The Sequential Imperative, w której Autor poszukuje zależności między gramatyką, językowymi aspektami poznania i funkcjonalną specjalizacją mózgu, a wszystko w kontekście ewolucji gatun-ków. W dalszej kolejności Martin Hilpert omawia monografię Adele Goldberg o prowokującym tytule Explain Me This, poświęconą gramatyce konstrukcji, a dokładniej sądom co do gramatyczności wyrażeń dokonywanych przez użytkowników języka. W końcu Adam Głaz zajmuje się pozycją, w której znaczącą rolę odgrywa aspekt kulturowy – jest to zredagowany przez Berta Peetersa niewielki tom pt. Heart- and Soul-Like Constructs across Languages, Cultures, and Epochs, zawierający analizy przeprowadzone w ramach modelu Naturalnego Metajęzyka Semantycznego Anny Wierzbickiej. Mamy nadzieję, że recenzje rzucają cenne światło na treść i poziom omawianych publikacji, a różnorodność tych drugich znajdzie uznanie w oczach Czytelników.

Małgorzata Fabiszak, Karolina Krawczak i Adam GłazRedaktorzy tomu

BibliografiaBartmiński, Jerzy 2012: Zmiany obrazu Polaka we współczesnym dyskursie

publicznym. W: Jezykowe podstawy obrazu swiata. Lublin: Wydawnictwo Uniwersytetu Marii Curie Skłodowskiej, 208-225.

Brandt, Line, Per Aage Brandt 2005: Making sense of a blend: A cognitive-semiotic approach to metaphor. Annual Review of Cognitive Linguistics 3, 216-249.

Collignon, Oliver, Frederic Gosselin, Sanjoy Roy, Dave Saint-Amour, Maryse Lassonde, Franco Lepore 2008: Audio-visual integration of Emotion expression. Brain Research 1242, 126-135.

De Gelder, Beatrice, Jean Vroomen 2000: The perception of emotions by ear and by eye. Cognition & Emotion 14, 3, 289-311.

Górska, Elżbieta 2018: From music to language and back. Language, Culture, Mind and Society 2, 82-103.

Kövecses, Zoltan 1990: Emotion Concepts. New York: Springer-Verlag. Kövecses, Zoltan 2004: Metaphor and Emotion: Language, Culture, and Body in

Human Feeling. Cambridge: Cambridge University Press.Kövecses, Zoltan 2008: On metaphors of emotion: A reply to Ayako Omori.

Metaphor and Symbol 23, 3, 200-203.

11Od Redakcji

Kress, Gunther R. 2010: Multimodality: A Social Semiotic Approach to Contempo-rary Communication. London: Routledge.

Kuhl, Ole 2007: Musical Semantics: Cognitive Musicology and the Challenge of Musical Meaning. Bern: Peter Lang.

Langacker, Ronald W. 1987: Foundations of Cognitive Grammar. Stanford: Stan-ford University Press.

Langacker, Ronald W. 2008: Cognitive Grammar. A Basic Introduction. New York: Oxford University Press.

Langacker, Ronald W. 2016: Nominal Structure in Cognitive Grammar: The Lublin Lectures. Adam Głaz, Hubert Kowalewski, Przemysław Łozowski (red.). Lublin: Maria Curie-Skłodowska University Press.

Langacker, Ronald W. 2017: The Function of Trees. W: Przemysław Łozowski, Adam Głaz (red.) 2017: Route 66: From Deep Structures to Surface Meanings: A Festschrif t for Henryk Kardela on His 66th Birthday. Lublin: Wydawnictwo Uniwersytetu Marii Curie-Skłodowskiej, 73-91.

Lakoff, George, Mark Johnson. 2003. Metaphors we Live By. London: The Uni-versity of Chicago Press.

Massaro, Dominic W., Peter B. Egan 1996: Perceiving affect from the voice and the face. Psychonomic Bulletin & Review 3, 2, 215-221.

Niewiara, Aleksandra 2009: Kształty polskiej tozsamosci. Katowice: Wydaw-nictwo Uniwersytetu Sląskiego.

Pell, Marc David 2005: Prosody–face interactions in emotional processing as revealed by the facial af fect decision task. Journal of Nonverbal Behavior 29, 4, 193-215.

Tambor, Jolanta 2008: Mowa Górnoslazaków oraz ich swiadomosc jezykowa i et-niczna, wyd. 2. Katowice: Wydawnictwo Uniwersytetu Sląskiego.

Van Fraassen, Bas C. 1980: The Scientific Image. Oxford: Clarendon Press. Van Fraassen, Bas C. 2002: The Empirical Stance. New Haven: Yale University




Hubert KowalewskiMaria Curie-Skłodowska University in Lublin

Structural empiricist Cognitive Grammar

The article sketches a structural empiricism approach to Ronald Langacker’s Cognitive Gram-mar (CG). In philosophy of science, structural empiricism is a view on the goals of science and adopting this approach helps to solve important problems in methodology of CG and linguistics by and large. The article analyses three such problems. The first one is the evidence required for certain theoretical postulates of CG. From the structural empiricist perspective, purely theoretical concepts postulated by CG do not require evidence apart from the overall empirical adequacy of the theory. The second problem is the usefulness of diagrams used by Langacker as illustrations accompanying his analyses. On the syntactic view of scientific theories (compatible with struc-tural empiricism), carefully designed diagrams can be mathematically rigorous formalizations of theoretical claims. The third problem is the legitimacy of constructed expressions in linguistic theorizing. In structural empiricism, constructed expressions can be treated similarly to idealiza-tions in natural sciences, i.e. a cognitive grammarian may use them fruitfully for model-making without committing herself to the belief in their authenticity.

Received: 25.10.2018. Reviewed: 07.09.2019. Accepted: 22.10.2019. Published: 31.12.2019.

1. Introduction

Many linguists rightly believe that the question of whether linguistics is a “real” science is sterile. The reason for this is not the shape or research prac-tices in modern linguistics, but the general feeling in philosophy of science

ABST

RACT

KeywordsCognitive Grammar, structural empiricism, philosophy of science

https://doi.org/10.32058/LAMICUS-2019-001

13Structural empiricist Cognitive Grammar

that the so-called demarcation problem, an attempt to provide a list of nec-essary and sufficient properties distinguishing science from non-science, cannot be solved in a satisfactory way. It is difficult to say whether linguistics is a science, because it is hard to specify what “standard” science is, so no solid reference point is available. Nonetheless, even if linguistics is not a “hard” science, we can still learn something from our colleagues in natural science departments by examining their practices and their solutions to epistemic and methodological problems. We do not need to do this for purely “idealistic” reasons, for example in order to decide whether linguistics “deserves” the honorary title of a natural science. We can do this for more opportunistic rea-sons: to see whether a philosophical account of practices in natural sciences can help us solve the problems we find in our discipline. In this spirit, this article offers an interpretation of Ronald Langacker’s Cognitive Grammar (henceforth CG, cf. e.g. Langacker 1987, Langacker 2008) along the lines of structural empiricism, a vision of science proposed by the philosopher Bas van Fraassen. More specifically, the aim of the article is to show how view-ing Cognitive Grammar through the lens of structural empiricism can help inform the debates on practical epistemic and methodological issues of the theory. It is open to debate whether similar conclusions are valid for other theories within Cognitive Linguistics. Given the multitude and diversity of the theories, a comprehensive philosophical discussion spanning the entire field would require a much larger publication. This is one of the reasons why the thematic scope of the article is limited specifically to Ronald Langacker’s Cognitive Grammar.

The first issue concerns the problem of evidence required for certain postulates made by Cognitive Grammar (and any other linguistic theory, for that matter). This problem is addressed straightforwardly in structural empiricism, which is the view that accepting a theory requires only the belief in the theory’s empirical adequacy rather than explicit evidence for every part of the theory. The second problem pertains to the role of diagrams in Ronald Langacker’s works. I partly disagree with Langacker that diagrams perform a merely heuristic role and I defend a stronger view: at least some of the diagrams are fully-f ledged, mathematically rigorous expressions of underlying theoretical models.

The final issue is more general and not limited to Cognitive Grammar. This is the problem of sources of data used for theory-making in linguistics, especially of the legitimacy of expressions constructed specifically for the purpose of theorizing. It appears that CG relies on constructed expressions to a large extent, so even though the problem is wider in scope, it applies to Langacker’s theory as well. My argument has two parts. Firstly, I investigate the alethic modal character of constructed expressions. Secondly, by drawing an analogy to research practices in natural sciences, I defend the usefulness


of constructed examples within the framework of structural empiricist lin-guistics. I do not analyze in detail the specifics of Cognitive Grammar and some familiarity with the theory on the part of the reader is assumed.

To the best of my knowledge, Ronald Langacker does not identify him-self as a structural empiricist. In my view, some of his claims can be easily and meaningfully interpreted along the lines of this approach, but it must be emphasized that these interpretations are my own and may not enjoy Lan-gacker’s endorsement. This is particularly worth remembering in the parts of the article where the conclusions are more radical than those Langacker seems to support (particularly in Section 4). Moreover, the conclusions of this article are not restricted to Cognitive Grammar, at least not in principle. The line of argumentation is in most parts fairly general and may be applied to other theories outside cognitive linguistics as broadly understood. I will, however, investigate the issues as they stand in CG and focus on the solu-tions proposed by Ronald Langacker (especially in Sections 3 and 4), partly because of my personal interest in this theory and the fact that I first came to be aware of them in the context of CG, and partly because a comprehensive discussion of other theories would require a much longer treatment.

2. Whence structural empiricism?In the context of linguistics, the perennial conf lict between two major views in epistemology, rationalism and empiricism, is evoked by Randy Harris in The Linguistics Wars (1993). While admitting that any slogan-like summaries of these complex views are bound to be grossly simplified, Harris tentatively defines them in the following way:

Empiricism: most knowledge is acquired through the senses. Rationalism: most knowledge is not acquired through the senses. (Harris 1993: 66)

The author mentions the two views in the context of “the linguistics wars” between Bloomfieldian structuralists (sympathetic to empiricism) and early Chomskyan generativists (siding with rationalism). Leonard Bloomfield explicitly embraced the then-popular version of empiricism proposed by a group of philosophers of science known as logical positivists (cf. Bloomfield 1938). Yet Bloomfield chose his allies poorly: logical positivism was heavily criticized from at least the 1930s and by the late 1950s it was clear that their version of empiricism was untenable for a number of reasons. The collapse of logical positivism left Bloomfield without much support from philosophy of science and helped the Chomskyan revolution of the 1950s and the 1960s.


Logical positivists’ empiricism, which Bloomfield found so inspiring, is “the second wave” of modern empiricism after John Locke’s, David Hume’s, and George Berkeley’s “first wave” originating in the 18th century. The “first-wave” empiricism was, as Harris notes, a theory about the source of knowl-edge. The “second-wave” empiricism of logical positivists was an attempt at providing solid foundations for scientific knowledge, largely by regimenting the language of scientific theories. The current “third-wave” empiricism, proposed and developed from the 1980s by the philosopher Bas van Fraassen, is a view on the goals of science. The third-wave empiricists have learned the lesson from the failure of logical positivists: the latest formulation of the stance is free from the problems riddling Bloomfield’s approach and, as I will show, may serve as philosophical background for discussions about modern linguistics.

The third wave begins with van Fraassen’s The Scientific Image (1980), where the philosopher advocates the view called constructive empiricism. There is a wide agreement in philosophy of science that scientific theories talk about both observable and unobservable phenomena. To take a handful of obvious examples, trees, humans, and linguistic expressions in the form of acoustic utterances and inscriptions are observable, while electrons, elec-tromagnetic fields, entropy, and cognitive domains are not. Most philoso-phers agree that science aims at accounting for observable phenomena (with a possible caveat for idealized objects discussed in more detail in Section 5), but the status and role of unobservable and purely theoretical entities is much more dubious.

The two opposing views are epistemological realism and instrumental-ism. Epistemological realists argue that the goal of science is to produce a true picture of the world, that is, a true picture of both the observable and the unobservable. A realist physicist, if she accepts modern physics as a success-ful theory, believes that electrons and electromagnetic fields exist and that they have the properties ascribed to them by modern physics. Analogically, a realist cognitive grammarian, if she believes that CG is a successful theory, would likewise believe that cognitive domains exist and that they have the properties ascribed to them by CG. An instrumentalist, in turn, claims that a goal of science is to correctly account only for the observable phenomena and unobservable theoretical phenomena are merely “useful fictions” for making accurate predictions and classifications. An instrumental physicist would believe that modern physics correctly describes observable phenomena (including phenomena produced in laboratory equipment, like measurement results), but electrons and electromagnetic fields are not real. Analogically, an instrumentalist cognitive grammarian would claim that CG correctly describes linguistic expressions, but cognitive domains do not exist: they are merely tools for producing correct descriptions of expressions. More spe-


cifically, instrumentalists believe that the unobservable theoretical entities postulated by science are not construed literally, i.e. they are not constructed in such a way that they putatively correspond to entities in the real world. For an instrumentalist, when theoretical concepts like electron or cognitive domain are postulated, they are not meant to have any counterpart in the real world, but are intended as part of the conceptual machinery yielding correct predictions or descriptions of the observable world. Strictly speaking, on the instrumentalist reading theoretical entities are not false, because they do not bear a proper relationship to extra-theoretical reality. Electrons and cognitive domains are more similar to tools, like hammers and screwdrivers, than logical propositions; they may be useful or useless, but not true or false.

Van Fraassen’s constructive empiricism from The Scientific Image is a middle path between the two extremes. It shares the realist conviction that theoretical entities are construed literally (i.e. they may correspond to some-thing in the extra-theoretical reality), but rejects the realist claim that the goal of science is to produce a true picture of the world (i.e. of both the ob-servable and the unobservable phenomena). Van Fraassen is closer to instru-mentalism in this respect, stipulating that “[science] aims to give us theories which are empirically adequate; and acceptance of a theory involves as belief only that it is empirically adequate” (van Fraassen 1980: 12) and elaborates:

The idea of a literally true account has two aspects: the language is to be literally construed; and so construed, the account is true. This divides the anti-realists into two sorts. The first sort holds that science is or aims to be true, properly (but not literally) construed. The second holds that the language of science should be literally construed, but its theories need not be true to be good. The anti-realism I shall advocate belongs to the second sort. (van Fraassen 1980: 10)

An empirically adequate theory is essentially a theory offering a correct de-scription of observable phenomena. More technically, “[to] present a theory is to specify a family of structures, its models; and secondly, to specify cer-tain parts of those models (the empirical substructures) as candidates for the direct representation of observable phenomena” (van Fraassen 1980: 64). Figure 1 is a schematic sketch of a scientific model (for the sake of clarity labels have been added to the theoretical entities constituting the model). The cloud on the left of the dotted vertical line stands for appearances, i.e. the aspect of observable phenomena available to our senses. The elements on the right of the dotted line constitute a model, featuring an empirical substructure isomorphic to appearances.


In order to clarify the notion of empirical adequacy, in Scientific Representation (2008) van Fraassen updated constructive empiricism with scientific struc-turalism, and structural empiricism was born. The key tenets of structural empiricism are:

I. Science represents the empirical phenomena as embeddable in certain ab-

stract structures (theoretical models).

II. Those abstract structures are describable only up to structural isomorphism.

(van Fraassen 2008: 238)

He adds:

In the empiricist version, the structuralist slogan is clearly and substantially qualified. It does not take reliance on theory for us to know many things about water – about observable phenomena in general – and so the slogan must be read as meaning, at best, that all we know through science is struc-ture. Nor does the formulation as I. and II. imply that there is in nature, or in the phenomena, a form/content or structure/quality distinction to be drawn. (van Fraassen 2008: 238)

Thus, structural empiricism is the view that the aim of science is to produce empirically adequate theories and an empirically adequate theory is one whose models are structurally isomorphic to observable phenomena.1

Structuralism in philosophy of science should not be confused with the structuralism of Ferdinand de Saussure (cf. Saussure 1966 [1916]). Despite

1 Isomorphism is, in fact, a very strong requirement demanding structural identity between two configurations. Ronald Giere proposes a less restrictive version of adequacy, demanding mere structural similarity between a model and its target phenomenon (cf. Giere 1999, ch. 7, Giere 2006).

Figure 1. A scientif ic model (adapted f rom Monton & Mohler 2014: n.p.)


the fact that both de Saussure and structuralist philosophers of science em-phasize the importance of structure understood as a configuration of rela-tions, philosophical structuralism does not entail linguistic structuralism. The most obvious difference between the two is that de Saussure proposed a structuralist understanding of a language: he defined a language (langue) as a system with the structure constituted by relationships between signs. Philosophers of van Fraassen’s ilk propose a structuralist understanding of a scientific theory: they claim that what scientific models are capable of capturing is the structure of the target entity. A linguist embracing philo-sophical structuralism is not bound to believe that a language is a Saussurean system of signs; in fact, she may embrace any understanding or definition of language. She merely holds that the models of linguistic phenomena ref lect the structure of linguistic phenomena, whatever these phenomena may be in the extra-theoretical reality.

Viewed through the lens of structural empiricism, Cognitive Grammar is therefore a family of models of a linguistic system. Like any other scien-tific theory, it features theoretical entities and processes (cognitive domains, constructional schemas, compositional paths, subjectification, etc.) and em-pirical substructures isomorphic (in a weaker version: similar) to appearances of observable phenomena. It is somewhat unclear what should count as ob-servable phenomena in linguistics. However, interesting and fundamental as the problem may be, it is also largely independent of the main topics of this article. For our purposes, I will adopt a broad and intuitive understanding of observables, which includes primarily linguistic expressions spoken, written, or signed, as well as replicable intuitions of speakers concerning linguistic data (e.g. judgments about grammaticality, acceptability, or semantic anom-alies). This understanding is not without problems. For example, intuitions of speakers are not publicly observable in any typical sense. One solution to the problem is to extend the notion of observation so that it embraces intro-spection; another is to treat only the reports about intuitions as observables; yet another is to emphasize the importance of replicability of observation rather than public access to the phenomenon under observation. Each solu-tion has its advantages and disadvantages, but a more detailed treatment of this topic is beyond the scope of this article. The aforementioned provisional understanding of observables including linguistic expressions and speakers’ intuitions will suffice throughout the remainder of the article.

3. Evidence for theoretical entitiesNominal Structure in Cognitive Linguistics (Langacker 2016), a record of a series of lectures given in 2014, features the following exchange between Ronald Langacker and a person from the audience:


Question: In the beginning of your talk you talked about this rock as an in-stance of grouping. Is there any evidence coming from language psychology that it’s really something that is done?

[Ronald W. Langacker]: Well, surely the answer is no (…) There’s no moti-vation for people to take up this challenge. All I’m telling you is a pretty story–but it’s a coherent story and everything fits. And given that so many nouns are transparently group nouns, which is part of the story, it lends credence to the total story.

(…)This is my attempt to say what that story might be, but I can’t claim that it’s empirical in any kind of experimental sense. (Langacker 2016: 99-100)

On the observable/unobservable spectrum, grouping would have to be classi-fied as an unobservable phenomenon. We do not observe the process of men-tal or perceptual grouping in a way we observe physical processes through sense organs. Neither extending the notion of observation to include intro-spection would help; we do not have a conscious introspective experience of our perceptual mechanisms gradually forming a concept or a percept of a group from previously ungrouped objects. Viewed in this light, the question from the audience is essentially a question about evidence, probably experi-mental confirmation, for a theoretical phenomenon postulated in CG. Lan-gacker’s answer is very much in the spirit of structural empiricism. Firstly, the author rhetorically refers to theory as a story, which is compatible with the constructivist proclivities of Fraassenian empiricists, emphasized by the term from The Scientific Image: constructive empiricism. A theory is a family of models and models are artificial constructs designed for specific aims rather than pieces of knowledge discovered in observation or experimentation. Sec-ondly, a structural empiricist insists that acceptance of a theory depends on how well the theory accounts for observable phenomena (i.e. how well its empirical substructure “fits” the observables phenomena) and not on how well the theoretical parts of the theory are supported by evidence. This is entirely consonant with Langacker’s answers quoted above. In a structural empiricist interpretation of Cognitive Grammar, the theoretical parts of the story do not require additional evidence other than the empirical adequacy and coherence of the overall theory.

This point is subtle and requires a few words of clarification. Firstly, it does not automatically follow from the structural empiricist reading of CG that mental grouping is not a suitable subject for scientific inquiry. Perhaps in some fields of science, for instance in the psychology of perception, the process of mental grouping has a more prominent role and is more than a theoretical concept accounting for observable phenomena. In such a case, scientists may have enough motivation to look for experimental evidence that the process really takes place. This, however, is not the case in Cogni-


tive Grammar in the sense that the lack of experimental evidence for mental grouping does not lead to the failure of the theory as a whole. After all CG is a theory of language intended to correctly account for linguistic phenom-ena and not a theory of perception intended to account for whether mental grouping actually happens.

Secondly, it is worth noting that the original formulation of construc-tive empiricism is silent about the beliefs concerning theoretical entities, as evidenced in the following quote “science aims to give us theories which are empirically adequate; and acceptance of a theory involves a belief only that it is empirically adequate” (van Fraassen 1980: 12). Effectively, this silence results in a voluntaristic epistemology, laid out especially in The Empirical Stance (van Fraassen 2002). Constructive and structural empiricism leaves much freedom with respect to beliefs about theoretical entities, although the view is sometimes misunderstood as involving normative guidelines on what scientists should believe in. Effectively, a structural empiricist cognitive grammarian may hold a wide range of beliefs about theoretical phenomena like mental grouping and cognitive domains, for example:

• she may believe that the phenomena exist and CG presents their true picture;

• she may believe that the phenomena exist, but CG does not present their true picture;

• she may believe that the phenomena are construed literally, but they do not really exist and function as useful fictions in CG;

• she may believe that the phenomena are construed literally, but she believes that it does not matter whether CG presents their true pic-ture as long as CG is empirically adequate;

• she may not have any particular beliefs about the status of theoret-ical phenomena.

Since acceptance of a theory involves only the belief that the theory is em-pirically adequate the acceptance does not enforce the choice of any of the above beliefs. Even the last option from the list, not having any particular beliefs, is available, since philosophical unref lectivity is not necessarily a vice in a practicing researcher. Perhaps somewhat surprisingly, the first op-tion is available as well, because commitment to empirical adequacy does not preclude the belief in the truth about theoretical entities for reasons other than strictly scientific. Nancy Cartwright aptly explains this subtlety of Fraassenian empiricism:

Van Fraassen, we should note, does not tell us we should not believe in what theory says about the unobservable. He merely says that that is not the proper, or, I take it, ‘justified’ attitude. But there is no suggestion that an attitude that is not justified is wrong or is to be avoided or is opposed


to rationality. Failing to have a justification for an attitude is in no way the same as having a justification for rejecting that attitude. There is vast space for choice. And choice is a good thing. What is important (at least if we wish not to be self-deceived) is not to conf late the beliefs and concerns we choose by ourselves with those that we are supported in. (Cartwright 2007: 41)

4. “Those diagrams”2

Another issue that can be clarified within the structural empiricist frame-work is the function of diagrams in Ronald Langacker’s works. The author notices some misconceptions surrounding the diagrams and explains his approach in the passage worth quoting in extenso:

From the frequent use of quasi-pictorial diagrams, some critics of CG have leaped to the incorrect conclusion that semantic structure is claimed to be entirely visual or spatial in nature. A related misconception is that CG can only deal with visuospatial notions. On the contrary, the essential con-structs proposed for semantic description (e.g. various kinds of prominence) are applicable to any cognitive domain and independent of any particular mode of presentation. (…) The actual intent of these diagrams is rather more modest: to allow certain facets of conceptual organization to be represented in a format that is both user-friendly and explicit enough to serve as a basis for semantic and grammatical analysis. (Langacker 2008: 12)

A structuralist approach to the diagrams does not only dispel the mis-conceptions mentioned in the passage, but also affords a stronger defense than the one proposed by the author.

Structural empiricism is at home with the so-called semantic approach to scientific theory, proposed originally by Patrick Suppes (cf. Suppes 1960). The semantic view is usually contrasted with the older syntactic view:

The syntactic picture of a theory identifies it with a body of theorems, stated in one particular language chosen for the expression of that theory. This should be contrasted with the alternative of presenting a theory in the first instance by identifying a class of structures as its models. In this second, semantic, approach the language used to express the theory is neither basic nor unique; the same class of structures could well be described in radically dif ferent ways, each with its own limitations. The models occupy centre stage. (van Fraassen 1980: 44; my emphasis – HK)

The emphasized part in the above passage is crucial for our purposes. In the semantic approach, no single formalization or notational convention (called “language” by van Fraassen) has priority over another and all of them

2 The title alludes to Langacker’s section title (cf. Langacker 2008: 9).


have inherent strengths and limitations. This also applies to formal sciences like mathematics, as Feyerabend (2011: 82) demonstrates taking the Pythag-orean theorem as an example. The most familiar modern formulation of the theorem is algebraic:

b2 = 2a2

for the isosceles right triangle. An advantage of this formulation is compact-ness and a disadvantage is abstractness: in order to understand the notation, one needs to understand how a and b correspond to the elements of the tri-angle, how the operations of multiplication and exponentiation work, and how multiplication and exponentiation are conventionally written in the above formula. The algebraic formulation describes symbolically certain geometric relations that holds for every isosceles right triangle. In order to illustrate this relation, one can exhibit an object in which the relation holds, for example a picture of the geometric objects in Figure 2.

The picture is not, strictly speaking, equivalent to the algebraic notation, because the relations expressed in the notation hold for all isosceles right tri-angles, not only to the one in Figure 2. The picture merely presents an entity that satisfies the relations expressed by the algebraic formula. The significant advantage of the picture is that the relations are more “visible,” albeit not without some exercise of imagination: if one projects the upper-right half of the smaller square inside the larger square, the two halves of the smaller square occupy a half of the larger square. Thus, after the projection is com-pleted it is clear that the field of the larger square is twice as large as the field of the smaller square. No prior knowledge of the conventions of algebraic notation and the axioms of mathematics is needed to grasp the relation. The algebraic notation and the picture in Figure 2 are not equivalent in all respects, for instance the algebraic notation is general, covering all isosceles right triangles, and the picture is specific, it shows relations of this particu-lar configuration only. Nevertheless, the two notations are equivalent with respect to the capacity of expressing the relation described by the Pythag-

Figure 2. Visual rendering of the Pythag-orean theorem (adapted f rom Feyerabend 2011: 82)


orean theorem. In the semantic approach to a scientific theory, none of the styles of formalization is more basic unless the meaning of “basic” is further qualified. The content of a scientific theory can be expressed in a symbolic propositional description or by providing an object (i.e. a model) of which the statements are true (cf. van Fraassen 1980: 41-44).

In the context of CG, Langacker touches upon the problem of alternative formalizations Cognitive Grammar. A Basic Introduction (2008: 9-12), where he defends legitimacy of diagrams in comparison to more “propositional” for-mat of representation. While arguing for general usefulness of diagrams, Langacker is characteristically cautious. For example, he claims that “[while] even the most carefully drafted [diagrams] fall considerably short of math-ematical rigor, the process of producing them forces the analyst to examine myriad details that are commonly ignored in semantic and grammatical descriptions”, the diagrams “do not amount to a mathematically respectable formalization,” and they are “heuristic in nature” (all three quotations from Langacker (2008: 10)).

I agree with Langacker that his diagrams are heuristic in nature, but the same is true for all models expressed in all formal languages. All mod-els are constructions whose aim is to facilitate our understanding of the target. Strictly speaking, Figure 2 does not depict the relations of the sides of the isosceles right triangle (these relations are abstract, so they cannot be depicted graphically); it merely offers a picture of geometric figures that exhibits these relations. However, the relations are not the symbols b2 = 2a2 either; the formula merely describes properties of the relations by means of an algebraic formula. In this sense, both the picture and the formula express selected properties of the isosceles right triangle in their own terms. I will use the term structural profile to refer to a configuration of relations singled out for the purpose of description and representation in a model. We are now in a position to define equivalence of models expressed in different formalisms more technically: models expressed in different notations are semantically and epistemically equivalent when they express the same structural profile of the target object. Conversely, if two models do not express the same struc-tural profile, the models are not semantically and epistemically equivalent.

As already mentioned, semantics of notations should be distinguished from pragmatics, i.e. the usefulness for a particular purpose. Semantically equivalent notations may differ considerably, for instance, with respect to simplicity, elegance, intuitiveness, or amenability for arithmetic operations. These pragmatic virtues invest notations with the heuristic value and con-sequently the value of each notation may be different. Yet the difference in the heuristic value does not automatically render one notation more basic or more fundamental than the other. For example, Figure 2 is a more intu-itive expression of the Pythagorean theorem, while the algebraic notation


is more amenable to arithmetical calculations, but neither is more basic in some absolute sense, and both are heuristic in one way or another. In other words, no style of formalization is more “default,” rational, or scientifically respectable in and of itself, although some styles may aspire to intellectual sophistication more than others.

At first glance, there is a tension between van Fraassen’s empirical adequacy (i.e. isomorphism between appearances and the empirical sub-structure of a model) and the semantic approach to theories. If models are expressible in different languages, many types of things can function as models: Frigg & Hartman (2017) enumerate physical and fictional objects, set-theoretical structures, descriptions, and equations as entities tradition-ally referred to as models, and add models with gerrymandered ontologies. Giere (1999: ch. 7) is stricter with terminology and seems to reserve the term model for an abstract set-theoretical object defined by (but not identical with) diagrams, descriptions, equations, etc. Giere has good reasons for regiment-ing terminology, but I will adopt the polysemous, although somewhat messy, terminology of Frigg and Hartmann, since it appears to be closer to the un-derstanding of the term model in ordinary language.

Since the items of Frigg & Hartmann’s list are ontologically different, how can they be isomorphic to each other? How can an frictionless pendulum, a purely fictional object, serve as a model of an actual physical pendulum? How can a diagram used in CG serve as a model of a conceptual structure? Van Fraassen’s answer in Scientific Representation (2008) is that isomorphism between objects of different ontologies may hold, because models ref lect the structural profiles of their targets, and the profile is an abstract and broadly mathematical notion that is not inherently bound to any particular physical or conceptual object. The fact that structure is broadly mathematical does not mean that it is immediately amenable to arithmetic calculations, but that it is a configuration of relations describable in terms of mathematical set theory. It should be noted that van Fraassen does not reduce all knowl-edge to mathematics, even broadly understood. He merely claims that “all scientific representation is at heart mathematical. [Structural empiricism] is a view not of what nature is like but of what science is” (van Fraassen 2008: 239). A good scientific model is therefore an object that expresses a structural profile of its target. In the unregimented terminology used in this article, the structural profile of the Pythagorean theorem, the algebraic notation, and Figure 2 all count as models, even though the structural profile is a model of a special kind: it is an abstract entity constituting the core of the theory embeddable in other models. For this reason, objects of different ontologies may function as models by virtue of having the same structural profile as their targets. For instance, a fictional frictionless pendulum may serve as a


model of a physical pendulum, because the objects share the same configu-rations of relations, which is to say that they are mathematically equivalent in some important and relevant respects.

On this view, Langacker’s diagrams are effective models of conceptual phenomena to the extent to which the diagrams represent structural pro-files of the conceptual phenomena. Since (under a structuralist reading) a structural profile is a set-theoretical object, a diagram is a mathematical representation in the broad sense and it is mathematically rigorous to the extent to which it correctly depicts (some relevant aspect of) the structural profile intended for depicting. It is also semantically, but not necessarily pragmatically, equivalent to other conventions of formalization. Let us il-lustrate this with an example.

In “The function of trees” Langacker (2017) discusses various ways of visualizing dependency relations between the head and its dependent in that cute girl with freckles. A model of dependency should express information about: 1) the relata, 2) the asymmetry of the relation, and 3) the prominence of girl as the overall head of the expression. The structural profile could be described verbally resulting in a somewhat laborious collection of sentences:

• Girl is the overall head of the expression that cute girl with freckles.• There is a dependency relation between that and girl; that is depen-

dent on girl.• There is a dependency relation between girl and with freckles; with freck-

les is dependent on girl.• …

Alternatively, a quasi-predicate-logical notation can be devised for the same purpose, for example:

• H(girl, that cute girl with freckles)• D(girl, that)• D(girl, D(with, freckles))• …where H(a, b) element stands for a being the head for b and each D(a, b)

element stands for a dependency relation in which b is dependent on a. The propositional descriptions of this sort are analogs of the algebraic formu-lation of the Pythagorean theorem: they are verbal or symbolic character-izations of the structural profile. Yet instead of offering a description, an object instantiating the structural profile can be offered, for example the dependency tree in Figure 3.


The dependency tree shows the same structural information as the descrip-tions: the relata are placed at the end of the lines, prominence of girl is sig-naled by vertical organization of the items (the head is the highest element in the hierarchy), and the vertical organization logically entails asymmetry of the relation (by necessity, one element cannot be above and below an-other element at the same time, so the vertical alignment entails that the relationship works in one direction only). The dependency tree expresses the structural profile not by describing it, but by exhibiting the profile in its structure; in other words, the relations in the structural profile are the relations holding between the elements of the tree. Langacker also proposes the following diagram illustrating dependency relations:

The diagram in Figure 4 is richer is structural details, because it presents information about the grounding relation (in the box with that) and about internal semantic structure of words (circle for singular noun, ellipse for plural noun, etc.). Due to this surplus information, the diagram is not a strict semantic equivalent of the dependency tree, and the propositional descriptions. However, the dependency tree is still embeddable into the di-agram, because all relations expressed by the dependency tree are present in the diagram: participants of the relation appear at the end of arrows, the directionality of arrows signals asymmetry of dependency relation, and girl is marked as the head with the heavy-line box. Langacker explicitly states that

Figure 4. Dependency diagram for that cute girl with f reckles (adapted f rom Langacker 2017: 81)

Figure 3. Dependency tree for that cute girl with f reckles (adapted f rom Langacker 2017: 74)


“[it] is readily seen that Figure [3] is equivalent to [4] apart from its greater detail”3 (Langacker 2017: 81).

To summarize, under the structural empiricist interpretation, Cognitive Grammar is a family of models of a linguistic system. In the most abstract sense, the models of CG are essentially a collection of set-theoretical models (structural profiles) intended to be isomorphic to the structural profile of the linguistic system. These set-theoretical models can be described by means of propositional descriptions or manifested in concrete objects, like diagrams. Obviously, the structural profile of the entire linguistic system cannot be depicted in a single diagram due to practical limitations of the medium and the complexity of the linguistic system, but portions of the system’s structure can be depicted for the purpose of the analysis at hand. In order to decipher the relevant structure from the diagram, familiarity with diagram’s visual conventions and practice are needed, but this is true for every format of rep-resentation used in science. Under the structuralist reading, diagrams are mathematically rigorous in the sense that a properly designed diagram is a manifestation of a set-theoretical structure captured by Cognitive Grammar. A diagram may be designed unrigorously, when the author fails to depict the relation intended for depicting or uses visual conventions inconsistently, but the visual format itself is not inherently unrigorous.

5. Sources of dataThe problem of legitimate sources of data for linguistic analysis is not

specific to CG and it is not discussed directly in mainstream philosophy of science. Therefore, we are now departing slightly from Cognitive Grammar and structural empiricism proper into wider waters of methodological and philosophical debates. In linguistics, the discussion on what counts as legit-imate data is long and complex, so it is not possible to provide a comprehen-sive summary4 or decisive cut-and-dry conclusions in this article. Instead, I will attempt to examine how certain ideas of structural empiricism and other constructivist stances inform the debate in linguistics and lend some support to the methods of data collection used in CG. In particular, I will focus on the role of idealizations in scientific model-making.

By doing this, I depart slightly from the way in which sources of data are traditionally discussed by linguists. While my argument does contribute to previous discussions in the community of linguists, instead of arguing for merits and demerits of constructed expressions as sources of data, I will point out the connections to a larger philosophical debate about idealizations.

3 Obviously, Langacker’s statement presupposes familiarity with CG formalisms and practice in using them. To an untrained eye, the correspondence may not be readily visible.4 For an overview, see Scholz, Pelletier, & Pullum (2016: 3).


The key question here is “How is it possible to learn something about the ac-tual state of affairs by examining counterfactual states of affairs?” Neither here, nor anywhere else in the article do I wish to prescribe certain research practices or claim that the use of idealizations warrants scientificness (it does not). My intention is rather to make inroads into the epistemic role of constructed expressions in the context of the epistemic role of counterfac-tuals in other sciences.

A convenient starting point is one of the tightly intertangled threads of the debate on methodology in linguistics: should linguists rely exclusively on actual attested data (typically derived from corpora) or should expressions constructed for the purpose of linguistic theorizing be admitted as legitimate data? Halliday & Matthiessen are enthusiasts of authentic data from corpora:

What people actually say is very different from what they think they say; and even more different from what they think they ought to say (…) Simi-larly, what people say or understand under experimental conditions is very different from what they say or understand in real life (…) The difference is less marked in writing, although it is still there (…) But it is in speech that authenticity becomes critical. (Halliday & Matthiessen 2004: 34)

In cognitive linguistics, similar views are expressed, among others, by Bar-low & Kemmer:

[T]he primary object of study is the language people actually produce and understand. Language in use is the best evidence we have for determining the nature and specific organization of linguistic systems. Thus, an ideal usage-based analysis is one that emerges from observation of such bodies of usage data, called corpora (…) Because the linguistic system is so closely tied to usage, it follows that theories of language should be grounded in an observation of data from actual uses of language. (Barlow & Kemmer 2002: xv)

The authors note that “[speaker] intuitions about constructed examples are an invaluable tool, provided that such data are treated with all appropriate care,” but promptly add that “even with such judicious use, intuitions about constructed data cannot be treated as the sole, or even primary, source of evidence as to the nature and properties of the linguistic system” (both quo-tations from Barlow & Kemmer 2002: xv).

Arguments for the use of authentic data are often made against gener-ative linguists and their supposed overreliance on constructed expressions. Generativists reply:

More than five decades of research in generative linguistics have shown that the standard generative methodology of hypothesis formation and empirical verification via judgment elicitation can lead to a veritable gold-


mine of linguistic discovery and explanation. In many cases it has yielded good, replicable results, ones that could not as easily have been obtained by using other data-gathering methods, such as corpus-based research (…) [Consider] the fact that parasitic gap constructions (…) are exceedingly rare in corpora (…) [These] distributional phenomena would have been entirely impossible to distill via any non-introspective, non-elicitation based data gathering method. (den Dikken et al. 2007: 336)

The linguistic material analyzed in many of Langacker’s texts appears to be either constructed or at least picked in order to illustrate points under anal-ysis. For this reason, the analyses can be easily attacked on the grounds that they are somehow unreliable due to the inauthenticity of data. Consider the following sentences used in Cognitive Grammar. A Basic Introduction (Langacker 2008: 87) illustrating predictions made on the basis of semantic descriptions:

(1) (a) The clock is {sitting / standing / lying} on the table. (b) The {vase / ?pen / ?football / ?*watermelon /*mat /*peach} is standing on the

table.

According to the author, the choice of the verb in (1a) depends on the shape of the clock construed metaphorically in terms of the human body, while the oddity of some nouns in sentence (1b) results from the incompatibility be-tween the semantic characterizations of stand (implying vertical orientation) and nouns which are typically at rest in horizontal position. Since Langacker does not provide any attested source, let us suppose that the sentences are constructed or cherry-picked. This suspicion is particularly strong for the anomalous nouns in (1b); since the author admits that sentences like *The peach is standing on the table is atypical, it is unlikely to be produced in re-al-life circumstances. If the argumentation of proponents of authentic data is brought to its logical conclusion, sentences of this sort should not be admitted as valid material for analysis. More generous proponents of authentic data may turn the blind eye on (1a), as it does not look overly exotic and is likely to the produced in natural conversation, but (1b) is less likely to be authentic; less generous ones may balk at the very idea that potential authenticity of data can be evaluated through armchair speculation.

What strategies are available for defending the use of constructed sen-tences, like the ones in (1a) and (1b) (assuming they are constructed after all)? One strategy consists in pointing out that the very distinction between authentic/actual and artificial/constructed data is somewhat strained. It is quite obvious that an overwhelming majority of actually produced expres-sions have never been recorded in any form, so it can hardly be argued that an expression does not count as authentic simply because its use cannot be at-tested. I will not resort to this strategy for two reasons. Firstly, the argument


formulated like this is a caricature: I presume that even the most sanguine opponents of constructed data would not seriously equate the authentic with the attested. Instead, what they defend (if I understand correctly) is caution and prudence in data collecting. It would be absurd to claim that an expression in (1a) has never been produced just because it is not attested, yet since (1a) is not attested, there is no way of knowing whether it was ac-tually produced. Thus, it is more cautious and prudent to rely exclusively on attested expressions, which are certainly authentic.

Secondly, and more importantly, I will not resort to this strategy, be-cause I defend the view that even inauthentic expressions may be useful in the study of language. As a preliminary step, I will strengthen the argument of the proponents of authentic data by focusing on non-occurrent expres-sions, rather than the merely unattested ones. I will define non-occurrent expressions as expressions that have never been produced. Obviously, for purely practical reasons, we can never know for sure whether an expres-sion is non-occurrent. To help our imagination, let us envision a linguistic version of Laplace’s demon, a supernatural being who hears and sees ev-ery expression produced in natural circumstances by every person in every language since the dawn of speech, and collects all these expressions in a corpus (“natural circumstances” excludes expressions concocted by linguists for purely theoretical purposes). Then, a non-occurrent expression is simply one that is not in Laplace’s demon’s corpus. This little thought experiment strengthens the case for the proponents of authentic data, since now a lin-guist who wishes to analyze non-occurrent expressions cannot argue that possibly the expressions are merely unattested rather than inauthentic. Given the Laplace’s demon’s corpus, we know for sure that the expressions have never been produced in a real-life circumstances. Let us now turn to the main part of my argument.

Non-occurrence of certain linguistic expressions as defined in the pre-vious paragraph is one of several kinds of counterfactuality (i.e. non-actual-ity). Counterfactual situations and scenarios may differ in terms of modal5 strength. To borrow Marc Lange’s (2016) examples, no mother has ever man-aged to distribute evenly 23 strawberries among 3 children (without cutting any) and no one has ever managed to untie the trefoil knot.6 Situations of equal distribution of 23 strawberries among 3 children and untying the tre-foil knot are non-actual, because they are prevented by the formal (broadly mathematical) properties of the number 23 and the trefoil knot respectively. In other words, due to these formal properties, it is inconceivable how 23

5 Here, I use the word modal in the alethic sense, i.e. referring to the notions of possibility, ne-cessity, actuality, and the like.6 In knot theory, a trefoil knot is the simplest nontrivial knot. Cutting or ripping apart does not count as untying. More technically, untying a knot is an operation where only Reidemeister moves, as defined in knot theory, are permitted.


strawberries could be distributed equally among 3 children (without cutting any) and how the trefoil knot could be untied. Inconceivability of a scenario implies logical impossibility. Hence, the equal distribution of strawberries and untying the trefoil knot are counterfactual due to logical impossibility, i.e. the impossibility of imagining a coherent scenario in which the tasks could be completed successfully.

Other scenarios are counterfactual in different ways. For example, no one has ever managed to produce a frictionless pendulum, or an isolated system (a physical system that does not exchange heat and work with its environment) featured in the second law of thermodynamics, or a perfect f luid (a f luid with no viscosity and no heat conduction). Frictionless pendu-lums, isolated systems, and perfect f luids are counterfactual, because it is impossible to produce them in practice: no physical object can be thermically insulated in such a way that is exchanges no heat at all with its environment and even a pendulum swinging in perfect vacuum (hence experiencing no friction from air) would be affected by the friction at the pivot. Nonetheless, these objects are conceivable, i.e. one can imagine a logically coherent sce-nario in which these objects are in working order. Conceivability combined with “in practice” impossibility typically implies natural impossibility, when a conceivable scenario is ruled out by laws of nature.7

Finally, some scenarios are counterfactual because of certain historical circumstances, even though they are logically and naturally possible. No human has ever landed on Mars (as of 2019) and no human has ever seen a living dinosaur, but these scenarios are not ruled out by the laws of logic or nature. Humans simply happened not to live in the days of dinosaurs and the technology enabling travel to Mars happened not to be developed by 2018. Observing a living dinosaur or travel to Mars are impossible, but this impossibility is modally weaker than the logical and the natural type: the scenario is precluded by contingencies of historical developments and not by any fundamental laws of universe. This is the so-called temporal impossibility pertaining to what is “impossible now,” in given historical circumstances.

Counterfactuality of non-occurrent expressions does not appear to fit into any of the above classes. Sentences in (1a), assuming they are indeed non-occurrent, do not appear to be impossible. Most sentences in (1b) are perhaps unlikely to occur in natural circumstances, but they are not straight-forwardly impossible either. In fact, Langacker convincingly argues that they could appear in certain situations:

7 Strictly speaking, it is debatable whether frictionless pendulums and isolated systems are nat-urally impossible, since laws of physics do not rule them out; what does rule them out are practical difficulties of constructing these objects. Yet this kind of “practical” impossibility is much closer to natural rather than logical or temporal impossibility, so I will conf late natural and practical impossibility for the sake of simplicity.


[Some sentences in (1b)] are acceptable (…) if we imagine that the pen is standing on end (possible with certain pens) or that the football is on a kick-ing tee. Because it has rounded ends, a watermelon can only lie on a table unless we concoct a bizarre context (e.g. it might be impaled on a spike) (…) encyclopedic knowledge tells us that mats are sometimes rolled up, and a rolled-up mat could well be stood on end; under this interpretation we can perfectly well describe it as standing. (Langacker 2008: 87)

Even such bizarre expressions as the famous Colorless green ideas sleep furi-ously devised for a purely theoretical purpose by Noam Chomsky does not appear to be impossible in any of the above senses, although perhaps Furiously sleep ideas green colorless (both examples from Chomsky 1957: 15) may be ruled temporally impossible in the sense that given the historical development of English, the expression could not have been produced and considered an English expression. Thus, assuming that Langacker’s (1a) and (1b), as well as Chomsky’s Colorless green ideas… are counterfactual in virtue of non-oc-currence (they are absent from Laplace’s demon’s corpus), their counterfac-tuality does not belong with logical, natural, and temporal impossibility. In fact, despite their non-occurrence, all of the sentences appear possible, even if not all of them are very likely to be produced in real-life situations. A non-occurrent possibility remains counterfactual, but it is modally weaker than any kind of impossibility. In sum, various kinds of counterfactuality form a modal scale arranged from the strongest logical impossibility to the weakest non-occurrence:

logical impossibility > natural impossibility > temporal impossibility > non-occurrence

Many linguists seem to embrace the conviction that linguistics should study only actual, real-life events of language use, which do not belong any-where on the above counterfactuality scale. However, this conviction over-looks the fact that many useful scientific models involve clearly counter-factual scenarios.8 The already mentioned frictionless pendulums, isolated systems, and perfect liquids are examples of naturally impossible objects. Their role is not merely heuristic or pedagogical; some of them are at the very heart of theories in natural science. For instance, the first and the second law of thermodynamics describe the behavior of a thermodynamically isolated system, which is a naturally impossible fiction performing an important epistemic role despite the fact that its counterfactuality is modally stronger than the mere non-occurrence of constructed sentences.

8 Of course, an enthusiast of authentic data may simply reject the usefulness of counterfactual scenarios in science by claiming that linguists, as opposed to natural scientists, cannot learn anything from the scenarios. This, however, would require at least demonstrating how the epis-temological situation of a linguist differs from the epistemological situation of a natural scientist.


The use of fictional models in natural sciences can be accounted for in philosophical terms. In the framework of structural empiricism, isolated systems, frictionless pendulums, and perfect liquids are treated just like any other theoretical objects, except their theoreticity does not follow from unobservability, as is the case with electrons and electromagnetic fields, but from their fictional character. Since a structural empiricist is not commit-ted to believing in the actual existence of real-world objects corresponding to theoretical concepts, but only to the empirical adequacy of the theory featuring these concepts, the structural empiricist may accept the theory while withholding belief in the actual real-world existence of the objects postulated by the theory. An idealized model may serve as a tool for predict-ing observable phenomena when predictions derived from the model are adjusted with respect to the factors omitted in the idealization. In physics, a real-life example of this procedure (calculating the properties of an actual signal in an idealized model of an electric amplifier) is described by Nancy Cartwright (1983: 107-112)

In linguistics, an analogical procedure is Noam Chomsky’s (1965: 11) analysis of a pair of expressions quoted here as (2a) and (2b). For the sake of discussion let us once again assume that the expressions are constructed and non-occurrent in the sense defined above. This assumption is corroborated by their intended unacceptability in natural circumstances:

(2) (a) I called the man who wrote the book that you told me about up.(b) the man who the boy who the student recognized pointed out is a friend of mine.

Chomsky uses (2a) and (2b) to illustrate the difference between grammat-icalness and acceptability. Essentially, the author argues that (2a) and (2b) are grammatical, because they can be derived from an idealized model of linguistic competence (idealized grammar). When the derivation is adjusted with respect to factors affecting acceptability, (2a) and (2b) are deemed un-acceptable. Here, the factors proposed by Chomsky are the principles “Nest-ing of long and complex elements reduces acceptability” for (2a) and “Re-peated nesting [especially self-embedding] contributes to unacceptability” (Chomsky 1965: 12-13). Thus, Chomsky recognizes that an idealized model of linguistic competence alone does not allow for accurate derivations of acceptability judgments. The idealized grammar allows for deriving (2a) and (2b), and predicts their grammaticalness, but, as Chomsky rightly notes, native speakers would not consider these expressions acceptable. It is only after adjusting the idealized derivations with respect to acceptability that more empirically adequate predictions regarding acceptability can be made.


By offering this example, I do not subscribe to Chomsky’s analysis, ac-cept his analytic framework, or suggest that the analysis is the state-of-art in generative linguistics. The example is only meant to illustrate that the idealizations in linguistics can be used similarly to idealizations in natural sciences. A structural empiricist linguist does not have to be committed to the belief in the authenticity of sentences (1a), (1b), (2a), and (2b) in order to claim that linguists can learn something interesting from analyzing them. Regardless of whether the sentences are attested, occurrent (actually pro-duced in contexts others than linguistic theorizing) or non-occurrent, they may be treated as parts of a larger theory, whose ultimate goal is to offer an empirically adequate account of observable phenomena. By analogy, a structural empiricist physicist is not committed to the belief in the existence of electrons or isolated systems, or to the belief that unadjusted predictions derived from the theories featuring electrons and isolated systems are correct descriptions of the observable phenomena. Instead, she may be committed to the belief that adjusting the idealized predictions with respect to the factors omitted in the idealization yields more empirically adequate predictions.

By arguing in favor of constructed expressions, I do not attempt to un-dermine the significance of actual, occurrent, and attested expressions in linguistics. After all, in structural empiricism the key criterion of theory’s success is how well the theory accounts for observable phenomena, which includes phenomena observed in real-life circumstances. Rather, my argu-ment is meant to emphasize the importance of idealized and counterfactual phenomena in scientific theorizing. Bluntly speaking, since natural scientists benefit from the use of counterfactual objects, why can’t linguists? Non-oc-current expressions can be viewed as valid research material, especially if we bear in mind that the counterfactuality of the expressions is a modally weaker than that of many models fruitfully used in natural sciences. In ef-fect, the discussion in this section is a call for methodological tolerance and pluralism based on appreciation of effective solutions in other sciences.

6. ConclusionPhilosophers of science, and philosophers in general, are often viewed as people confined to ivory towers and divorced from down-to-earth problems of “the real life.” This view is not always false. Undoubtedly, a linguist does not have to study philosophy of science to be a good researcher, just like a speaker does not have to study the grammar of her language to use the language f luently. Yet, philosophical theories can be more than intellectual playthings with no real importance for scientific practices.

In this article, I demonstrated how viewing science through the lens of structural empiricism can help to solve concrete problems not only in Cog-


nitive Grammar, by in linguistics by and large. The problem of evidence for theoretical concepts loses most of its force, because a structural empiricist may deny that theoretical concepts require evidence other than empirical adequacy of the theory. This seems to be in tune with Ronald Langacker’s answer in the passage from Section 3: a coherent story that explains all (ob-servable) facts is all we need. Objections to diagrams as proper devices of formalizations vanish in the semantic approach to theories, where a model is expressible in many semantically equivalent ways. What philosopher and semioticians are left with is the task of studying how different styles of formalization achieve their epistemic and cognitive efficacy. The problem of sources of data does not receive a straightforward solution, but a better understanding of the role of idealizations in science helps to vindicate ex-pressions constructed for the purpose of linguistic theorizing as valuable material for analysis. These problems are certainly less spectacular than the one (rightly) dismissed by the reviewer in the introductory section: “whether linguistics or any branch of it is entitled to call itself a science.” Yet from the point of view of everyday research practice, these smaller problems may be even more important.

References

Barlow, Michael, Suzanne Kemmer (eds.) 2002: Usage Based Models of Language. Stanford: Center for the Study of Language and Information Sciences.

Bloomfield, Leonard 1938: Linguistic Aspects of Science. Chicago: Chicago Uni-versity Press.

Cartwright, Nancy 1983: How the Laws of Physics Lie. Oxford-New York: Oxford University Press.

Cartwright, Nancy 2007: Why Be Hanged for Even a Lamb? In: Bradley Mon-ton (ed.) 2007: Images of Empiricism. Essays on Science and Stances, with a Reply from Bas C. van Fraassen, 32-45. Oxford-New York: Oxford Uni-versity Press.

Chomsky, Noam 1957: Syntactic Structures. The Hague: Mouton.Chomsky, Noam 1965: Aspects of the Theory of Syntax. Cambridge: The MIT

Press.Den Dikken, Marcel, Judy B. Bernstein, Christina Tortora, Raffaella Zanut-

tini 2007: Data and Grammar: Means and Individuals. Theoretical Lin-guistics 33 (3), 335-52. URL: https://doi.org/10.1515/TL.2007.022 ED 16 October 2018.

Feyerabend, Paul. K. 2011: The Tyranny of Science. Cambridge-Malden: Polity.


Frigg, Roman, Stephan Hartmann 2017: Models in Science. The Stanford Ency-clopedia of Philosophy, Edward N. Zalta (ed). URL: https://plato.stanford.edu/archives/spr2017/entries/models-science/ ED: 16 October 2018.

Giere, Ronald N. 1999: Science without Laws. Chicago: University of Chicago Press.

Giere, Ronald N. 2006: Scientific Perspectivism. Chicago: University of Chi-cago Press.

Halliday, Michael, Christian Matthiessen 2004: An Introduction to Functional Grammar. London-New York: Routledge.

Harris, Randy A. 1993: The Linguistics Wars. New York-Oxford: Oxford Uni-versity Press.

Langacker, Ronald W. 1987: Foundations of Cognitive Grammar. Stanford: Stan-ford University Press.

Langacker, Ronald W. 2008: Cognitive Grammar. A Basic Introduction. New York: Oxford University Press.

Langacker, Ronald W. 2016: Nominal Structure in Cognitive Grammar: The Lublin Lectures. Adam Głaz, Hubert Kowalewski, Przemysław Łozowski (ed.). Lublin: Maria Curie-Skłodowska University Press.

Langacker, Ronald W. 2017: The Function of Trees. In: Przemysław Łozowski, Adam Głaz (eds.) 2017: Route 66: From Deep Structures to Surface Meanings: A Festschrif t for Henryk Kardela on His 66th Birthday. Lublin: Wydawnictwo Uniwersytetu Marii Curie-Skłodowskiej, 73-91

Lange, Marc 2016: Because Without Cause: Non-Causal Explanations in Science and Mathematics. New York: Oxford University Press.

Monton, Bradley, Chad Mohler 2014: Constructive Empiricism. The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed). URL: http://plato.stan-ford.edu/archives/spr2014/entries/constructive-empiricism/ 16 October 2018.

Saussure, Ferdinand de 1966 [1916]: Course in General Linguistics. Trans. Wade Baskin. New York, Toronto and London: McGraw-Hill.

Scholz, Barbara C., Francis Jeffry Pelletier, Geoffrey K. Pullum 2016: Phi-losophy of Linguistics. The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed). URL: https://plato.stanford.edu/archives/win2016/entries/linguistics/ 16 October 2018.

Suppes, Patrick 1960: A Comparison of the Meaning and Uses of Models in Mathematics and the Empirical Sciences. Synthese 12 (2-3), 287-301.

Van Fraassen, Bas C. 1980: The Scientific Image. Oxford: Clarendon Press.Van Fraassen, Bas C. 2002: The Empirical Stance. New Haven: Yale University




Streszczenie

Gramatyka Kognitywna w ujęciu strukturalistyczno-empirycznym

Artykuł szkicuje strukturalno-empirycystyczne podejście do gramatyki kognitywnej Ronalda W. Langackera. Współczesny empiryzm, określany pierwotnie mianem empiryzmu konstruktywnego, to stanowisko w ramach filozofii nauki rozwijane przez Basa van Fraassena od lat 80. XX wieku. Pod-stawowe założenie tego stanowiska głosi, że język teorii naukowych kon-struowany jest dosłownie – co oznacza, że terminy zawarte w teoriach na-ukowych mogą odnosić się do obiektów w rzeczywistości pozateoretycznej

– lecz warunkiem zaakceptowania teorii nie jest jej prawdziwość, a jedynie adekwatność empiryczna. W tym rozumieniu teoria adekwatna empiryczne prezentuje wierny obraz zjawisk obserwowalnych, w przeciwieństwie to teorii całkowicie prawdziwych, prezentujących wierny obraz zjawisk za-równo obserwowalnych, jak i nieobserwowalnych. Empiryzm rozumiany w ten sposób stoi w sprzeczności zarówno z realizmem epistemologicznym, zgodnie z którym warunkiem akceptacji teorii jest jej całkowita prawdziwość, jak i z instrumentalizmem, zgodnie z którym elementy teorii odnoszące się do zjawisk nieobserwowalnych nie są konstruowane literalnie. Filozoficzny strukturalizm to z kolei przekonanie, że teorie naukowe odzwierciedlają strukturę badanych zjawisk, rozumianą jako sieć relacji, milczą zaś na temat metafizycznej natury elementów powiązanych tymi relacjami.

Proponowanie strukturalno-empirycystycznego ujęcia Langackerow-skiej gramatyki kognitywnej nie jest jedynie czczym ćwiczeniem z zakresu filozofii nauki i może przyczyniać się do rozwiązania praktycznych pro-blemów metodologicznych w językoznawstwie. W artykule skupiam się na omówieniu trzech takich problemów.

Pierwszym z nich jest zagadnienie dowodów na istnienie zjawisk, do których odnoszą się terminy teoretyczne, takich jak proces grupowania men-talnego zaproponowany przez Langackera między innymi w celu sformuło-wania semantycznej definicji rzeczownika. W empiryzmie strukturalnym brak twardych dowodów na istnienie takich zjawisk nie dyskwalifikuje teorii, bowiem jej akceptacja zależy jedynie od adekwatności empirycznej (stopnia, w jakim teoria prawidłowo oddaje zjawiska obserwowalne). Co za tym idzie, strukturalno-empirycystyczny językoznawca pracujący w paradygmacie gramatyki kognitywnej nie jest zobowiązany, by dostarczyć dowód na to, że proces grupowania mentalnego rzeczywiście zachodzi, jeśli teoria postulu-jąca grupowanie mentalne jest empirycznie adekwatna.


Drugim z omawianych problemów jest zagadnienie wykresów stosowa-nych w publikacjach Langackera. Sam autor odżegnuje się od stwierdzenia, że wykresy są ścisłymi i sformalizowanymi wyrażeniami treści teoretycz-nych, lecz dla empirysty strukturalnego starannie przygotowany wykres może funkcjonować jako wysoce sformalizowany model teoretyczny. Em-piryzm strukturalny wpisuje się w tzw. semantyczne podejście do teorii naukowych, w którym teoria to grupa modeli danego zjawiska opisywalnych przy pomocy różnorakich formalizacji. Podejście semantyczne przeciw-stawiane jest starszemu podejściu syntaktycznemu, w którym teoria jest tworem wyrażalnym jedynie w ściśle zaksjomatyzowanym języku formal-nym. W ujęciu semantycznym model może być wyrażony quasi-algebra-iczną notacją logiczną lub opisem sformułowanym w języku naturalnym, ale również w formalizacji graficznej, na przykład w postaci diagramu. W podejściu strukturalnym notacja logiczna, opis językowy i diagram mogą być równoważnymi „ekspresjami” modelu, jeśli tylko opisują bądź wyrażają tę samą strukturę.

Ostatnim zagadnieniem jest kwestia źródeł danych do analiz języko-znawczych, a w szczególności pytanie, czy wyrażenia językowe konstru-owane doraźnie na potrzeby takiej analizy stanowią wartościowy materiał badawczy. W podejściu strukturalno-empirycystycznym wyrażenia takie mogą być traktowane jako idealizacje, czyli zjawiska, które wprawdzie nigdy nie zaszły lub zajść nie mogą, lecz mimo to dostarczają naukowcowi ważnych informacji na temat badanego zjawiska. Przykładami idealizacji w naukach ścisłych są układy izolowane termicznie (jedynie do nich odnoszą się prawa termodynamiki) lub wahadła, w których nie występuje tarcie (jedynie w ta-kich wahadłach można zaobserwować prawo zachowania energii). Mimo że w praktyce nie da się zbudować układu izolowane termicznie (to jest takiego, który nie wymienia ciepła ze swoim otoczeniem) ani wahadła, w którym nie występuje tarcie, pomiędzy elementami mechanicznymi, idealizacje są używane nie tylko do celów dydaktycznych, lecz stanowią zasadniczą treść fundamentalnych teorii współczesnej fizyki. Układy izolowane i wahadła, w których nie występuje tarcie to obiekty kontrfaktyczne w dość silnym sen-sie tego słowa, ponieważ ich istnienie wykluczone jest przez prawa fizyki. W porównaniu z nimi doraźnie konstruowane wyrażenia językowe – za-kładając, że rzeczywiście nigdy nie zostały wypowiedziane i napisane w „naturalnych” okolicznościach – są kontrfaktyczne w sensie znacznie słabszym. Jeśli więc obiekty naturalnie niemożliwe mogą pełnić użyteczną rolę w naukach ścisłych, nie należałoby z zasady wykluczać użyteczności zdań skonstruowanych na doraźne potrzeby analizy językoznawczej. Języko-znawca pracujący w ramach empiryzmu strukturalnego może potraktować takie wyrażenia jako idealizacje i podejść to nich podobnie jak do obiektów


teoretycznych – nie musi więc wierzyć, że „zaistniały” one w naturalnych okolicznościach, a powinien jedynie przyjmować, że teoria, które wyrażenia te opisuje, jest empirycznie adekwatna.


Małgorzata MisiakUniwersytet Wrocławski

Co to znaczy być Łemkiem? W kręgu pytań tożsamościowych. Ujęcie etnolingwistyczne2

1 Artykuł odwołuje się do badań przedstawionych w książce: M. Misiak, 2018, Między Popradem a Osławą. Tożsamość kulturowo-językowa Łemków w ujęciu etnolingwistycznym, Profil, Wrocław. Tekst jest przedrukiem fragmentów książki ułożonych w zmodyfikowanej kolejności.

Artykuł dotyczy zagadnień tożsamościowych. Została przedstawiona w nim próba odtworzenia tego obszaru samoświadomości Łemków, w którym zawiera się odpowiedź na pytanie, co to zna-czy być Łemkiem? Pytanie postawiono specyficznej próbie respondentów. Stanowią oni pokolenie odchodzące. Można je określić mianem „pokolenia przejściowego”, gdyż jego doświadczenie biograficzne objęło swych zasięgiem czasowym kilka przełomów dziejowych.

Materię badawczą stanowił zbiór tekstów wywołanych, na które składają się zapisane, nie-kiedy wielogodzinne (najdłuższa relacja obejmuje 4 godziny nagrania), narracje łemkowskie. Zastosowana filologiczna analiza tekstów odwołuje się do założeń tekstologii. W pracy wykorzy-stano głównie aparat pojęciowy z zakresu struktury i semantyki tekstu. Wyzyskane wywiady łem-kowskie należą natomiast do typu narracji określanych mianem historii mówionej (oral history).

Artykuł ze względu na specyfikę podejmowanej problematyki sytuuje się na pograniczu nauk: lingwistyki, antropologii, socjologii, etnologii z szeroko zarysowanym tłem historycznym. Podstawowym odniesieniem pozostaje jednak lingwistyka z jej odmianami: etnolingwistyką, socjolingwistyką czy lingwistyką kognitywną.

W wyniku przeprowadzonych analiz zebranego materiału w pierwszej kolejności odkodo-wano zawarte w testach wyobrażenia, sądy i wartości współtworzące świat mentalny narratorów, a następnie wskazano ich wzajemne relacje, uporządkowano i pogrupowano w większe struktury.

Z materiału wyodrębniono w ten sposób cztery wzajemnie się uzupełniające, ale też przeci-nające się zbiory / profile określające łemkowską tożsamość kulturową. Są to: 1/ ŁEMKO – OFIARA HISTORII (profil grupujący syndromy i deskryptory z punktu widzenia określonej generacji,

ABST

RACT

1


41Co to znaczy być Łemkiem?...

a zatem w ujęciu generacyjnym); 2/ ŁEMKO – WYGNANIEC (rekonstruowany z punktu widze-nia wysiedleńca, czyli osoby zmuszonej do zmiany miejsca zamieszkania); 3/ ŁEMKO ETNICZ-NO-KULTUROWY (odnoszący się do osoby należącej do społeczności łemkowskiej), wreszcie 4/ ŁEMKO – CHŁOP (zespół syndromów zebranych z punktu widzenia mieszkańca wsi, z punktu widzenia chłopa).

Przeprowadzone analizy pokazały w sposób jednoznaczny, że nie istnieje odpowiedź finalna na pytanie o tożsamość kulturowo-językową. Każdy nowy wątek może stanowić jej dopełnienie. Tożsamość, każda tożsamość, ma to do siebie, że jest procesem, więc i określenie jej kształtu zmienia się pod wpływem czasu i kontekstu.

Artykuł wpisuje się w rozległy zakres badań nad problematyką łemkowską. Stanowi nowe ujęcie problemu definiowania tożsamości językowo-kulturowej. Przedstawiona analiza została przeprowadzona z perspektywy uwikłań historyczno-społecznych Łemków.

Otrzymano: 27.03.2019. Zrecenzowano: 03.10.2019. Przyjęto: 11.11.2019. Opublikowano: 31.12.2019.

Problematyka łemkowska, podobnie jak kwestia tożsamości kulturowej, od kilkunastu lat pozostają w zainteresowaniu wielu badaczy. Stanowią one pola badawcze dla uczonych reprezentujących czasami bardzo odległe ob-szary dociekań naukowych. Liczba, i to stale rosnąca, opracowań poświęco-nych tym zagadnieniom świadczy o tym, że pozostają one tematami, jak do tej pory, niewyczerpanymi, stawiającymi wciąż nowe wyzwania, pobudza-jącymi do ref leksji. W centrum wszelkich dociekań nieodmiennie pozostaje jednak człowiek - w swojej zmienności, wielobarwności, wielopłaszczyzno-wości trudny do jednoznacznego zdefiniowania. I to On sprawia, że wszelkie prowadzone badania, okazuje się, są ciągle tylko (albo aż) dążeniem do celu

– poznania Innego, Drugiego, Obcego. Andrzej Bobkowski pisał, że „być Polakiem to naprawdę nieraz pie-

kielnie skomplikowane” (Niewiara 2009: 9). Na czym polega owo „piekielne skomplikowanie” próbowała zrozumieć cytująca go na kartach swojej ob-szernej monografii Aleksandra Niewiara. Z kolei Jolanta Tambor dociekała tego, „co mieszkańcy Górnego Śląska myślą i komunikują na zewnątrz na temat swojego pochodzenia, swojej przynależności narodowej i etnicznej, kontaktów z przedstawicielami innych kultur, na temat historii i jej wpływu na dzisiejszy stan świadomości, i wreszcie ich stosunku do języka” – innymi słowy – dociekała istoty bycia Ślązakiem (Tambor 2008: 10). Inaczej, ale prze-cież również zmierzając w tym samym kierunku, postępował Jerzy Bart-miński, który wraz z zespołem etnolingwistów lubelskich rekonstruował obrazy Polaka, Rosjanina, Ukraińca, Żyda (zob. Bartmiński 2006) itd. Samo-świadomość tak siebie, jak i innych jest jedną z nieprzemijających wartości w ludzkim świecie.

Zgodnie ze stwierdzeniem Hannah Arendt: „Tego, kim ktoś jest lub był, możemy dowiedzieć się dopiero, poznając historię, której bohaterem jest on sam” (Arendt 2000: 205). Szukając odpowiedzi na pytanie: Co to znaczy być Łemkiem? odwołałam się zatem do narracji biograficznych przedstawicieli grupy łemkowskiej. Respondenci, którzy udzielili mi wywiadów narracyj-

Słowa kluczowe:etnolingwi-styka, historia mówiona, Łemkowie, tożsamość, wypędzenia


nych stanowią specyficzną grupę informatorów. Jest to pokolenie odcho-dzące, określane przeze mnie mianem „pokolenia przejściowego”, bowiem jego doświadczenie biograficzne objęło swym zasięgiem czasowym kilka przełomów dziejowych. Pokolenie to wyrastało w micie demokracji mo-narchii austro-węgierskiej, przyglądało się budowaniu II Rzeczypospolitej, widziało upadek państwa, doświadczyło działania systemów opresyjnych, a zanim osiągnęło stan względnej stabilizacji społeczno-materialnej, stało się uczestnikiem największego kataklizmu w najnowszej historii własnej wspólnoty – ekspatriacji.

Przyjętym w przedstawionych tu badaniach założeniem podstawowym, które może wydawać się wręcz banalne, było uznanie, że jeżeli człowiek mówi o czymś, to znaczy, że jest to dla niego istotne (por. Niewiara 2009: 10), a z drugiej strony – istnieje tylko to, co zostało wypowiedziane i/lub na-zwane. Zadanie, jakie sobie postawiłam, polegało w pierwszej kolejności na odkodowaniu zawartych w testach wywiadów wyobrażeń, sądów i wartości współtworzących świat mentalny narratorów, a następnie na wskazaniu ich wzajemnych relacji, uporządkowaniu i pogrupowaniu w większe struktury. Taka postawa badawcza wpisuje się w podejście etnolingwistyki kognityw-nej, która w sposób całościowy patrzy na język i myślenie, zawsze w centrum umieszczając człowieka. „Podkreśla się – pisze Jerzy Bartmiński (2018: 16), próbując wskazać istotę lubelskiej etnolingwistyki kognitywnej (LEK) – że skoro język i kultura to fenomeny par excellence ludzkie, ich badanie osta-tecznie zawsze prowadzi do człowieka. LEK jest więc w najgłębszej swojej istocie nastawiona na człowieka, na jego mentalność”. Z podobnych założeń wyrasta również wcześniejsza teoria – językowego obrazu świata (JOS), która była jedną z ważnych inspiracji koncepcji LEK. Według J. Bartmińskiego: JOS jest zawartą w języku interpretacją rzeczywistości dającą się ująć w postaci zespołu sądów o świecie, o ludziach, rzeczach, zdarzeniach (Bartmiński 2010: 155-178). JOS jest więc z natury obrazem subiektywnym, zbiorem przekonań i wartości rekonstruowanych z punktu widzenia konkretnego podmiotu, który zawsze uwikłany jest w jakiś czas i historię. Elementami JOS z kolei są stereotypy, definiowane jako „kolektywne wyobrażenia ludzi i rzeczy, także zdarzeń, obejmujące cechy traktowane jako ‘normalne’” (Bartmiński 2010: 155-178). Ostatecznie różne punkty widzenia i stereotypowe oglądy zjawisk składają się na całościowy obraz świata. „Wiarygodny opis tego obrazu musi także wziąć pod uwagę całość poznawczo-kulturową, rozdzielaną wyłącznie w celu rozpoznania szczegółowych, ‘punktowych zjawisk’ – zauważa Adam Głaz (2017: 51), odwołując się też do koncepcji poznania kulturowego Hutchinsa (1995) i Sharifiana (2011).

Materię badawczą stanowił dla mnie zbiór tekstów wywołanych, na które składały się zapisane, niekiedy wielogodzinne (najdłuższa relacja obejmuje 4 godziny nagrania), narracje łemkowskie. Wyzyskane wywiady


łemkowskie należą do typu narracji określanych mianem historii mówionej (oral history), są to relacje wspomnieniowe uzyskane od świadków historii w celu ich zarchiwizowania.

Materiał, zebrany zarówno w trakcie wywiadów przeprowadzanych przeze mnie osobiście, jak i znajdujących się w zbiorach Archiwum Łem-kowskiego Zespołu Pieśni i Tańca „Kyczera” w Legnicy, posłużył do poszu-kiwania odpowiedzi na pytania dotyczące łemkowskiej tożsamości, która

– jak każda tożsamość etniczna – wyraża się poprzez język, jak i zanurze-nie w tkance historyczno-kulturowej. Wyłania się stąd subiektywny obraz przeszłości wydobyty przy pomocy narzędzi z obszaru nauk filologicznych i socjologicznych, będący podstawą przedstawionej tu próby odtworzenia stanu tożsamości językowo-kulturowej. W zarejestrowanych wywiadach zaznacza się niezwykle silnie obecność autentyczności emocjonalnej, co stanowi o wyjątkowej wartości relacji.

Na korpus wykorzystanego materiału składają się 34 wywiady nar-racyjne, z czego 17 zarejestrowano w języku polskim, pozostałych zaś 17 w etnolekcie łemkowskim. Dwujęzyczność materiału wynika z dwujęzycz-ności Łemków. W języku polskim, którym się posługuje pokolenie przej-ściowe (dziś nawet częściej, jak samo zaznacza, niż własną mową), bardzo mocno zaznaczają się wpływy etnolektu. Etnolekt Łemków na obczyźnie jest również mocno przesiąknięty wpływami polskimi. Wywiady te zostały zarejestrowane w latach 1999-2000. Wszystkie relacje zostały utrwalone na nośnikach elektronicznych. Wywiady przeprowadzono z osobami urodzo-nymi w latach 1914-1940.

Analizowane opowieści usytuowane zostały w obrębie szerokiego tła historycznego. Przywołanie i omówienie faktów, wydarzeń historycznych, o których mówią respondenci jest niezbędne, by zrozumieć działanie Historii jako katalizatora problematyzowania kwestii tożsamościowej. Osadzenie w kontekście historycznym jest ważnym elementem w perspektywie po-znawczej, dotyczącej tematu (wywiad biograficzny), sposobu jego prezentacji, odtworzenia systemu wartości, zrekonstruowaniu doświadczeń języko-wo-kulturowych, obrazów INNYCH i SIEBIE. Na linii wyznaczonej przez historię, przez dzieje umieszczane są zatem trajektorie i punkty zwrotne. Kontekst historyczny rozumiany jest przeze mnie jako swoista kategoria in-terpretacyjna, tworzona z jednej strony poprzez wpisanie w tło historyczne, uwikłanie historyczne jednostki, wyznaczana wspólnotą doświadczeń histo-rycznych dzielonych z innymi społecznościami, ale również tworzony przez następne pokolenia. Kluczem do wytypowania uczestników wywiadu były jednorodne doświadczenia życiowe powalające na wskazanie tożsamego dla całej grupy miejsca pamięci (węzły pamięci) w ujęciu francuskiego historyka Pierra Nory, dla którego historia była w dużej mierze historią symboliczną i toczyła się w obrazach rekonstruowanych w pamięci. „Miejsca pamięci” to


pewne ślady historii, które z upływem czasu zaczęły odgrywać rolę symboli współtworzących poczucie wspólnoty członków danej społeczności; symboli składających się na tożsamość zbiorową jednostek. „Miejsca pamięci” z jed-nej strony istnieją w sposób konkretny, geograficzny, ale również w sposób symboliczny, są nimi postacie historyczne i fikcyjne, święta, obrzędy, utwory literackie itd. Wszystkie te elementy łączy jedna przestrzeń symboliczna, dodajmy – zakorzeniona w zbiorowej pamięci wspólnoty (Nora 1997).

Narratorzy są nosicielami pamięci grupowej określonej społeczności etnograficzno-etnicznej w perspektywie danej generacji. W kształtowaniu tej tożsamości bardzo ważną rolę należy przypisać pamięci etnograficzno-etnicznej (kulturowej), na którą składa się świadomość wspólnej historii, a więc świadomość ciągłości dziejów, budząca u członków wspólnoty za-angażowanie emocjonalne (zob. Łastowski 2007: 279-307). O stosunku do własnej przeszłości świadczą wspomnienia dawnych wydarzeń, opowieści z przeszłości, którymi seniorzy dzielą się ze swoimi potomkami, ale i ze-wnętrzne znaki pamięci realizowane we wszelkiego rodzaju przedsięwzię-ciach: budowie pomników, organizowaniu wystaw, festiwali kultury itd. Pamięć, a zwłaszcza jej rekonstruowanie, jest również niezwykle istotna z punktu widzenia dociekania istoty tożsamości etnicznej.

Odtwarzając samoświadomość Łemków, podjęłam próbę odpowiedzi na pytania:

1) co determinuje ich – Łemków postawy, wybory, zachowania?2) jak postrzegają Innych?3) czym się różnią od Innych?4) dlaczego są Obcy dla Innych?

Katalog postawionych przeze mnie pytań/zagadnień czerpie z zestawień opracowanych przez J. Bartmińskiego i A. Niewiarę (Bartmiński 2012: 208-225; Niewiara 2009: 19). Wspólne dla obojga badaczy pozostają kwe-stie związane z identyfikacją (kim jesteś?), miejscem (gdzie mieszkasz?), konceptualizacją czasu wspólnotowego (jaki jest twój czas?). A. Niewiara pyta o autostereotyp (jaki jesteś?), J. Bartmiński zaś o heterostereotypy (jacy są ludzie, z którymi wchodzisz w relacje społeczne?). W wyniku pro-wadzonych przeze mnie badań nad tożsamością Łemków, które obejmo-wały swym zasięgiem różne grupy generacyjne, doszłam do wniosku, że należy umieścić katalog pytań tożsamościowych w obrębie szerokiego tła społeczno-historycznego. Okazało się bowiem, że nie tylko okoliczności zewnętrzne wydobywają odpowiedzi na zadane pytania, ale – co wydaje się niezwykle ważne – również determinują postawy tożsamościowe, decydują o doborze elementów i ich hierarchizacji w obrębie łemkowskiego uniwer-sum tożsamościowego.

Małe, odrębne kulturowo grupy, jak Łemkowie, bardzo wyraźnie po-strzegają ludzi wokół siebie przez pryzmat ich przynależności do innych


grup: przynależności narodowo-etnicznej, etnograficznej czy religijnej. Jest to sposób na porządkowanie świata, pozycjonowanie swojej grupy w tym świecie, sposób konfrontacji z zastanym światem, w którym są INNI, wpisujący się w relacje MY-ONI. Wyzyskanie opozycji MY-ONI, będącej naj-bardziej podstawową kategorią opisu świata, w procesie identyfikacji et-nograficzno-etnicznej uznaję za strategię identyfikacyjną o charakterze wspólnotowym. W zrekonstruowanych obrazach INNYCH, jak w zwiercia-dłach przeglądają się właściwie sami Łemkowie. Ze stwierdzeniem Marka Zaleskiego, że „swojskość” i „inność/obcość” to „kluczowe zagadnienie opisu doświadczenia będącego udziałem żyjących w kulturze wielonarodowego pogranicza” (Zaleski 1994: 126) nie sposób polemizować. ‘Swojskość’ i ‘inność/obcość’ to kluczowe zagadnienie opisu doświadczenia będącego udziałem żyjących w kulturze wielonarodowego pogranicza. Dawna etniczna Łem-kowszczyzna, choć stanowiła zwarty teren osadnictwa rusińskiego, była za-wsze terenem pogranicza kulturowego. Opowiadając o INNYCH, Łemkowie charakteryzują samych siebie. Za każdym razem jest to swoista konfrontacja różniących się kulturowo społeczności. MY – to Łemkowie, a kim są ONI? ONI to nie MY, czyli INNI (inni kulturowo). Kim są INNI:

INNI mogą być SWOI, a więc nie-OBCY;

INNI mogą być OBCY, więc nie-SWOI;

INNI mogą być NIEZNANI.

Podstawowa jednostka językowa podlegająca analizie to wyrażenia cytatowe z wywiadów łemkowskich, będące dosłownym przytoczeniem relacji. Definiuję je jako najmniejsze możliwe do wydzielenia cząstki tekstu, które sygnalizują istnienie cechy uwidaczniającej się poprzez działanie pod-miotu. Wyrażenie cytatowe wskazuje tym samym na najmniejszy możliwy do opisania kontekst sytuacyjny konotujący daną cechę. M. Brzozowska traktuje wyrażenie cytatowe jako „minimalny kontekst, który komunikował jedną w rzadkich przypadkach dwie cechy semantyczne” (Brzozowska 2006: 39), natomiast w proponowanym ujęciu wyrażenia cytatowe pozwalają na uchwycenie również dynamiczności cechy. Dzięki temu można stwierdzić nie tyle, co to znaczy być jakimś, ile w jakich sytuacjach ta cecha się przeja-wia. Przyjęłam, że co najmniej trzykrotne potwierdzenie cechy w wyraże-niach cytatowych stanowi o jej nieprzypadkowości występowania. Grupa wyrażeń cytatowych pozwala na sformułowanie deskryptora, który stanowi metawyrażenie nadbudowane nad wyrażeniami bliskimi/tożsamymi semantycznie

(Brzozowska 2006: 39). Pogrupowane w wiązki cechy/deskryptory składają się na jednostki złożone semantycznie, czyli syndromy. Syndromy mogą być przedstawiane na przykład w ujęciu generacyjnym, społecznym czy


etnicznokulturowym. Z kolei każda z cech w obrębie syndromu może być charakteryzowana poprzez określony aspekt (np. społeczny, religijny itd.).

Obrazy, czyli – innymi słowy – wyobrażenia INNYCH odtwarzane dzięki narracjom, realizują konkretne syndromy INNYCH. Syndrom ozna-cza zespół cech charakteryzujących dany obiekt (INNEGO). Syndromy dają się pogrupować w profile. Wykorzystałam pojęcie syndromu i profilu wy-pracowane przez lingwistów lubelskich: J. Bartmińskiego, Irenę Lappo i Urszulę Majer-Baranowską (Bartmiński, Lappo, Majer-Baranowska 2002: 105-152). Za profile uznają oni odmiany danego obrazu konstruowane z okre-ślonej perspektywy, „kreowane przez jakiś czynnik dominujący, dominantę. (…) różne profile są tworzone na zasadzie derywacji wychodzącej od bazo-wego zespołu cech semantycznych w obrębie znaczenia” – otwartego ze-społu cech (Bartmiński, Lappo, Majer-Baranowska 2002: 145). Wyłaniające się z narracji obrazy SIEBIE i INNYCH mieszczą się w kategorii potocznych wyobrażeń, prowadzą bowiem do rekonstrukcji autostereotypu ŁEMKA i łemkowskiego stereotypu INNYCH. Stanowią więc sumę odpowiedzi na pytanie: „jaki jest X?”.

Narracje składają się z dwóch zasadniczych przedziałów czasowych: pierwszego przebiegającego do momentu wysiedlenia w 1947 r. i drugiego

– od momentu wysiedlenia. Sam moment wysiedlenia ludności z Łem-kowszczyzny stanowi osobny epizod czasowy. INNI, czyli nieŁemkowie, pojawiają się w każdej z sekwencji czasowych. W odniesieniu do czasów etnicznej Łemkowszczyzny są to przede wszystkim Polacy – ich najbliżsi sąsiedzi, dalej: Żydzi i Cyganie. Na obczyźnie ONI to wszyscy pozostali przesiedleńcy, a więc Ukraińcy (bo „my nie jesteśmy Ukraińcami”), Polacy – z Polski centralnej (którzy przyjechali „z góry”, czyli z północy, patrząc na mapę Polski) i ci zza Buga („niby też Polacy”), Niemcy (pozostali jeszcze po wojnie). Żydzi należą do obrazów przeszłości, wpisani są w łemkowski pejzaż międzywojnia, a ich historia kończy się wraz z nadejściem II wojny światowej. Podobnie jest z Cyganami. W okresie wojennym pojawiają się natomiast Niemcy, Rosjanie, Ukraińcy. Ukraińcy wzmiankowani są również we fragmentach dotyczących czasów przedwojennych, choć zdominowali w zdecydowany sposób okres powysiedleńczy. Z wszystkich grup narodo-woetnicznych jedynie Polacy towarzyszą Łemkom przez wszystkie fazy ich historii osobistej: przedwojenną, wojenną i powojenną. INNI pojawiają się w narracjach w planie bliskim i planie dalekim2. Plan bliski obejmować bę-dzie najbliższe otoczenie informatora, miejsca mu znane, odwiedzane przez

2 Terminów: „plan bliski”, „plan daleki” (dodatkowo „plan najbliższy”), choć w innym rozumieniu, używa A. Niewiara (2009), rekonstruując samoświadomość Polaków na przestrzeni XVI-XX w. Pozwalają jej one opisać warianty polskich map mentalnych. Plan daleki oznacza świat obejmu-jący swym zasięgiem Europę, plan bliski – odnosi się do obszarów najbliższych Polsce (nie tylko w sensie geograficznym), plan najbliższy – skupia się na obszarach realizujących się w sferach prywatnych.


niego (rodzinna wioska, sąsiednie wioski i miasteczka). INNI będą więc zna-nymi mu konkretnymi osobami. Plan daleki wykraczać będzie poza rejony znane, często poza Łemkowszczyznę, będzie się odnosił do całego kraju, do Polski. INNY będzie wówczas postrzegany w kategorii reprezentanta okre-ślonej zbiorowości, nosiciela cech stereotypowych. Żydzi, Cyganie i Polacy w planie bliskim to sąsiedzi Łemków.

Łemkowie dzielą ludzi w sposób schematyczny na dobrych i złych. Jest to typowy ludowy prostolinijny sposób postrzegania rzeczywistości, w której człowiek postępuje uczciwie albo nieuczciwie. Z narracji przebija ludowa mądrość życiowa, oparta na wiedzy potocznej, wsparta określonymi cechami charakteru, jak: dobroduszność, łatwowierność, pozytywne nastawienie do życia i bliźnich. Na postawę tę wpływ miała niewątpliwie również chłopska religijność.

Żydzi charakteryzowani są przez trzy syndromy. Najważniejszym syn-dromem jest „Żyd handlarz” – był pośrednikiem handlowym, miał zmysł do handlowania, posiadał sklep, był człowiekiem interesu, ale był również chciwy i potrafił bogacić się na cudzej krzywdzie. „Żyd sąsiad” był postrze-gany jako dobry człowiek, który pomagał Łemkom, żył z nimi w zgodzie i dawał im zarobić. „Żyd religijny” z kolei rygorystycznie przestrzegał sza-batu, nie jadł wieprzowiny i modlił się długo i gorliwie.

„Cygan sąsiad” to sąsiad pracujący, dobry, niekradnący, porządny, choć biedny, chodzący po żebrach. Cyganie zarabiali grą na instrumentach (syn-drom „Cygan muzykant”) i byli kowalami („Cygan kowal”). Podawane cechy z jednej strony wpisują się w powszechne potoczne wyobrażenie o Cyga-nach, a z drugiej strony wchodzą w polemikę ze stereotypem. „Łemkowski” Cygan pracował – wiedza potoczna mówi, że Cygan nie pracuje; nie kradł – a „normalnie” kradnie; był porządnym człowiekiem – stereotypowy taki nie jest, skoro wiadomo, że kradnie i nie pracuje.

Historia wzajemnych kontaktów polskołemkowskich dzieli się bardzo wyraźnie w narracjach na dwa okresy: okres przed wojną i okres po wojnie. Syndromy „Polaka przedwojennego” i „Polaka powojennego” bardzo się od siebie różnią. Zasadnicza różnica polega na znacznie większym nacechowa-niu negatywnym obrazu Polaka powojennego. Polak przedwojenny w planie bliskim realizował syndrom „Polaka sąsiada”. W tym ujęciu „Polak sąsiad”: czasami chodził do cerkwi, był wyznania rzymskokatolickiego, przyjaźnił się z Łemkami, można się było z nim dogadać, nie robił Łemkom krzywdy. W planie dalekim pojawił się syndrom „Polaka niesąsiada” (realizującego schemat INNY i OBCY, NIE SWÓJ). Dowodzi on braku konf liktu etnicznego łemkowskopolskiego. Polak przedwojenny w oczach Łemków nie stanowił dla nich zagrożenia. Syndrom ten, zapewne ze względu na usytuowanie właśnie w planie dalekim, obejmuje tylko dwa deskryptory: był tolerancyjny i nie robił krzywdy.


Relacje polskołemkowskie uległy gwałtownemu pogorszeniu po zakoń-czeniu II wojny światowej. Ewolucja obrazu Polaka przebiegała w kierunku zdecydowanie negatywnym. Sytuacja historyczna (wysiedlenie Łemków) wykreowała nowy syndrom: „Polaka korzystającego z łemkowskiej krzywdy” (zagrabił połemkowski majątek, mścił się na swoich dawnych sąsiadach, zabrał Łemkom cerkiew). W odniesieniu do okresu powysiedleńczego dla Łemków rozszerzeniu ulega zakres nazwy narodowej „Polak”. Jeśli przed wysiedleniem kategoria przynależności narodowej była bardzo wyrazista, to w realiach powojennych ta wyrazistość uległa zatarciu. Łemkowie na obczyźnie spotkali Polaków, którzy nie przystawali do ich obrazu Polaka ukształtowanego i wyniesionego z realiów dawnej Galicji. Dla okresu tuż-powojennego charakterystyczna jest ogromna mobilność ludności polskiej wywołana pojałtańskimi zmianami granic. Tak zwane oficjalnie „Ziemie Odzyskane”, nieoficjalnie zaś nazywane „Polskim Dzikim Zachodem” (zob. Halicka 2015), stały się terenem nowej kolonizacji, zostały zasiedlone miesz-kańcami poszczególnych rejonów Polski centralnej, ludnością dawnych Kre-sów Wschodnich II RP, reemigrantami m.in. z Jugosławii, Niemiec, Belgii czy Francji, ale również Łemkami. Kategoria „prawdziwego Polaka” odnosiła się do typu, znanego Łemkom z ich własnego przedwojennego doświadczenia. W nowych – powysiedleńczych, społeczno-politycznych – realiach, pojawił się syndrom „Polaka sąsiada na obczyźnie”. Syndrom niezwykle skompliko-wany co do układu elementów, w którym skupiły się cechy sprzeczne: pozy-tywne i negatywne. Tak spolaryzowany układ oddaje istnienie stanu przej-ściowego we wzajemnych relacjach łemkowskopolskich. Łemko przybył na

„Zachód” z silnym poczuciem krzywdy, toteż niektóre zachowania polskich sąsiadów stan ten utrwalały, inne zaś skłaniały do weryfikacji dotychcza-sowych odczuć. Po raz kolejny dała znać o sobie prosta ludowa prawda, że człowiek jest dobry i zły. „Polak sąsiad na obczyźnie”: źle traktował Łemków, wykorzystywał, okradał z dobytku, robił na złość, choć potrafił im pomagać, przezywał od Ukraińców, UPA, ale też „złemczał się” (nauczył się mówić po łemkowsku), wspólnie z Łemkami chodził na zabawy. Relacje polsko-łemkowskie, silnie napięte w pierwszych latach, z biegiem czasu okrzepły, ustabilizowały się. Zatarła się nieuzasadniona, podsycana propagandowo wrogość „do Ukraińca”, przeważyła, wynikająca z codziennej sąsiedzkiej obecności, wzajemna akceptacja. W tle pozostało wprawdzie poczucie kul-turowej odmienności, ale przeważyło zdroworozsądkowe spojrzenie na dru-giego człowieka. Po latach spędzonych w nowym otoczeniu „Polak sąsiad” to ten, który po prostu żyje pośród Łemków. Wzajemne relacje powróciły do stanu normalności, a zatem rekonstrukcja łemkowskich syndromów Po-laka wskazuje na dokonujący się proces zakorzeniania na obczyźnie, a tym samym modyfikację strategii tożsamościowej. Łemko to też ten, który ma nowych sąsiadów Polaków i są to wprawdzie INNI, ale nie OBCY, lecz SWOI.


Obraz własny Łemka, który wyłania się z wywiadów, zaskakuje wie-lopłaszczyznowością oraz liczbą składających się nań elementów. Obfitość uzyskanego w ten sposób materiału spowodowała, że pojawiła się potrzeba wprowadzenia dwustopniowego opisu – obok syndromów o charakterze podstawowym, bazowym zostały uwzględnione jednostki szersze, czyli profile. Są to: 1/ Łemko – ofiara historii (profil grupujący syndromy i de-skryptory z punktu widzenia określonej generacji, w omawianym materiale będzie to pokolenie przejściowe; doświadczenia historyczne każdej gene-racji Łemków będą w odmienny sposób wypełniać ów profil); 2/ Łemko – wygnaniec (rekonstruowany z punktu widzenia wysiedleńca, czyli osoby zmuszonej do zmiany miejsca zamieszkania); 3/ Łemko etniczno-kul-turowy (odnoszący się do osoby należącej do społeczności łemkowskiej), wreszcie 4/ Łemko – chłop (zespół syndromów zebranych z punktu wi-dzenia mieszkańca wsi, z punktu widzenia chłopa). Te cztery wzajemnie się uzupełniające, ale też przecinające się zbiory określają łemkowską toż-samość kulturową. Profil generacyjny bierze pod uwagę doświadczenie bio-graficzne danego pokolenia. Narratorzy byli świadkami, a zarazem uczest-nikami przełomowych wydarzeń dziejowych, „wielka historia” działa się obok ich chyży. Profil kulturowy wskazuje na tradycyjne cechy kulturowe Łemków, czyli język, religię, obyczaje, strój, wierzenia ludowe – zabobony. Ostatni z profili – społeczny – dotyczy relacji międzyludzkich Łemków z innymi społecznościami, czyli innymi słowy cech wynikających z rela-cji łemkowsko-polskich, łemkowsko-ukraińskich, łemkowsko-żydowskich, jak i łemkowsko-łemkowskich. Wzorce kulturowe, odzwierciedlone w ję-zyku używanym przez Łemków (etnolekcie/języku polskim), umożliwiają tym samym podjęcie próby odpowiedzi na pytanie o kształt ich kulturowej tożsamości.

Syndromy, czyli określone wiązki cech, mieszczą się w poszczególnych przywołanych tu profilach. Profile nie stanowią bytów samoistnych, gdyż w pewnych obszarach zachodzą na siebie. Syndromem dominującym, obec-nym w obrębie wszystkich trzech płaszczyzn okazał się syndrom „Łemka skrzywdzonego”, będący zarazem najważniejszym elementem profilu Łemko

– ofiara historii. Według celnej obserwacji Ewy Michny, „pamięć o dozna-nych w wyniku przesiedlenia krzywdach moralnych i materialnych jest dziś jednym z podstawowych czynników integrujących społeczność ukraińską i łemkowską w Polsce” (Michna1995: 51), trudno się zatem dziwić, że deskryp-tory składające się na syndrom „Łemka skrzywdzonego” były obecne w każ-dej bez wyjątku relacji.

Łemko był krzywdzony przez historię i przez los. W przekroju gene-racyjnym doświadczał krzywd jako jednostka i jako grupa. Łemko wie, że cierpienie dotykało jego wspólnotę w przeszłości, zetknął się z krzywdą bezpośrednio i boi się powtórzenia złych doświadczeń w przyszłości. Ba-


dana grupa narratorów jest generacyjnie usytuowana na pograniczu epok, stanowi swoiste pokolenie przejściowe − jej doświadczenie biograficzne bowiem mieści się między doświadczeniem pierwszej wojny światowej, a doświadczeniem ekspulsji w 1947 roku. W przypadku generacji responden-tów wybrzmiewa niezwykle wyraźnie poczucie nieustannej, niezawinionej, nieuprawnionej krzywdy.

Konsekwencją przymusowego wysiedlenia okazało się również wyko-rzenienie, radykalne przecięcie więzów łączących Łemka z jego górską „małą ojczyzną”. W ten sposób w obrębie profilu Łemko wygnaniec pojawia się syndrom bazowy „Łemko wykorzeniony”. „Łemko wykorzeniony” to Łemko, który zamiast do cerkwi, chodzi do kościoła. ŁEMKO – WYGNANIEC to jed-nak przede wszystkim „Łemko tęskniący” tęskniący za młodością spędzoną na Łemkowszczyźnie, tęskniący za utraconą „małą ojczyzną”, tęskniący do swojej ziemi, chcący na swojej ziemi umrzeć, dumny i przywiązany do swojej rodzinnej wsi. W łemkowskiej tęsknocie mieści się i wspomnienie utraco-nej „małej ojczyzny”, rodzinnej wsi i ziemi, i żal za bezpowrotnie utraco-nym tamtym porządkiem społecznym. Przy czym, co warto dopowiedzieć, nie tyle chodzi o realia administracyjno-urzędowe, co o inny rytm życia i ówczesne wartości. Łemko wygnaniec sytuuje się również w opozycji do następnego zbioru syndromów − profil ten można opisać jako Łemko et-niczno-kulturowy. Mieszczą się w tym profilu syndromy „Łemko radosny” (śpiewał, m.in. w chórze cerkiewnym, tańczył) i „Łemko religijny” (greko-katolik albo prawosławny, uczestniczący w praktykach religijnych, wierny swojemu wyznaniu). Charakterystyczna dla społeczności wiejskich była skłonność do ulegania najrozmaitszym zabobonom, niekiedy wpisanym w ludowy zwyczaj. Stąd cecha „bycie zabobonnym”, będąca uzupełnieniem obrazu Łemko etniczno-kulturowy.

Za niezwykle istotny dla określenia obszaru łemkowskiej etniczno-kul-turowej przynależności należy uznać deskryptor „bycie stamtąd”. Wiąże się on z łemkowskim terytorium etnicznym i zawiera w sobie określenie horyzontu przestrzeni geograficznej, w gruncie rzeczy bardzo zawężonej, ograniczonej z reguły do najbliższych okolic. Wynikało to z faktu, że rodzina, krewni bliżsi i dalsi, zazwyczaj zamieszkiwała w obrębie jednej wioski.

Ostatni, czwarty profil gromadzi deskryptory określające łemkowską tożsamość etniczno-kulturową z punktu widzenia mieszkańca wsi, a zatem z perspektywy chłopa. Ten profil to Łemko chłop, przy czym składające się nań syndromy odnoszą się nie tylko do stanu zasobności gospodarującego na swoim Łemka, ale i postawy życiowej z tym związanej. „Łemko pragma-tyczny” to następny syndrom oddający chłopską filozofię życiową. Mieści się w nim interesowność, np. małżeństwo było traktowane jako transakcja, sposób na ustawienie się w życiu, a uczucia schodziły na dalszy plan. Prag-matyzm graniczył niekiedy z naiwnością, jak wówczas, gdy Łemko myślał,


że gdy zmieni wyznanie na „polskie”, czyli rzymskokatolickie, to nie zostanie wysiedlony. Pragmatyzm przejawiał się również wtedy, gdy Łemko po wysie-dleniu uznał, że po przejściu na emeryturę może bez negatywnych dla sie-bie konsekwencji uczestniczyć w praktykach religijnych swojego wyznania. W profilu tym mieści się również syndrom „Łemka cwaniaka”, który oszu-kiwał i był sprytny. Kolejne syndromy bazowe, składające się na ostatni profil, ukazują rozwarstwienie społeczne obecne na przedwojennej Łem-kowszczyźnie. Łemko chłop to zarówno „Łemko biedujący”, jak i „Łemko bogaty”. Z analizy uzyskanych z relacji wynika, że poza ciągiem deskrypto-rów układających się w wyraziste profile pozostają syndromy i pojedyncze cechy (deskryptory), będące elementami współtworzącymi tożsamościowy autostereotyp Łemka. Ich katalog otwiera syndrom świadczący o łemkow-skiej zdolności do podejmowania ryzyka, nawet w obliczu realnego zagro-żenia – jest nim „Łemko odważny”, który przeciwstawił się okupantowi niemieckiemu, pomagając Żydom czy biorąc czynny udział w działaniach konspiracyjnych.

Wyłaniający się z wykorzystanych relacji obraz Łemka, składający się na jego etniczno-kulturową tożsamość, obejmuje aż 68 deskryptorów. Aż 38% z nich sytuuje się w obrębie profilu Łemko ofiara historii, przy czym 34% wchodzi w skład syndromu bazowego „Łemko skrzywdzony”. Zastanawia, że ponad pół wieku po ostatnim najbardziej traumatycznym wydarzeniu, wpisanym w łemkowską historię – przymusowym wysiedleniu, ten syndrom wciąż najmocniej definiuje łemkowską tożsamość. Tłumaczy to niewątpliwie generacyjna płaszczyzna pokolenia respondentów, które określiłam mianem przejściowego. Okazuje się, że w świadomości łemkowskiej bagaż tragicz-nych historycznych losów wspólnoty odcisnął tak trwały ślad, że do tej pory nie zniwelowały go późniejsze – również pozytywne – doświadczenia.

W 1947 dla Łemków ich historia pomimo kilku wcześniejszych drama-tycznych przyspieszeń (np. Thalerhof – który był w okresie I wojny świa-towej obozem dla internowanych Łemków czy okres II wojny światowej – por. Wirchniański 2014) biegnąca spokojnym, uporządkowanym rytmem, utraciła linearny charakter, przybierając kształt punktów i przeskoków. Od momentu wysiedlenia w 1947 roku Łemkowie znaleźli się w sytuacji nie-ustannego problematyzowania swojej tożsamości. Wysiedlenia doświad-czyły wprawdzie trzy ówczesne pokolenia: dziadków, rodziców i dzieci, które można scharakteryzować jako odpowiednio pokolenia schodzące, aktywne i wchodzące w życie, ale to najmłodsi rozpoczynali dorosłość zawieszeni pomiędzy tym, co minęło (mijało na ich oczach), i tym, do czego musieli się zaadaptować na obczyźnie. Uznać więc ich należy za klasyczne pokolenie przejściowe. Ich relacje stanowią swego rodzaju tożsamościowy palimpsest. Okazały się one tekstem semantycznie wielowarstwowym, toteż po „zeskro-baniu” jednostkowej opowieści biograficznej jak na starożytnym pergaminie


dało się odczytać tożsamość kulturowo-językową zbiorowości. Uzewnętrz-niła się ona w wielosegmentowej opozycji my – oni: Łemkowie – Inni (Po-lacy, Żydzi, Cyganie, Ukraińcy, Ruscy, Niemcy), nasza wiara – wasza wiara (grekokatolicy/prawosławni – katolicy), nasza mowa – wasza mowa (łem-kowszczyzna/polszczyzna Łemków – polszczyzna Polaków), nasza ziemia

– wasza ziemia (Łemkowszczyzna – obczyzna), nasza historia (Thalerhof/akcja „Wisła”) – wasza historia. Po raz kolejny okazało się, że bez znajomo-ści kontekstu historycznego jest niepełne, o ile w ogóle możliwe, dotarcie do istoty łemkowskości.

W relacjach obok wątków oczekiwanych i oczywistych (np. II wojna światowa, wysiedlenie, obczyzna), sięgających do pamięci i doświadczenia osobistego, pojawiły się również przywołania wydarzeń świadczące o prze-jęciu przez narratorów pamięci rodzinnej (czasy monarchii austro-węgier-skiej czy I wojny światowej). Historia oplotła dzieje Łemków zarówno jako społeczności, jak i jednostek – niczym pajęczyna, z której nie byli (nie są?) w stanie się uwolnić. W szczególny sposób kształt ich tożsamości kulturo-wo-językowej wyznaczyła tym samym trajektoria wypędzonego. Zmieniła status ich etnolektu: z poziomu dialektu na mowę pogranicza i – przynaj-mniej dla części – mikrojęzyka literackiego. Bezpowrotnie wykorzeniła ich z ojcowizny, przerywając gwałtownie i w skali masowej ciągłość zasiedlenia w Beskidach, między Popradem a Osławą. Kategoryczność tego wyroku Historii nie zmieni już nawet realna możliwość powrotu na ziemie przod-ków. Łemkowszczyzna etniczna przekształciła się w kategorię historyczną. Łemkom pozostało już tylko do niej tęsknić. Na obczyźnie wyostrzyło się natomiast ich poczucie odrębności religijnej. Jak jednoznacznie wynika z relacji, w pierwszym okresie powysiedleńczym przynależność do kościoła rytu wschodniego stała się swoistą metonimią tożsamości, uzewnętrznianą w kontaktach zewnętrznych: leksemy grekokatolik/prawosławny to niejako synonimy etnominu Łemko. Diametralnie zmienił się też społeczny kon-tekst funkcjonowania Łemków: ze zwartej wiejskiej społeczności stali się oni społecznością diasporyczną, a sąsiedzkie kontakty z Innymi zawęziły się do Polaków.

„Wysiedlenie – 1947” stanowi w strukturze narracji łemkowskich zda-rzenie przełomowe. Dzieli wszystkie indywidualne historie (jak i w ogóle dzieje Łemków) na dwie części (przed rokiem 1947 i po nim). Wprowadza, używając terminologii Fritza Schützego (1997), trajektorię wypędzonego. Od tego momentu Łemkowie nie mają wpływu na swój los, za nich zdecydowano i o opuszczeniu ojcowizny, i o tym, gdzie będzie ich nowy dom. To jest wyda-rzenie, które wprowadziło poczucie braku bezpieczeństwa, dezorganizację i cierpienie. Od tego momentu zaczyna się ref leksja nad własną identyfi-kacją, ref leksja wymuszona okolicznościami historycznymi. Poszczególne przedstawione wcześniej strategie łemkowskiej autoprezentacji nie dają się


w prosty sposób przypisać konkretnym przedmiotom narracji. Są obecne, z różnym natężeniem, w każdej sekwencji opowieści.

Łemko symbolicznie wciąż związany z Łemkowszczyzną, jednocześnie na nowo buduje swą kulturowo-językową tożsamość na obczyźnie. Prze-siedlenie tym samym dokonało istotnej korekty w zbiorze elementów, skła-dających się na łemkowskie uniwersum tożsamościowe. Pojawił się przede wszystkich, charakterystyczny dla nowych warunków, dualizm widoczny we wszystkich płaszczyznach życia, co przekłada się na: dwureligijność (obok obecności w kościołach greckokatolickich czy prawosławnych pojawiło się uczestnictwo w nabożeństwach katolickich), dwujęzyczność (obok posłu-giwania rodzimym etnolektem znacznie częściej niż na Łemkowszczyźnie sięga się po język polski), biterytorialność (obok ojczyzny prywatnej w wy-miarze symbolicznym Łemko zyskał realną małą ojczyznę na obczyźnie).

Nie istnieje odpowiedź ostateczna na pytanie o tożsamość kulturowo-językową. Każdy nowy wątek może stanowić jej dopełnienie. Tożsamość, każda tożsamość, ma to do siebie, że jest procesem, więc i określenie jej kształtu zmienia się pod wpływem czasu i kontekstu.

BibliografiaArendt, Hannah 2000: Kondycja ludzka. Tłum. Anna Łagodzka. Warszawa:

Fundacja Aletheia.Bartmiński, Jerzy 2010: Pojęcie językowego obrazu świata i sposoby jego operacjo-

nalizacji. W: Przemysław Czapliński, Anna Legeżyńska, Marcin Telicki (red.) 2010: Jaka antropologia literatury jest dzisiaj możliwa. „Poznańskie Studia Polonistyczne”, Poznań, 155-178.

Bartmiński, Jerzy 2012: Zmiany obrazu Polaka we współczesnym dyskursie publicznym. W: Językowe podstawy obrazu świata. Lublin: Wydawnictwo Uniwersytetu Marii Curie Skłodowskiej, 208-225.

Bartmiński Jerzy, Lappo Irina, Majer-Baranowska Urszula 2002: Stereotyp Rosjanina i jego profilowanie we współczesnej polszczyźnie. Etnolin-gwistyka 14, 105-152.

Bartmiński, Jerzy 2018: O profilowaniu pojęć z punktu widzenia etnolingwistyki kognitywnej. W: Agnieszka Libura, Daria Bębeniec, Hubert Kowalewski (red.) 2018: Dociekania kognitywne. Kraków: Universitas, 13-48.

Bartmiński, Jerzy (red.) 2006: Język – wartości – polityka. Zmiany w rozu-mieniu nazw wartości w okresie transformacji ustrojowej w Polsce. Raport z badań empirycznych. Lublin: Wydawnictwo Uniwersytetu Marii Curie-Skłodowskiej.

Brzozowska, Małgorzata 2006: O przebiegu badań nad zmianami w rozu-mieniu nazw wartości 1990-2000. W: Bartmiński, Jerzy (red.) 2006: Język – wartości – polityka. Zmiany w rozumieniu nazw wartości w okresie


transformacji ustrojowej w Polsce. Raport z badań empirycznych. Lublin: Wy-dawnictwo Uniwersytetu Marii Curie-Skłodowskiej, 44-62.

Głaz, Adam 2017: Worldview as cultural cognition. LaMiCuS 1, 34-53.Halicka, Beata 2015: Polski Dziki Zachód. Przymusowe migracje i kulturowe oswa-

janie Nadodrza 1945-48. Kraków: Universitas.Hutchins, Edwin 1995: Cognition in the Wild. Cambridge, MA: MIT Press.Michna, Ewa 1995: Łemkowie. Grupa etniczna czy naród? Kraków: Nomos.Misiak, Małgorzata 2018: Między Popradem a Osławą, Tożsamość językowo-kul-

turowa Łemków w ujęciu etnolingwistycznym. Wrocław: Profil.Niewiara, Aleksandra 2009: Kształty polskiej tożsamości. Katowice: Wydaw-

nictwo Uniwersytetu Śląskiego.Nora, Pierre 1997: Les lieux de mémoire, Gallimard, Paris.Sharifian, Farzad 2011: Cultural Conceptualisations and Language: Theoretical

Framework and Applications. Amsterdam, Philadelphia: John Benjamins.Schütze, Fritza 1997: Trajektorie cierpienia jako przedmiot badań socjologii in-

terpretatywnej, „Studia Socjologiczne” 144 (1): 11-56.Tambor, Jolanta 2008: Mowa Górnoślązaków oraz ich świadomość językowa i et-

niczna, wyd. 2. Katowice: Wydawnictwo Uniwersytetu Śląskiego.Wirchniański, Piotr 2014: Łemkowski Thalerhof. Magury’14, 6-66.Zaleski, Marek 1994: Naprzód w przeszłości, czyli pamięć o wielonarodowej Rzecz-

pospolitej w literaturze po roku 1939. W: Wiesław Władyka (red.) 1994: Inni wśród swoich. Warszawa: Instytut Badań Literackich PAN, 120-129.

Abridgement

What does it mean to be a Lemko? On the questions of identity. Ethnolinguistic approach

This article concerns the questions of identity. It is an attempt to recon-struct this area of the Lemko’s self-knowledge. It offers an answer to the question: What does it mean to be a Lemko? This question was asked to a specific group of respondents. They constitute a passing generation referred to as a “generation in transition”. Their biographical experience covered sev-eral historical breakthroughs. This generation was growing up in a myth of the Austro-Hungarian Empire, observed the creation of the 2nd Republic of Poland, its collapse, and finally experienced two repressive totalitarian systems. Before it reached a relatively stable social and financial position, it had been a participant in the biggest cataclysm in the newest history of this community – expulsion.

Due to the nature of the investigated phenomenon, this work is situated in-between various scientific fields: linguistics, anthropology, sociology, and


ethnology with a broad historical background. However, linguistics together with its variants: ethnolinguistics, sociolinguistics or cognitive linguistics remains the main frame of reference. The goal of this study is to decode im-ages, opinions and values constituting the mental world of the story-tellers as expressed in their words and then to show mutual relationships, order and grouping of these images in bigger structures. The analysed material consists of a collection interviews with the Lemkos, which were recorded and transcribed. They could last several hours, the longest story is 4-hour long. The data was analysed within the philological approach and was based on the procedures used in text linguistics. In this article, I mainly use the conceptual apparatus related to the structure and semiotics of the text. In-terviews with the Lemkos analysed here belong to a narrative type referred to as oral history consisting of recollections of the witnesses collected in order to archive them.

The analysed material was used to investigate questions about the Lem-kos’ identity which (as any ethnic identity) is expressed through language and immersed in the historic and cultural context. As a result, we can observe the emerging subjective image of the past extracted with the use of tools from philology and sociology. This image ref lects the state of preservation of linguistic and cultural integrity. In the recorded interviews, one can find strong emotional authenticity, which constitutes an exceptional value of those stories.

The interviews were selected for analysis if they represented homogene-ous life experiences, which allow us to identify a place of memory (memory nodes) identical for the whole ethnic group. The story tellers are carriers of the collective memory of the ethnographic-ethnic community from a per-spective of a given generation.

Small and culturally distinct groups (such as Lemkos) perceive people around them through their affiliation to other groups: national-ethnic af-filiation, ethnographic affiliation or religious one. It is a way of ordering the world, positioning of our own group in that world, a way of confronting the world surrounding us where there are others entering into a relation-ship we – they.

The self-image of a Lemko arising from the interviews surprises with its multidimensional character and the number of elements constituting it. An abundance of the material created a need to introduce a two-phase description: apart from basic syndromes (basis character) we also take into account larger units – i.e. profiles. They are (1) Lemko – a victim of history (the profile grouping syndromes and descriptors from a point of view of a given generation, i.e. in a generation-oriented approach; (2) Lemko – an exile (reconstructed from the point of view of a displaced person, i.e. a per-son forced to change their place of permanent residence; (3) An ethnic and


cultural Lemko – related to a person belonging to Lemkos’ community; and finally (4) Lemko – a peasant – a group of syndromes collected from a point of view of a village inhabitant; from a peasant’s point of view. These four com-plementary, but also overlapping groups determine Lemkos’ cultural identity. The generation profile takes into account biographic experience of a given generation. Story tellers were witnesses and at the same time participants of crucial historical events, and the “big history” happened right next to their cottage. The cultural profile shows Lemkos’ traditional cultural features – i.e. language, religion, customs, traditional dress, and folk superstitions. The last profile – a social one refers to Lemkos’ mutual relationships with other communities; in other words, it refers to features resulting from Lemko-Polish, Lemko-Ukrainian, Lemko-Jewish or generally Lemko-non-Lemko relationships as well as Lemko-Lemko ones. Cultural patterns ref lected in the language used by Lemkos (ethnolect / Polish language) enable us to try to answer the question about the shape of their cultural identity.

Syndromes – i.e. certain groups of features – are incorporated in the abovementioned profiles. Let us emphasize once again that these profiles are not independent phenomena as they overlap in certain areas. The syndrome of an “aggrieved Lemko” is the dominant one and at the same time the most important element of the profile Lemko – a victim of history. Descriptors constituting the syndrome of an “aggrieved Lemko” were present in all re-lationships with no exceptions.

The Lemko was aggrieved both by the history and the fortune. In his generation profile he experienced bad fortune both as an individual and as a group. He knows that his community suffered in the past and is afraid that this bad experience may repeat itself in the future. The informants come from a generation, which is situated between two eras and is a kind of a generation in transition – its biographic experience covers a period between the First World War and the expulsion in 1947. In case of this generation the feeling of long-standing grievance remains dominant.

In 1947 the history lost its linear character for Lemkos and became a f low of jerks and jumps. Despite some earlier dramatic cases of acceleration (Talerhof, the war), it was only in 1947 (expulsion) Lemkos found themselves in a situation where their identity was continuously problematised. Three generations experienced the expulsion: grandparents, parents and children; and they can be characterised as a passing generation, an active generation and a new generation respectively. The youngest generation started their mature lives suspended between something that passed (which they were witnesses of) and something to which they had to adapt to in a foreign land. Therefore, they should be recognized as a typical generation in transition. Their relationships constitute a kind of an identity-oriented palimpsest. Simi-larly to an ancient parchment – it is possible to read cultural and language


identity of a community after removing the layers of an individual biographic story. It appeared in a form of multiple we – they oppositions: Lemkos – Others (Poles, Gypsies, Ukrainians, Russians, Germans); our faith – your faith (Greek Catholics / Orthodox – Catholics); our language – your language (Lemkivshchyna / Polish of Lemkos – Polish of Poles); our land – your land (Lemkivshchyna – foreign land); our history – your history (Thalerhof / the Operation “Vistula”). And again, it turned out that it is not fully possible to recognize the core of Lemkos without the historical context and background.

In all those stories, apart from expected and obvious topics (e.g. The Second World War, expatriation, foreign land) related to personal memory and experience, there are also recollections of the past events that the tellers themselves cannot have witnessed (a period of the Austria-Hungary Monar-chy or the First World War). This means that they retell their family tradition and memory. The history became a sort of a web that entwined Lemkos both as a society and as individuals. A web, from which they were not (or they still are not?) able to release themselves.

The trajectory of an exile defined the shape of their linguistic and cul-tural identity in a special way. It changed the status of their ethnolect: from a level of a dialect to a level of a language in a borderland and – at least for a part of the community – a literary microlanguage. That trajectory of exile deprived them of their homeland irretrievably, suddenly and on a massive scale, breaking the continuity of settlement in the Beskidy mountains, a re-gion between the Poprad and Osława rivers. Even a possibility of return to their homeland cannot change this verdict of History. The ethnic Lemkivs-hchyna turned into a historical category. Lemkos can only miss it.

Their feeling of religious independence, however, became even stronger in a foreign land. As it can be seen in their stories, in the first period after displacement, their Eastern Christianity affiliation became a certain me-tonymy for their identity observed in out-group relationships: lexemes Greek Catholic / Orthodox have become synonyms for the ethnonym “Lemko”. Also, the social context of Lemkos’ activity changed dramatically: from an origi-nally dense rural society they became a diasporic one and the scope of their neighbourhood relationships with Others narrowed down to Poles.

The 1947 expulsion is a crucial point in a structure of Lemkos’ stories. It divides all individual stories (as well as the overall history of Lemkos) into two periods (before 1947 and after that year). It introduces the trajectory of an exile (following the terminology by Fritz Schütze). From that moment Lemkos have no inf luence on their fortune; it was decided for them that they had to leave their homeland and where their new home would be. And this event provoked the sense of insecurity, disorganization and suffering. It is only from that moment on that the ref lections upon their own identity appear and these are the ref lections forced by historical events. All previ-


ously presented strategies for Lemkos’ self-presentation cannot be easily connected with narrative topics. They are present (with different intensity) in each sequence of the story.

The Lemko is still related to Lemkivshchyna and at the same time they create their cultural and language identity in the foreign land from scratch. Thus, the expulsion was a crucial revision of the collection of elements con-stituting the Lemkos’ identity universe. It led to a dualism, which can be ob-served in many different spheres: bi-religiosity (apart from Greek Catholic or Orthodox affiliation there appeared a new Catholic affiliation), bilingualism (apart from using their own ethnolect it is increasingly more common for Lemkos to use Polish), so called bi-territoriality (apart from a private sym-bolic homeland the Lemko have a real small homeland in the foreign land).

There is no final answer to the question about cultural and language identity. Each new thread may be complement the already created picture. Identity, each identity, is a process, so its shape changes with time and in context. The analysis presented here has been carried out from the perspec-tive of the Lemkos’ historic and social entanglements. It determined the method of data gathering, analysis and interpretation. The article belongs to the broad scope of research about the Lemkos. It is proposes a new way of defining linguistic and cultural identity.


Weronika DrosdowskaAdam Mickiewicz University, Poznań

Multimodal construction of fear in a TV series – American Horror Story: Cult – Verbal and visual mode

Conceptual Metaphor Theory assumes that metaphors go beyond language and shape our conceptual systems. Conceptual metaphors have been researched multimodally and found in the verbal, visual, gestural, and other modes. Lately, metaphors have been studied in multimodal sources of data, e.g. movies. This article examines how the emotion of fear is constructed in multimodal discourse looking at American Horror Story: Cult – a television series representing the genre of horror. The point of departure is Zoltán Kövecses’ (1990, 2004) description of linguistic metaphors an metonymies of emotions. This paper aims to answer the question whether his findings are applicable to both the verbal and visual modes. Specifically, I looked at which fear metaphors and metonymies are present in both modes, and which are mode-specific. For the analysis of linguistic data the Meta-phor Identification Procedure designed by the Pragglejaz Group (2007) has been employed, as well as keywords derived from Kövecses’ list of metaphors. To analyze the moving image, the analytic tools developed for cinematic language by Edgar-Hunt, Marland and Rawle (2010) were used. The study has shown that Kövecses’ core list of fear metaphors is applicable both to the verbal and the visual mode. New metaphors were also identified. Additionally, it was shown that the visual mode allows for more specificity, while the verbal mode allows for unconventional conceptualization in relation to particular discourse, here, political discourse.


1. IntroductionThere are many ways of expressing meaning which tend to appear not sep-arately, but in combination (Bezemer, Jewitt & O’Halloran 2016: 11-12). They

ABST

RACT

Keywordsmetaphor, me-tonymy, fear, audiovisual mode, multimodality


61Multimodal construction of fear...

are called modes and are defined as culturally and socially shaped resources for making meaning (Kress 2010: 79). A mode could also be seen as a system of signs perceived by senses and thus interpreted. However, because of the complex nature of modes, a precise definition or a list of modes cannot be provided. Nevertheless, pictorial signs, written signs, spoken signs, gestures, sounds, music, smells, tastes, touch have been enumerated by Forceville (2009: 22-23) as some of the existing modes.

Conceptual Metaphor Theory (CMT) brought to light a new way of think-ing about metaphor and human conceptual system. It is based on the as-sumption that metaphors, apart from being present in language, are also a part of human conceptual system which shapes our experience of the world (Lakoff & Johnson 2003: 3, 4). Initially, conceptual metaphors were found in language, but current approaches acknowledge that metaphors can also be expressed visually and through other modes. Each mode may create specific metaphorical expressions (Kress 2010: 154-155), as has been shown in stud-ies of metaphors in the verbal (Kövecses 1990, Yusuf 2016), visual (Forceville 2005, Shinohara & Matsunaka 2009) and gestural (Cienki & Müller 2008, Jelec 2014) modes.

The source of data for multimodal analysis of Conceptual Metaphor has shifted from still image, e.g. static advertisements, branding, logos, and comics (Forceville 2016a on car advertisements) to moving image, such as animated cartoons (Popa 2015) and feature films (Forceville 2015). Contem-porary movies are a combination of image and sound which means that the study of metaphor needs to take into account both visual and auditory data. Such data has been analyzed in various ways, e.g. Forceville (2015) studies animated films focusing on the moving image and ignoring the verbal layer, while Popa (2015) analyzed animated political cartoons both in the verbal and visual mode, however, she treated the visual as still images. The most recent type of data for such investigations come from television advertisements (Forceville 2007) and movies (Forceville 2016b). It is vital to note that, while most of these studies focus on multimodal metaphors, therefore metaphors in which the source and target belong to different modes, it is also a common practice to analyze multimodal data with respect to metaphors constructed separately in each mode, e.g. the verbal and the gestural (Cienki & Müller 2008).

The present study follows the trend of studying multimodal sources in-cluding the verbal layer and the moving image by focusing on a so far relatively unexplored type of data – a television series. American Horror Story: Cult, apart from being an example of multimodal discourse, is the season of the series that portrays different dimensions of fear, namely fear typical for the horror genre, fear related to phobias, and fear induced in political contexts in order to control people. The analysis concerns the verbal and the visual mode on their own in order to investigate the meaning making potential of each of the two


modes, separately. The point of departure for the analysis are Kövecses’ (1990, 2004) metaphors and metonymies of fear. Since Kövecses limited his insight to the verbal mode, I search for verbal and visual manifestations of metaphors and metonymies of fear, asking whether his theory is applicable to both modes. In particular, I look at which fear metaphors and metonymies are present both in the verbal and the visual mode, and which are mode-specific. Additionally, I check if they appear in both modes in the same scene or in different scenes.

The next section of the article contains lists of metaphors and metonym-ies of fear based on the work of Kövecses (1990, 2004). Section 3 discusses the choice of data and provides the methods of analysis. For analysing language the Metaphor Identification Procedure is used along with keywords derived from metaphorical expressions provided in Section 2. Visual metaphors are searched for on the basis of the description of cinematographic tools and their potential for meaning making (Edgar-Hunt, Marland & Rawle 2010). Section 4 contains the analysis of the material and presents the metaphors that were identified in the verbal and the visual mode. Section 5 includes a discussion of the findings and the conclusion of the paper.

2. Verbal and visual metaphors and metonymies of fearFear as an abstract emotion concept was described by Zoltán Kövecses in Emotion Concepts (1990) where he analyzed studies of emotion and the language of emotion. The following list includes the metaphors of fear as proposed by Kövecses (1990: 74-78) along with selected example sentences and short ex-planations modeled on the ones provided by Kövecses:

(1) fearisafluidinacontainerThe sight filled her with fear.

This metaphor assumes that fear is a liquid existing inside a container; the container is the body of the person who experiences the emotion.

(2) fearisaviciousenemy(humanoranimal)The fear that things wouldn’t work out continued to prey on her mind.

In this metaphor fear is a creature that threatens and even harms the experiencer.

(3) fearisatormentorThey were tortured by the fear of what was going to happen to their son.

Here, fear is a person causing torture, pain.


(4) fearisanillnessI have recovered from the shock slowly.

Such metaphors conceptualize fear as causing or being an illness itself.

(5) fearisasupernaturalbeing(ghost,etc.)She was haunted by the fear of death.

In this metaphor fear is a supernatural entity that causes mental struggle.

(6) fearisanopponentHe was wrestling with his fear.

This metaphor treats fear as an opponent in a fight for taking back the con-trol over one’s emotions.

(7) fear(danger)isaburdenFear weighed heavily on them as they heard the bombers overhead.

Fear in this metaphor is an unpleasant experience, while getting rid of fear is pleasant. It could be brought down to simpler terms – the fear might be a heavy object; disposing of it constitutes relief.

(8) fearisanaturalforce(wind,storm,flood,etc.)Fear swept over him.

When comparing fear to a natural force, we put the experiencer in a position undergoing the effects of an uncontrollable force.

(9) fearisasuperiorFear dominated his actions.

This metaphor puts the emotion and the experiencer in a relation of power in which the emotion makes one perform or prevents them from performing certain actions.

In a later work Kövecses (2004: 23) revised the list by changing some la-bels (fear is a hidden enemy, fear is an opponent in a struggle and fear is a social superior) and adding two more metaphors:

(10) fearisinsanityJack was insane with fear.


To my mind this metaphor could be treated as a subordinate to the fear is an illness metaphor because insanity is roughly a synonym of mental illness.

(11) thesubjectoffearisadividedself I was beside myself with fear.

Here, the self that conventionally is located inside the body moves out of it when one loses control over their emotions.

Kövecses (2004: 23-24, 191) also discusses conceptual metonymy. Met-onymies related to emotions are straightforwardly based on physiological aspects of reacting to a given emotion. While they are necessary for under-standing the complex concepts of emotions, they are not sufficient. Met-aphors, on the other hand, are more significant as they complement the understanding of fear by conceptualizing the causes and nature of fear, char-acteristics of one’s relationship with fear, and feelings caused by the cessation of fear (Kövecses 1990: 76, 79). While conceptual metaphors are the main point of interest of the present study, considering some metonymies has proven to be helpful in the analysis of data. The following are fear metonymies as proposed by Kövecses (1990: 77-78): drop in body temperature for fear, physical agitation for fear, increase in heartbeat for fear, blood leaving face for fear, skin shrinking for fear, hair straightening out for fear, inability to move for fear, inability to breathe for fear, inability to speak for fear, (involuntary) release of bowels or bladder for fear, sweating for fear, nervousness in the stomach for fear, dryness in the mouth for fear, screaming for fear, ways of looking for fear, startle for fear, flight for fear.

3. The study: Constructing fear in American Horror Story: Cult

3.1 Data

The study analyses American Horror Story: Cult. The series represents the genre of horror, which tries to scare both the characters and the audience. This is why it was selected as a source of data on metaphors and metonymies of fear. The tropes and themes used in the series are well-known, which suggests that the metaphors are culturally entrenched.

Season 7 of the series was selected for the analysis because fear plays a special role here: apart from the dimension of fear which can be generally expected from the genre of horror, it presents fear in terms of phobias and politics. 11 episodes (around 45 minutes per episode) tell a story of Alyson Mayfair-Richards who has to face her fears and phobias; and Kai Anderson


who starts a cult in order to gain political power, which shows that fear can be used to control people (Ferber 2017). This is why this season was expected to present a high number of various conceptualizations of fear.

3.2 Methods

3.2.1 MIP

In order to find and analyze metaphorical linguistic expressions that are employed to express the conceptualization of fear, I adopted the MIP as in-troduced by the Pragglejaz Group (2007). MIP or “metaphor identification procedure” is a tool that allows for identification of metaphorically used words with respect to the given context in different types of spoken and written discourse. It offers more reliability than pure intuition (Praggle-jaz Group 2007: 1-3). The procedure of identifying metaphors includes the following steps: going through a given text for general understanding; dis-tinguishing particular lexical units in the text and resolving their meaning in the context; determining whether given lexical units could have a more basic meaning in a different context; and if it does so, marking the word as used metaphorically (Pragglejaz Group 2007: 3). I used MIP to analyze the utterances that concern fear.

3.2.2 Keywords

In order to find linguistic manifestations of fear metaphors described in Section 3, I looked for specific keywords (Table 1) relating to those metaphors. I elicited the keywords from the example sentences provided by Kövecses (1990, 2004) for each metaphor, in which the words pointing to the given metaphor were written in italics. I allowed for variation of their grammatical form when locating them in the data.

Metaphor Keywords

FEAR IS A FLUID IN A CONTAINER rise in, fill, contain, full ofFEAR IS A VICIOUS (HIDDEN) ENEMY

(HUMAN OR ANIMAL)lurk, creep up, choke, hound, prey on, harness

FEAR IS A TORMENTOR torture, tormentFEAR IS AN ILLNESS/DISEASE sick with, recover from, get over, plague

FEAR IS A SUPERNATURAL BEING (GHOST, ETC.)

haunt, spooky, dark, ghastly

FEAR IS AN OPPONENT (IN A STRUGGLE) wrestle with, overcome, take hold of, grip, control, besiege, seize, suppress, struggle with, fight,

win out, attack, in the clutch of, incapacitate, overtake

FEAR (DANGER) IS A BURDEN relieved, weigh (heavily) on, (sigh of) relief, alle-viate, burden

Table 1. Key-words cuing the use of metaphor


FEAR IS A NATURAL FORCE (WIND, STORM, FLOOD, ETC.)

sweep over, engulf, surge, f lood, overwhelm, come over, carry away

FEAR IS A (SOCIAL) SUPERIOR prevent from, drive, dictate, rule, dominate, reign

FEAR IS INSANITY insane (with)

THE SUBJECT OF FEAR IS A DIVIDED SELF beside (oneself)

Some metaphors were expressed without the use of the suggested key-words, which required the use of MIP in other stretches of the dialogue list in order to identify the underlying conceptual metaphors. What is more, I did not exclude the possibility of encountering linguistic expressions that could rely on different metaphors than those suggested by Kövecses.

3.2.3 Filmic Techniques

When analysing the moving image I was guided by Edgar-Hunt, Marland and Rawle’s (2010) film theory. Table 2 lists the filmic techniques derived from the literature and possible metaphors that could be expressed by them.

Filmic techniques Metaphor

Low/high angle shot FEAR IS A (SOCIAL) SUPERIOR

A character appearing from outside the frame/a character looking at another char-

acter located outside the frame

FEAR IS HIDDEN ENEMY

Long shots (LS) / extreme long shots (XLS) (depicting landscape)

FEAR IS A NATURAL FORCE (WIND, STORM, FLOOD, ETC.)

Dolly/hand-held shot (depicting a chase/escape)

FEAR IS A TORMENTOR

Zoom (often used with villains in a fight) FEAR IS AN OPPONENT (IN A STRUGGLE)

Low-key lighting (or darkness, night) FEAR IS A HIDDEN ENEMY

Here, I also did not exclude the possibility of finding different metaphors of fear as portrayed by the listed and other filmic techniques, acting, and other elements constituting the visual layer of the data.

4. Results: metaphors of fear in American Horror Story: Cult

4.1 Verbal metaphors

The first example of a metaphorical linguistic expression based on an under-lying conceptual metaphor comes from the first episode of the series. In the

Table 2. Filmic techniques and possible metaphors


scene of interest, Kai Anderson interviews his sister Winter, which is a usual practice before accepting one to join his cult. One of the routine questions he asks concerns fear. In this scene Kai phrases it as follows: Now, tell me, woman what fills your heart with dread? As can be easily noticed, the phrase contains one of the keywords listed in Table 1, namely, the word fills which points to the fear is a fluid in a container metaphor. The lexical unit fills in this context has the meaning of making a person feel an emotion, but it also has a more general meaning that is possible in a different context – to fill is to occupy a physical container by a physical substance. This confirms that the word is used metaphorically and the sentence expresses the metaphor fear is a fluid in a container. What is more, there are other lexical units that are used metaphorically creating the mappings between the domain of fear and the domain of f luids in containers. The noun dread whose basic meaning is ‘intense fear’ becomes the liquid, while heart which is a label for ‘a human body organ’ becomes the container.

Kai Anderson gives many political speeches in which he talks about fear. In the first episode he says that fear is what humans love the most and further characterizes it as “The fear that over time we have honed and polished and built up, brick by brick, until it stands before us every day as tall as the Trump Tower”. In this sentence, Kai clearly speaks about fear in a metaphorical way. While he does not use any of the keywords from the reference lists, he clearly uses two metaphorical expressions. The words hone and polish create map-pings between the domain of fear and something valuable, a treasure. There-fore, the metaphor here could be fear is a valuable commodity or fear is treasure. The second metaphor is expressed with the use of the words: build, brick, stands, tall and tower which point to the domain of building. Therefore, the probable metaphor here is fear is a building. In the same speech, Kai also says the following sentences: “Fear is currency. It has value.” The first one is actually formulated in a way we formulate conceptual metaphors, thus simply providing us with a new metaphorical mapping which has not been identified before, fear is currency. The second sentence again describes fear, in more general terms, as a valuable commodity.

In episode four, during a conversation with one of his followers, Beverly Hope, the leader of the cult presents a different approach by saying that:

(…) the fear in a small town in Michigan can infect the country, the world in a few days. Now, fear, fear isn’t like a virus. When fear finds more hosts, it gets stronger, scarier. The tiny fear in one woman turns into a beast that swallows the world by the time it spreads across the country. Great men and women have been weaponizing fear forever (…).

While he does not use any of the aforementioned keywords, the lexical units: infect, virus, hosts, spreads suggest the presence of the metaphor fear is an


illness, and: beast, swallows are motivated by fear is a vicious enemy (an-imal) or fear is a tormentor. What is more, the last sentence hints at a metaphor which was not identified in the sources discussed in Section 2 – fear is a weapon.

Another character who often talks about fear is Ally Mayfair-Richards. In episode nine she uses the same metaphor as Kai did – fear is a fluid in a container. This happens when she says: “But then, I realized that those fears were filling a hole I had inside me, and you can’t take those fears away without filling that hole with something else”. The keyword here is again fill. While fear is the f luid, the body is the container. The hole in the body which is filled with fear is probably some emotional need of the character.

In episode one, in a situation which is to be described in detail in the analysis of the visual mode, when Ally is really scared, her wife asks her:

“Are you insane?” After some time Ally admits – “I think I’m going insane. I think I’m losing my mind”. The keyword insane as well as the expression to lose one’s mind point to the metaphor fear is insanity or a more general one, namely, fear is an illness, as insanity is a synonym for mental illness which is a type of medical condition.

Earlier in the same episode, Ally talks to Ivy in their restaurant about the fact that her fears have been distracting her from work. Ally replies to the accusations made by her wife in the words: “I am so sorry, and I promise you I’m going to get this under control”. The lexical unit control is one of the keywords for the metaphor fear is an opponent. The character intends to gain control over her opponent – fear. In the sentence the word fear is implied by the pronoun this. In episodes five and six, when her therapist talks to her about her fear, he says: “That is the first step to conquering any fear, because it’s up to you to pull yourself out” and “The fear, Allyson, don’t let it win.” He also uses the fear is an opponent metaphor which is hinted at by the key-word win and the word conquer which is synonymous to it. During a news broadcast, in episode five, Beverly also employs the fear is an opponent metaphor which she expresses through the utterance: “(…) reporting to you from a neighborhood gripped in terror”. Terror is a synonym of the word fear, while gripped is one of the keywords for the metaphor (grip).

The analysis of the verbal mode given above shows further support to most of the source domains identified by Kövecses (1990, 2004). Other source domains which were indentified are: building, valuable commodity, or, more specifically, treasure / currency; and weapon. These metaphors seem to be less conventional and specific to the data which presents fear as a means for controlling people. Let’s now move on to the analysis of the visual mode.


4.2 Visual metaphors

The first scene with significant presence of metaphors in the visual mode appears in the first episode. The scene involves a couple having a picnic, and Twisty the clown, a character that has emerged in one of the previous sea-sons of the series. The scene is a case of intertextuality, as it is a depiction of a comic book read by Oz, one of the characters from the main plot. During the picnic the woman says that she feels as if someone was watching them. Her partner jokes that it might be Twisty the clown, which scares her. The joke, however, turns out to be true, and we see Twisty for the first time. The frame is occupied only by a part of his body: his stomach, the lower parts of his hands and the upper parts of his legs. Later, when the clown is approach-ing the couple, he is followed by the camera (a dolly shot) which shows the back of his head. The fact that we do not see the face of the approaching intruder, and that the camera follows his path, supports the presence of the metaphor fear is a hidden enemY. It is constructed on the principle of a basic metonymy: the part for the whole (Lakoff and Johnson 2003: 36) as parts of the clown’s body represent the entire clown. Simultaneously, this metonymy strengthens the hidden enemy metaphor, as in this shot we do not see his face but other parts of the body partially covered by the under-growth. The dolly shot, as mentioned before, can be related to the fear is a tormentor metaphor. When Twisty reaches the couple, the camera shows them from a high angle laying on the ground, and then it quickly shifts to a low angle shot to show the clown. Such use of camera angle introduces another metaphor: fear is a social superior. The woman starts breathing heavily and screaming out at the sight of the clown. These symptoms include the metonymies: inability to breathe for fear and screaming for fear. Next, the man wants to defend himself and shoots the attacker multiple times. The bullets hit the clown but he does not suffer any damage. Such a reaction is not typical for a human being, therefore the metaphor here is fear is a supernatural being. In return, Twisty heavily abuses the man, which is a clear depiction of the fear is a tormentor metaphor. The entire scene of the fight also brings to mind the opponent and vicious enemy source domains. In the meantime, the woman begins to escape, while the camera follows her in a hand-held shaky shot. Soon the clown starts to follow her. This might depict the metaphor fear is a vicious enemy or a specification of it – fear is a predator. This scene and the metaphor present in it are built upon the metonymy flight for fear. Later, the woman hides in the back of an old car, separated from the rest of the car with bars. Taking into consideration the fact that she cannot escape, because following the path which she used to get into the car would be too dangerous, and that she takes up almost the whole of the frame and looks as if trapped in a cage, it can be


identified as a representation of the metaphor: fear is a trap which was not listed by Kövecses. The lighting in this part changes drastically, from a well-lit sunny outdoor surroundings to a dark, gloomy space with low-key lighting. This kind of lighting was predicted to signal the hidden enemy metaphor (Table 2), however this scene brings to mind one more metaphor which is directly related to the change in lighting: fear is darkness. This metaphor was not proposed by Kövecses. Later, the woman looks in the direction of the space outside of the diegetic world and the clown jumps into the frame from this space, which together with the shadows formed by the low-key lighting again reinforces the metaphor fear is a hidden enemy. The fact that Twisty vigorously attacks her can be interpreted as another instance of the fear is a tormentor metaphor. In the described scenes Twisty as a whole stands for the source of fear but different aspects of the character create different metaphors. This kind of conceptualization seems to be based on the the part for the whole metonymy.

The next scene occurs later in the first episode. It takes place after Ally decides to forget about her fears and worries connected with the political situation in her country and plans a relaxing dinner with her wife at their restaurant. At the beginning, Ivy brings her a plate and when she opens it up, Ally’s phobias come back – she sees blood spilling out of small holes in the food and then she sees a clown. At the beginning of the scene, we see Ally’s head from behind and the camera slowly zooms in on her, which gives an impression of seeing through the eyes of an approaching entity. The fact that the entity is unknown creates a feeling of threat. Later, it turns out that it was Ivy bringing her the meal. This situation involves two metaphors. The first is fear is a hidden/vicious enemy because the camera movement creates an impression of a hostile person unseen to the viewer approaching the character. This can be further supported by the fact that this part of the scene could be described in the words (with a slight modification as to the gender of the experiencer): Fear slowly crept up on [her] which is an example that Kövecses (1990: 75, 2004: 23) provides as a linguistic representation of the fear is a hidden/vicious enemy metaphor. In the metaphor, Ally should be mapped as the victim of the approaching enemy, while the mapping of the hidden enemy goes first to the unknown entity and after the person is revealed, it can be ascribed to Ivy. The fact that the attacker turns out to be the person whom we would not expect, one that is considered to maintain a friendly relation with the object of the “attack”, brings to mind a metaphor that has not been mentioned in Kövecses’ repertoire – fear is an unexpect-ed/a disguised person. These metaphors are specifications of the fear is a hidden enemy metaphor. Later in that scene Ally takes the cover off her plate. What, as we later learn, is a regular dish, appears to her as having small holes with blood spilling out of them. Next, there is a quick zoom in


the direction of Ally’s face, as if she was being attacked by an unseen force. In this frame, Ally starts breathing heavily (inability to breathe for fear metonymy) which strengthens the depiction of her fear. Further, Ally also notices a clown. The fact that Ally sees the holes in her food is caused by her trypophobia, the fear of small holes, one of the phobias she developed be-cause of her fear connected to the changes in politics. She also has another phobia – a fear of clowns.

The subsequent quick approach of the camera could therefore be un-derstood as an “attack” of fear which is motivated by the metaphor fear is a vicious enemy which could be specified as fear is an attacker. Even though the linguistic examples for fear is a vicious enemy indicate the idea of approaching and also acting violently towards the patient, the fear is an attacker metaphor would point more precisely to the abrupt attack expressed by the camera movement in this scene. Finally, the fact that the phobias caused by fear come back points to the metaphor fear is insanity or fear is an illness, as suggested before. The metaphor fear is insanity as expressed through the verbal mode in connection to this scene was de-scribed in the previous section.

The last scene that I am going to analyze comes from episode six. It takes place in the house of Sally Kef ler, a new city council candidate who stood up to Kai. Ally comes to her to seek help. While the women are talk-ing, they suddenly get scared because a window is smashed by an anony-mous character from outside of the house. This action involves the fear is a hidden/vicious enemy metaphor. People dressed as clowns, with masks making their faces unrecognizable, enter the frame. The interpretation of this scene is based on the aforementioned metaphor fear is a disguised person which is a specification of fear is a hidden enemy. At the sight of the intruders, Ally immediately starts to escape (flight for fear metonymy). She runs through a corridor which, as opposed to the room, has no lights, and the camera follows her in a dolly shot with much shakiness. This scene depicts f light from the chasing fear in dark surroundings, which points to the metaphors: fear is a predator, a specification of fear is a vicious enemy, and, again, fear is darkness. At the end of the corridor, Ally enters a door and falls to the ground as if weighted down by the fear, which illustrates the metaphor fear is a burden.

Meanwhile, in the room, before Sally gets shot, some interesting use of camera can be observed. Kai leans down over Sally who lies on the carpet. While his face is shown straight-on from an extremely low angle as if seen through Sally’s eyes, Sally’s face is shown in the opposite manner, which establishes a power relation with the use of the metaphor fear is a social superior. Sally acknowledges her inferior position in this situation and tries to alleviate Kai’s superiority by spitting in his face. Later, after writing Sal-


ly’s suicide note, Kai stands up and approaches the woman. At one point, a part of his body covers almost the entire frame, disenabling the viewers from seeing any of the other characters. This is an instance of the fear is a hidden enemy metaphor. When Sally is shot, for a brief moment we go back to Ally who screams out at the sound of the firing gun (screaming for fear metonymy). Then, one of the clowns, who we later learn is Ivy, heads to the corridor, probably towards Ally’s hideaway. The camera follows only her feet, which makes it seem as if she was a sneaking animal – fear is a vicious enemy (animal). The colors prevailing in the corridor and the bathroom, where Ally hides, are cold and bluish, and the material covering the walls and the f loor of the bathroom is covered with cold-looking tiles. This points to another metaphor not identified before: fear is coldness.

The analysis of the visual mode also provides support for most of the source domains from Kövecses (1990, 2004). However, apart from those listed by Kövecses’, the following source domains were found: predator, trap, darkness, unexpected/disguised person, attacker, coldness.

5. Discussion and conclusionTable 3 below presents a summary of the results of the analysis of the verbal and visual modes.

Source domains and vehicles

Verbal expressions Frequency in the verbal mode

Frequency in the visual mode

VICIOUS (HIDDEN) ENEMY

beast, swallows, con-trol, win, conquer, grip

5 9

- OPPONENT (IN A STRUGGLE)

control, win, conquer, grip

4 1

- ANIMAL beast, swallows 1 1

- TORMENTOR beast, swallows 1 3

- PREDATOR* - - 2

- UNEXPECTED/ DISGUISED

PERSON*

- - 2

- ATTACKER* - - 1

ILLNESS infest, virus, hosts, spreads, insane, lose

one’s mind

3 1

- INSANITY insane, lose one’s mind 2 1

VALUABLE COMMODITY*

hone, polish, currency, value

3 -

- TREASURE* hone, polish 1 -

Table 3. Meta-phors and met-onymies of fear

in American Horror Story:

Cult


- CURRENCY* currency 1 -

FLUID IN A CONTAINER

fill 2 -

BUILDING* built up, brick, stands, tall, tower

1 -

WEAPON* weaponize 1 -

(SOCIAL) SUPERIOR - - 2

DARKNESS* - - 2

SUPERNATURAL BEING

- - 1

TRAP* - - 1

COLDNESS* - - 1

BURDEN - - 1

INABILITY TO BREATHE

- - 2

SCREAMING - - 2

FLIGHT - - 1

The source domains marked with an asterisk were not mentioned in the list in Section 2.

The findings of this study show that in the verbal mode of American Horror Story: Cult we can find most of the metaphors suggested by Kövecses, such as: fear is a fluid in a container which appeared twice; fear is an illness which appeared once or three times if we equal insanity with illness, oth-erwise fear is insanity having two appearances; fear is a vicious enemy (animal); and fear is a tormentor. The most common one was fear is an opponent which I identified four times. The prevalent use of this meta-phor may imply that fear is something one can conquer and control. This is plausible because this metaphor, in most cases, concerned Ally – the main character who attended psychotherapy to fight with her fears. It may also be counted as an instance of a more general metaphor: fear is a vicious enemy, which gives this metaphor the frequency of five. The second most frequent metaphor, fear is insanity, was also connected with her.

In the verbal mode I also found five metaphors which were not ac-counted for by Kövecses: fear is a valuable commodity, fear is trea-sure, fear is currency, fear is a building and fear is a weapon. The currency and treasure metaphors can be generalized to the superordinate metaphor fear is a valuable commodity, however they both have some distinct qualities. The fact that Kai used these unconventional metaphors in his political speeches emerges from his unusual approach to the concept of fear. While most metaphors treat fear as something negative, the poli-tician and cult leader ascribed positive value to it, because he used fear for his personal profit (gaining power). On the other hand, the metaphors con-


nected with Ally as the victim of fear were more conventional, because in her case, fear was construed as a negative emotion. The fear is a valuable commodity metaphor was mentioned by Koivisto-Alanko and Tissari (2006: 210), however, it concerned fear of god which probably does not encompass the distinct aspects of currency and treasure. A similar metaphor to fear is a building was discussed by Ansah (2014). The keyword that she provided was well-founded and she labeled the metaphor as fear is a construction.

The metaphor that prevailed in the visual mode was fear is a vicious/hidden enemy (9 times). It was mostly constructed by the use of camera movement and framing. I noticed many instances specifying this metaphor, i.e. fear is a tormentor (3) and fear is an opponent (in a struggle) (1) which were provided as separate metaphors in Kövecses’ works, fear is an unexpected/disguised person (2), fear is a predator (2), fear is an attacker (1), and fear is an animal (1). The fact that the same metaphor which could be expressed verbally, when shown in the visual mode seems to be more specific, is an argument in favor of the potential of the visual mode to make metaphorical realizations more concrete and specific than is the case with the verbal mode.

Other metaphors in the visual mode were: fear is a social superior (2), fear is darkness (2), fear is a supernatural being (1), fear is insanity/illness (1), fear is coldness (1), fear is a burden (1) and fear is a trap (1). The fear is a trap metaphor could be related to a more general metaphor: fear is a container which was also mentioned by Ansah (2014).

Two metaphors from Kövecses’ list which were not present in the ana-lyzed data are: fear is a natural force and the subject of fear is a di-vided self. The first metaphor did not occur, probably, because in a television series it is not common to show wide-shot depictions of nature. This is thus a genre-specific result. The second metaphor is generally not very frequent, as even Kövecses provides only one linguistic example of it.

The metaphor which occurred most often in both modes was the fear is a hidden/vicious enemy metaphor along with its specifications. According to Stefanowitch’s (2008: 78) comment to Kövecses’ list of metaphors of fear:

“it is not clear why fear is a vicious enemy and fear is a tormentor are posited as separate mappings rather than being subsumed under something like fear is an enemy, together with fear is an opponent in a struggle”. I agree with this stance, because, especially in the case of the visual mode, I experienced trouble determining which one of those metaphors was the basis for a given representation, and often, more than one metaphor seemed appli-cable. It is worth noting, however, that each of these metaphors has some dis-tinct features. In fact, in the visual mode I identified further specifications: fear is an unexpected/a disguised person, fear is a predator, fear is an attacker. Another metaphor which could be found in both modes was


fear is insanity, probably because it was related to one of the main themes of the analysed material. The scenes in which Ally experienced her phobias were portrayed visually, but oftentimes the illness was also discussed.

Some metaphors were found only in one of the two examined modes. The metaphors which were exclusive to the verbal mode were: fear is a building (construction), fear is a valuable commodity (currency/treasure) and fear is a weapon. All of them came from political speeches which are a verbal form of expression. The metaphors, therefore, were trig-gered by the particular discourse. Another verbal-only metaphor was fear is a fluid in a container, a highly abstract metaphor which often includes humans’ internal organs, which possibly makes it hard to portray in the live action television series which is supposed to present situations resembling real life. The source domains which were found only in the visual mode are: social superior, trap (container), burden, darkness, and coldness. Most of them were conveyed by filmic tools: framing, angle, lighting, and colors. They are all related to the conventional representation of fear and con-cern scenes which are typical for the genre of horror, unlike the ones which were related to politics. The metaphor fear is dark was also discussed in a paper by Winter (2014) who analyzed metaphors of fear in horror movies. He inferred this metaphor from evil is dark, following the assumption that the domain of evil encompasses the domain of fear (Winter 2014: 152). Omori (2008) mentions the relation between emotions and temperature, and links the emotion of fear with the domain of coldness. The existence of such metaphor was actually confirmed by Kövecses (2008) in a reply to Omori’s paper. The scenes which were presented as containing visual metaphors, also included metonymies (inability to breathe, screaming and flight for fear). It can, therefore, be assumed that especially in the visual mode met-aphor and metonymy interact with each other to provide a more complex understanding of the emotion. What is more, most of the described scenes did not contain metaphors in both modes at the same time. The only one concerned Ivy asking Ally about her insanity when she was experiencing her phobias (fear is insanity). In this scene the same metaphor was expressed in both modes.

To sum up, the findings reveal what metaphors were used to conceptu-alize fear in American Horror Story: Cult. The study has shown that Kövecses’ core list of fear metaphors is also applicable to the visual mode and led to the identification of metaphors which were not present on the list. It was shown that the visual mode allows for more specificity, while unconventional conceptualization is possible in the verbal mode in relation to particular discourse, here, political discourse.

The scope of this paper allowed only for the analysis of metaphor and metonymy in the two major modes: the verbal and the visual, but the TV


series could be further analyzed with regard to the mode of music, and the use of image schemata, which could provide an even fuller picture of how the emotion of fear can be constructed multimodally. Further studies with relation to other emotions are also possible.

Streszczenie

Konstruowanie strachu w serialu telewizyjnym American Horror Story: Cult – modalność werbalna i wizualna

Komunikacja jest zjawiskiem multimodalnym – istnieje wiele systemów semiotycznych pozwalających na tworzenie komunikatów (Bezemer, Jewitt & O’Halloran 2016: 11-12). Systemy te są społecznie i kulturowo ukształto-wane. Często występują jednocześnie w tym samym komunikacie (Kress 2010: 79). Teoria metafor pojęciowych (Conceptual Metaphor Theory, CMT) zakłada, że metafory istnieją nie tylko w języku, ale są częścią ludz-kiego systemu konceptualnego i wpływają na to, jak postrzegamy świat (Lakof f & Johnson 2003). Początkowo metafory pojęciowe znajdowano w języku, ale obecnie wiadomo, że można je również wyrazić za pomocą m.in. obrazu i gestów (np. Forceville 2005, Shinohara & Matsunaka 2009, Cienki & Müller 2008, Jelec 2014).

Artykuł ten zajmuje się metaforami strachu obecnymi w serialu te-lewizyjnym American Horror Story: Kult. Analiza dotyczy przedstawienia emocji – strachu w sposób werbalny oraz wizualny. Wybrany serial jako źródło multimodalne pozwala na analizę mowy, jak i obrazu, a ponadto przedstawia różne wymiary strachu: strach typowy dla gatunku, jakim jest horror, strach związany z fobiami, a także strach w kontekście politycz-nym. Analiza opiera się na metaforach oraz metonimiach strachu przed-stawionych przez Zoltána Kövecsesa (1990, 2004). Kövecses omawia meta-fory występujące jedynie w języku, a te badania sprawdzają, czy obecne są one również w warstwie wizualnej. Analiza poszukuje odpowiedzi na następujące pytania: (1) Czy metafory przedstawione są tylko werbalnie lub tylko wizualnie? (2) Które metafory mogą być wyrażone zarówno w ję-zyku jak i w obrazie? (3) Czy te same metafory występują w języku i obrazie w jednej scenie? (4) Czy w serialu pojawiają się metafory nieopisane wcze-śniej przez Kövecsesa?

W celu analizy dialogów pod kątem obecności metafor strachu użyłam metody identyfikacji metafory (Metaphor Identification Procedure, MIP), opracowanej przez Pragglejaz Group (2007). Aby zidentyfikować potencjalne


metafory, posłużyłam się słowami kluczowymi zaczerpniętymi z przykła-dów, które Kövecses podaje do każdej z omawianych przez niego metafor (Tabela 1). W celu odnalezienia metafor obecnych w obrazie, posłużyłam się teorią filmu Edgara-Hunta, Marlanda & Rawle’a (2010), co pozwoliło mi przewidzieć, w jaki sposób mogłyby zostać przedstawione poszczególne metafory (Tabela 2). Spodziewałam się również odnalezienia metafor innych niż te zaproponowane wcześniej przez innych badaczy.

Metafora Słowa kluczowe

STRACH TO PŁYN W POJEMNIKU rosnąć, wypełniać, opanować, pełen

STRACH TO AGRESYWNY (UKRYTY) WRÓG (CZŁOWIEK LUB ZWIERZĘ)

czaić się, podkradać się, dusić, prześladować, żerować na, ujarzmić

STRACH TO DRĘCZYCIEL torturować, dręczyć

STRACH TO CHOROBA chory, wyzdrowieć, dochodzić do siebie, nie dawać żyć

STRACH TO ISTOTA O CECHACH NAD-PRZYRODZONYCH (DUCH ITP.)

nawiedzać, upiorny, mroczny, koszmarny

STRACH TO PRZECIWNIK (W WALCE) borykać się z, pokonać, panować nad, schwytać, kontrolować, oblegać, zawładnąć, stłumić,

zmagać się z, walczyć, zwyciężyć, atakować, w uścisku, obezwładnić, ogarnąć

STRACH (NIEBEZPIECZEŃSTWO) TO CIĘŻAR

odczuwać ulgę, obciążać, westchnąć z ulgą, ulżyć

STRACH TO SIŁA PRZYRODY (WIATR, BURZA, POWÓDŹ ITP.)

ogarnąć, pochłaniać, fala, zalać, zasypać, porwać

STRACH TO OSOBA WYŻSZA RANGĄ powstrzymać przed, zmusić do, nakazywać, rządzić, dominować, panować

STRACH TO OBŁĘD oszaleć

OSOBA DOŚWIADCZAJĄCA STRACHU TO ROZDWOJONA POSTAĆ

wyjść z siebie

Techniki filmowe Metafora

„Ptasia”/„żabia” perspektywa (kamera skie-rowana w dół/w górę)

STRACH TO OSOBA WYŻSZA RANGĄ

Postać wchodząca w kadr spoza niego/postać spoglądająca na inną, znajdującą się

poza kadrem

STRACH TO UKRYTY WRÓG

Plan totalny STRACH TO SIŁA PRZYRODY (WIATR, BURZA, POWÓDŹ, ITP.)

Filmowanie „z ręki” (obrazujące pogoń/ucieczkę)

STRACH TO DRĘCZYCIEL

Tabela 1. Słowa kluczowe wska-zujące na użycie metafor

Tabela 2. Tech-niki filmowe i prawdopodobne metafory


Zbliżenie (często używane w przypadku walki)

STRACH TO PRZECIWNIK (W WALCE)

Niski klucz oświetleniowy (low-key) (albo ciemność, noc)

STRACH TO UKRYTY WRÓG

Podsumowanie wyników analizy znajduje się w Tabeli 3.

Domeny źródłowe i nośniki metonimii

Wyrażenia Frekwencja – kod werbalny

Frekwencja – kod wizualny

AGRESYWNY (UKRYTY) WRÓG

bestia, połykać, kontrolować, wygrać,

pokonać, schwytać

5 9

- STRACH TO PRZECIWNIK

(W WALCE)

kontrolować, wygrać, pokonać, schwytać

4 1

- ZWIERZĘ bestia, połykać 1 1

- DRĘCZYCIEL bestia, połykać 1 3

- DRAPIEŻNIK* - - 2

- NIESPODZIE-WANA/

ZAMASKOWANA OSOBA*

- - 2

- AGRESOR* - - 1

CHOROBA zarazić, wirus, nosiciel, rozprzestrzeniać się,

oszaleć, postradać zmysły

3 1

- OBŁĘD oszaleć, postradać zmysły

2 1

CENNY TOWAR* szlifować, polerować, waluta, wartość

3 -

- SKARB* szlifować, polerować 1 -

- WALUTA* waluta 1 -

PŁYN W POJEMNIKU

wypełniać 2 -

BUDOWLA* budować, cegła, stać, wysoki, wieża

1 -

BROŃ* używać jako broń 1 -

OSOBA WYŻSZA RANGĄ

- - 2

CIEMNOŚĆ* - - 2

ISTOTA O CE-CHACH NADPRZY-

RODZONYCH

- - 1

PUŁAPKA* - - 1

ZIMNO* - - 1

CIĘŻAR - - 1

Tabela 3. Metafory i me-

tonimie strachu w American

Horror Story: Kult


PROBLEMY Z ODDYCHANIEM

- - 2

KRZYK - - 2

UCIECZKA - - 1

Domeny źródłowe zaznaczone gwiazdką nie zostały wspomniane przez Kövecsesa.

Wyniki badania pokazują, że w dialogach w American Horror Story: Kult występuje większość metafor z listy Kövecsesa: strach to płyn w pojemniku, strach to choroba, strach to obłęd, strach to prze-ciwnik w walce, strach to agresywny wróg. Dwie najczęstsze meta-fory to: strach to przeciwnik w walce i strach to obłęd. Ta pierwsza wskazywałaby, że strach można pokonać i kontrolować. Co ciekawe, obie te metafory dotyczyły jednej z głównych postaci – Alyson Mayfair-Richards, która zmagała się ze strachem i fobiami. W dialogach znalazłam także pięć metafor, które nie zostały opisane przez Kövecsesa: strach to cenny to-war, strach to skarb, strach to waluta, strach to budowla, strach to broń. Wszystkie te metafory zostały użyte przez Kaia Andersona, który wykorzystał je w swoich przemowach politycznych. W odróżnieniu od kon-wencjonalnego podejścia do strachu, czyli traktowania go jako czegoś ne-gatywnego, ten bohater nadawał strachowi cechy pozytywne, ponieważ używał go, aby osiągnąć własne cele. Świadczy to o tym, że w przypadku konkretnego dyskursu, możemy natknąć się na nietypowe metafory.

W przypadku kodu wizualnego zidentyfikowane metafory to: strach to agresywny/ukryty wróg oraz metafory podobne, ale bar-dziej szczegółowe, tj. strach to dręczyciel, strach to przeciwnik w walce, strach to niespodziewana/zamaskowana osoba, strach to drapieżnik, strach to agresor, strach to zwierzę. Wnioskować można zatem, że metafory, które występują także w języku, przedsta-wione w formie wizualnej mogą być bardziej konkretne i szczegółowe. W warstwie wizualnej znalazłam również następujące metafory: strach to osoba wyższa rangą, strach to ciemność, strach to istota o ce-chach nadprzyrodzonych, strach to choroba/obłęd, strach to zimno, strach to ciężar oraz strach to pułapka.

Metafory, które opisane zostały przez Kövecsesa, a nie zostały zi-dentyfikowane w analizowanym materiale to: strach to siła przyrody oraz osoba doświadczająca strachu to rozdwojona postać. Meta-fory, która występowały zarówno w obrazie, jak i w dialogach to: strach to ukryty/agresywny wróg oraz podobne, szczegółowe metafory, a także strach to obłęd, czyli metafora, która powiązana była z jednym


z głównych wątków serialu. Metafory realizowane jedynie w kodzie werbal-nym to wymienione wcześniej metafory związane z dyskursem politycznym. Pojawiły się tam prawdopodobnie dlatego, że wiązały się z przemowami będącymi formą ekspresji werbalnej. Jedynie werbalna postać miała także metafora strach to płyn w pojemniku. Metafory przedstawione tylko w kodzie wizualnym to: strach to osoba wyższa rangą, strach to pułapka, strach to ciężar, strach to ciemność i strach to zimno. Wiązały się one z konwencjonalnym przedstawieniem strachu typowym dla gatunku horroru, a większość z nich oparta była na pracy kamery i na grze świateł i kolorów. Metaforom przedstawionym w warstwie wizualnej towarzyszyły także metonimie: problemy z oddychaniem, krzyk oraz ucieczka za strach, co pokazuje, że szczególnie w przypadku obrazu metafory i meto-nimie wchodzą w interakcję i tworzą bardziej złożone przedstawienie emo-cji. Tylko jedna metafora – strach to obłęd obecna była w jednej scenie zarówno w formie werbalnej, jak i wizualnej.

ReferencesAnsah, Gladys Nyarko 2014: Culture in embodied cognition: Metaphorical/

metonymic conceptualisations of FEAR in Akan and English. Metaphor and Symbol 29(1), 44-58.

Bezemer, Jeff, Jewitt Carey, Kay O’Halloran 2016: Introducing multimodality. London: Routledge.

Cienki, Alan, Cornelia Müller (eds.) 2008: Gesture studies. Vol. 3: Metaphor and gesture. Amsterdam: John Benjamins Publishing.

Edgar-Hunt, Robert, John Marland, Steven Rawle 2010: The language of film. Switzerland: AVA Publishing SA.

Ferber, Taylor 2017: The ‘AHS: Cult’ trailer has a deeper meaning you probably missed (https://www.bustle.com/p/the-ahs-cult-trailer-has-a-deep-er-meaning-you-probably-missed-78269) (date of access: 04 July 2019).

Forceville, Charles 2005: Visual representations of the idealized cognitive model of anger in the Asterix album La Zizanie. Journal of Pragmatics 37(1), 69-88.

Forceville, Charles 2007: Multimodal Metaphor in Ten Dutch TV Commer-cials. Public Journal of Semiotics 1, 19-51.

Forceville, Charles 2009: Non-verbal and multimodal metaphor in a cognitiv-ist framework: Agendas for research. In: Forceville, Charles J.,Eduardo Urios-Aparisi (eds.): Multimodal metaphor. Berlin: Mouton de Gruyter.

Forceville, Charles 2015: Metaphor and symbol: searching for one’s iden-tity is looking for a home in animation film. In: Pinar-Sanz, Maria-Jesus, (ed.) Multimodality and cognitive linguistics. Amsterdam: Benjamins, 27-44.


Forceville, Charles 2016a: Mixing in pictorial and multimodal metaphors? In: Gibbs, Raymond W. Jr, (ed.) Mixing metaphor. Amsterdam: Benja-mins, 223-240.

Forceville, Charles. 2016b: Visual and multimodal metaphor in film: Chart-ing the field. In: Kathrin Fahlenbrach (ed.), Embodied metaphors in film, television and video games: Cognitive approaches. London: Routledge, 17-32.

Jelec, Anna 2014: Are Abstract Concepts Like Dinosaur Feathers? Conceptual Meta-phor Theory and the conceptualisation strategies in gesture of blind and visually impaired children. Poznań: Wydawnictwo Naukowe UAM

Koivisto-Alanko, Päivi, Heli Tissari 2006: Sense and sensibility: Rational thought versus emotion in metaphorical language. In: Stefanowitsch, Anatol, Stefan Th. Gries (eds.) 2006: Corpus-Based approaches to metaphor and metonymy. Berlin: Mouton de Gruyter, 191-213.

Kövecses, Zoltán 1990: Emotion concepts. New York: Springer-Verlag.Kövecses, Zoltán 2004: Metaphor and emotion: Language, culture, and body in

human feeling. Cambridge: Cambridge University Press.Kövecses, Zoltan 2008: On metaphors of emotion: A reply to Ayako Omori.

Metaphor and Symbol 23(3), 200-203.Kress, Gunther R. 2010: Multimodality: A social semiotic approach to contemporary

communication. London: Routledge.Lakoff, George, Mark Johnson. 2003. Metaphors we live by. London: The Uni-

versity of Chicago Press.Omori, Ayoko 2008: Emotion as a huge mass of moving water. Metaphor and

Symbol 23(2), 130-146, DOI: 10.1080/10926480801944277.Page, Ruth (ed.) 2010: New perspectives on narrative and multimodality. New

York: Routledge.Popa, Diana E. 2015: Multimodal Metaphors in political entertainment. In:

Pinar-Sanz, Maria-Jesus, (ed.) 2015: Multimodality and cognitive linguistics. Amsterdam: Benjamins, 79-95.

Pragglejaz Group 2007: MIP: A method for identifying metaphori-cally used words in discourse. Metaphor and Symbol 22(1), 1-39, DOI: 10.1080/10926480709336752 .

Shinohara, Kazuko, Matsunaka, Yoshihiro 2009: Pictorial metaphors of emo-tion in Japanese comics. In: Charles Forceville, Eduardo Urios-Aparisi (eds.) 2009: Multimodal Metaphor. Berlin: Mouton de Gruyter, 265-296.

Stefanowitsch, Anatol 2008: Words and their metaphors: A corpus-based approach. In: Stefanowitsch Anatol, Stefan Th. Gries (eds.) 2008: Cor-pus-Based Approaches to Metaphor and Metonymy. Berlin/NewYork: Mouton de Gruyter, 1-16.


Winter, Bodo 2014: Horror movies and the cognitive ecology of primary met-aphors. Metaphor and Symbol 29(3), 151-170.

Yusuf, Aliyu Yakubu 2016: Conceptual Metaphor in the Language of Football Commentary: A Cognitive Semantic Study. Ganga Journal of Language and Literary Studies 5(4), 2-17.


Agnieszka Mierzwińska-HajnosMaria Curie-Skłodowska University in Lublin

Where Music and Text Meet: A Multimodal Analysis of Vissi d’Arte aria

The aim of this paper is to scrutinize the factors that determine the multimodal character of op-eratic performances while analyzing Vissi d’Arte soprano aria of Giacomo Puccini’s Tosca. By using blending theory (Fauconnier and Turner 2002) as well as Brandt and Brandt’s six-space model of conceptual integration (2005), we will try to show that the integration of auditory and verbal modes of communication that accompany each operatic performance should be approached, first and foremost, as a conceptual operation –“the ref lection of cognitive capacities of human beings” (Zbikowski 2009a: 448). It is also claimed that it is complex and dynamic conceptual operations that contribute to a better reception and understanding of arias in the speaker – hearer interaction.


(…) to analyze an opera-to analyze its constituent parts

and then to see how these fit together,

musically and dramatically- is a multifaceted process

that requires a considerable amount

of gear-changing

(Drabkin 2014: 245, after Rossi & Sindoni 2017: 62)

INTRODUCTION

Being a combination of a multitude of elements such as music, text, gestures, scenery, costumes, or dance, opera becomes one of the most complex and,

ABST

RACT

Keywordsoperatic performance, multimodality, conceptual blending, Vissi d’Arte aria


85Where Music and Text Meet: A Multimodal Analysis of Vissi d’Arte aria

simultaneously, most challenging meaning-making events (cf. Rossi & Sin-doni 2017). The way operatic performances are received and interpreted by the audience can barely be predicted: opera remains variable, dynamic, and is usually determined by many factors, the most important of which are var-ious modes of communication used. No less important is also the way these modes are processed in the interaction between the potential listener and the artist during the performance, as well as how they shape and communi-cate the ultimate message of a given operatic acting (Rossi & Sindoni 2017).

Studied mainly from a musicological perspective, usually limited to mu-sical, staging or theatrical aspects (cf. Rabb 2006), opera is still unappreciated as the empirical material in the linguistic analysis, especially the one carried out from a multimodal as well as cognitive perspective. This ‘lukewarm’ at-tention obtained from linguists seems to stem from two interrelated factors. Firstly, with its complexity and diversity, opera is considered too huge a challenge for a linguistic analysis. Secondly, with a whole gamut of variables, starting from the changes in the mise-en-scène, various soloists performing and interpreting the same aria, and finishing in the corrections made both in the libretto and the score for a particular performance (cf. Harris-Warrick 2016; Strohm 1985), opera becomes ‘an unstable medium’ (Drabkin 2014: 245), which also impedes an in-depth linguistic study.

Approaching opera from a linguistic perspective is as much challeng-ing as justifiable a task. It becomes even more justifiable when we adopt a cognitive stance according to which meaning becomes the central target of the linguistic endeavour (cf. Evans & Green 2006; also Kardela 1992), simul-taneously allowing for all the factors that determine it, i.e. the situational context in which a given meaning is likely to occur, the role of interlocutors responsible for encoding and decoding the necessary information in the act of communication, or various modes that allow us to enhance its potential.

Whether perceived as too complex or too unstable, opera seems to ful-fill all the abovementioned parameters and does remain a meaning-making event. Additionally, with a plethora of communicative modes involved, it also becomes a challenging multimodal event (Rossi & Sindoni 2017), the pivotal objective of which is to strengthen the meaning as conveyed in a given operatic performance or its part and thus boosting its reception among the audience.

The present paper is an attempt of a cognitive-cum-multimodal analysis of Vissi d’Arte soprano aria of Giacomo Puccini’s Tosca. Within the course of the proposed analysis we will try to discuss the importance of two modes of communication involved in the performance of a famous aria, music and text, and portray the aria as the result of blending multimodal inputs, the main objective of which is to contribute to a better reception of this form of art by the audience. Allowing for the fact that each operatic performance should


also be rendered a meaning-making event where we find “the ref lection of cognitive capacities of human beings” (Zbikowski 2009a: 448; also Kühl 2007), the analysis in question will be a cognitive linguistics-oriented study, based on the six-space model as proposed by Brandt and Brandt (2005; also Brandt 2009). While analyzing Vissi d’Arte we will try to show that each op-eratic performance is not only a multimodal event, but it is also the result of complex and dynamic conceptual operations that lead to better reception and understanding of an aria.

1. Operatic performance as a multimodal eventDespite the fact that the very structure of each operatic performance calls for the simultaneous activation and ‘co-deployment’ (cf. Rossi & Sindoni 2017) of various resources, opera has received limited attention in multi-modal studies, resulting so far in only a few credible publications.1 Mean-while, approaching opera as “an eminently multimodal event”, the meaning of which “can only be understood by reintegrating all the systems, resources, and components that work in combination in every operatic performative event” (Rossi & Sindoni 2017: 81) is not a novel observation as it coincides with the 17th century observation put forward by an Italian composer Marco da Gagliano, who first described opera as a form of ‘total art’ (after Rossi & Sindoni 2017: 63).

With all the breadth of resources involved, opera entirely fulfills the three key premises of multimodality. Firstly, its meaning is constructed with the aid of various resources (or modes), each having its “distinct potential-ities and limitations”(Jewitt, Bezemer & O’Halloran 2016: 3). Secondly, the process of meaning making results in the production of multimodal wholes, the decoding of which is inextricably bound up with the third premise, i.e. the exploration of all semiotic resources used in the process of making

“a complete whole” (Jewitt, Bezemer & O’Halloran 2016: 3). Nevertheless, claiming that it is enough to conform to the three premises to provide a good multimodal analysis of an operatic performance would be a too far-fetched generalization, as distinct variables have to be taken into consideration. The complexity of opera, and, what follows, the dilemma with the profile of mul-timodal analysis to be adopted is best portrayed in the following observation:

Each multimodal artifact can be read, understood and received differently, depending on the context and semiotic expertise of the receiver. An opera, for example, can be interpreted very differently by an occasional specta-

1 Much as the overwhelming majority of papers have been dedicated to opera, this medium is primarily explored from a musicological or performance studies perspective. Opera understood as a multimodal meaning making event has gained attention in the realm of linguistic and discourse studies only recently (see, e.g. Rossi & Sindoni 2017; Hutcheon & Hutcheon 2010).


tor or by a hardcore fan (…). The combination of resources and how they are read and interpreted by communities or societies is neither static nor predictable and may vary considerably across time and space according to context. (Rossi & Sindoni 2017: 61)

As stated above, it is virtually impossible to capture the essence of each and every mode and explore its contribution to an overall reception of a given operatic performance. Each attempt of a multimodal analysis should be profiled with reference to what it aims to convey since, as observed by Rossi and Sindoni, “no definitive analytical framework can be provided when any complex multimodal fact is involved” (2017: 65). Nevertheless, it seems that in the case of operatic performances two aspects of multimodality should always be taken seriously into account, viz. (i) multimodal communication and (ii) multimodal interaction (Jewitt, Bezemer & O’Halloran 2016). By multimodal communication we understand the way a given message is communicated by the speaker (i.e. the performer) to the recipient (the audience) with the aid of accompanying resources, as well as how the message obtained by means of applying various modes inf luences the final reception of a given performance. Multimodal interaction, in turn, explores the process of integrating various modes to obtain a meaningful whole. Both these aspects are very much in line with what cognitive linguistics offers to account for a context-depen-dent and dynamic meaning construction in the speaker-hearer interaction, especially while recalling the notion of conceptual blending (cf. Fauconnier & Turner 2002) as proposed in the revised form of the six-space Brandt and Brandt model (2005).

2. Material and Methods UsedThe multimodal character of the proposed analysis inheres in understand-ing communication and representation to be more than about language. As observed by Jewitt (2009: 12, see also Kress 2009):

Multimodality assumes that representation and communication always draw on a multiplicity of modes, all of which contribute to meaning. It fo-cuses on analyzing and describing the full repertoire of meaning-making resources that people use (visual, spoken, gestural, written, three-dimen-sional, and others, depending on the domain of representation) in different contexts, and on developing means that show how these are organized to make meaning.

Still, allowing for the fact that a credible holistic multimodal analysis of a given operatic performance cannot be offered in the form of a single original paper, a cognitive-cum-multimodal analysis proposed herein will limit itself to the study of two fundamental modes being the essence of opera, viz. (i)


music (the score), and (ii) text (the libretto), the material in question being Vissi d’Arte aria of Puccini’s opera Tosca.

The choice of modes for the analysis is not random here, but rather draws on a multiplicity of observations pertaining to an interplay between the two in the process of meaning construction (see, i.a. Zbikowski 2015, Maeder 2011, 2013, Lerdahl 2003). A daring statement made by McClary who notices that “meaning is not inherent in music but neither is it in language” (1991: 21) immediately compels us to approach music and text as interdepen-dent modes which, when combined together, build “bridges (…) between the musical form and representational contributions of the language involved” (Bateman, Wildfeuer & Hiippala 2016: 28). The complementary, or even ‘hy-brid’ (cf. Lerdahl 2003) character of the two modes in question is also stressed in a cognitive-oriented line of argumentation. Music and text are believed to stem from the same cognitive resources (cf. Zbikowski 2009b, also Jackendoff 2008), simultaneously being rendered as belonging to two distinct domains, each having a different function ascribed. Thus, the function of language (whichever form of representation, written or spoken, is recalled here) is mainly limited only to WHAT is communicated2, rather than HOW the mes-sage is communicated, which immediately makes the text ‘monodimensional’ (cf. Maeder 2013). Music, in turn, is usually defined as “the enhancement of affect associated with an activity” (Jackendoff 2008: 197). In this sense, it becomes the factor responsible for evoking emotions that accompany the text.

Taking into account the above-mentioned considerations, arias prove an excellent material for a multimodal-cum-cognitive analysis. It seems that the contribution of music and text, both being discussed in detail in subsequent parts of this paper, is indispensable not only for the interpretation but also for the reception of an aria.

No less important is also the methodology assumed for such an analy-sis. Drawing on the observations made by Zbikowski, according to whom

“certain combinations of words and music create a realm for the imagination that extends far beyond what words or music alone could summon” (2015: 158), the choice of conceptual blending theory (Fauconnier & Turner 2002), applied as the method for the analysis of all forms based on text-music in-terplay (songs, operas, masses, oratoria, etc.), is justified by earlier research (see, i.a. Zbikowski 2008, 2009b, 2015, 2018, also Pérez-Sobrino 2014).

Conceptual blending, considered a basic cognitive operation, adds a lot to a multimodal analysis as it can model input spaces exemplifying in-formation in different modalities. Nevertheless, assuming that conceptual blends are considered to be contextualized ad-hoc structures that might be rendered as intentional, discursive chunks of information which emerge in

2 As pointed out by Zbikowski, language “gives us the means to represent symbolically objects and relations” (2009b: 376).


a specific communicative situation and which are always context-relevant, the question arises whether the basic Fauconnier and Turner four-space net-work model is not too general an approach to capture the specificity of the analyzed aria. The limitations of the Fauconnier &Turner model originate in the lack of clear explanation how and in what circumstances conceptual blends are constructed (see, e.g. Oakley & Coulson 2008, Cienki 2008, Veale & O’Donoughe 2000). The statement that the whole blending process con-sists in making “new meanings out of old” (Nordquist 2019: 1) on the basis of the integration of mental spaces activated “for purposes of local under-standing and action” (Fauconnier & Turner 2002) in the process of think-ing and talking is not satisfactory enough as it does not explicitly state the importance of background knowledge as well as the role of interlocutors in a communicative situation. Therefore, in lieu of the classical Fauconnier & Turner model of conceptual integration (2002), a revised six-space model as proposed by Brandt & Brandt (2005)3 will be offered for the music-and-text analysis of Vissid’Arte aria, since blend analysis should take account of the situational context and the communicative purpose of constructing the blend assuming that a given blended unit is to mean “what it is intended to mean in a particular situation where it is uttered by someone” (Brandt and Brandt 2005: 219).

The Brandt & Brandt model is composed of six mental spaces, i.e. se-miotic space, reference space, presentation space, relevance space, and a two-stage blended space (i.e. consisting of the virtual space and the mean-ing space). All mental spaces are linked to one another via different kinds of relations, which make “a figurative and dynamic semantic network that is designed to derive the (…) meaning of the utterance” (Brandt & Brandt 2005: 220). A schematic representation of the Brandt & Brandt model is presented in the figure below:

Let us now proceed to the analysis of Vissi d’Arte aria to see how text and music can be interpreted by applying the Brandt & Brandt six-space model.3 See also Brandt (2005).

Figure 1. Brandt & Brandt’s six-space model (2005: 235)


3. Vissi d’Arte: The analysis

The analysis offered here is the study of the famous Vissi d’Arte soprano aria as presented at the end of act II of Giacomo Puccini’s Tosca. By applying the Brandt & Brandt six-space model (2005) we will try to examine in what way music and text integrate with each other to boost the content of a message comprised in the aria itself, as well as answer the question to what extent the integration of such two distinct modes may contribute to a better reception of the aria by the audience.

The pivotal theme of the aria is the blackmail the main character, Flo-ria Tosca, falls victim to. Caught between a true love towards her beloved, Mario Cavaradossi and lured by the sinsiter Baron Scarpia, the chief of Ro-man police and persecutor of Cavaradossi, she is forced to make the most difficult decision: in order to save her lover (sentenced by Scarpia to death for treason) she must sleep with the Baron. Vissi d’Arte is essentially Tosca’s cry of anguish as she is faced with the impossible choice: either way, she will have to betray Cavaradossi, in the form of sleeping with another man, or not doing everything she could to save his life.

Our considerations will begin with the reconstruction of the semiotic (base) space, considered by Brandt & Brandt the starting point for structuring the semantic content of the expected message where the interlocutors, i.e. the speaker and the hearer, share in the dynamic act of meaning construction. In the case of the analyzed aria, the semiotic space, or, more precisely, the very act of semiosis, takes place while listening to the operatic performance of Vissi d’Arte. Obvoiusly, a communicative situation as depicted here departs from its canonical form as represented by a “face-to-face two-participant interaction” (Langacker 2009: 153). Instead, we deal with one speaker, i.e. a soprano singer who interacts with the collective hearer, viz. the audience (cf. Di Giovanni 2018). By assuming the role of Floria Tosca, a soprano sing-er-the speaker strives to portray the anguish the main heroine experiences by means of music and text at her disposal, while the audience-the hearer attempts to appropriately decode the message conveyed in the aria, in con-cord with the generally agreed-upon model of successful communication4 and according to culturally shared patterns and schemas for expressing de-spair, activated in the reception of both the verbal and musical layer of Vissi d’Arte (cf. Zbikowski 2015, Machin & Richardson 2012).

It has to be noticed that the act of semiosis, though it depicts the here-and-now moment of Tosca lamenting her fate, does not exist per se. Put dif-

4 As pointed out by Heylen (2009: 2151), “a typical communicative action is normally produced with the intention that one or more other participants (the addressees, the audience, the ‘listeners’) attend to them, are able to perceive them, recognize the behaviour as an instance of a communica-tive action, try to understand them and possibly act upon them in one way or another; preferably with the effect that the producer of the communicative action had intended to achieve.”


ferently, it cannot be detached from the situation, i.e. the background which provides relevant aspects for the act of semiosis. In the case of Puccini’s Tosca, the whole plot, with both preceding and forthcoming events, constitutes the background for the analyzed aria. Situation in turn is included in the so-called pheno-world, the third sphere indispensable for an appropriate un-derstanding of the conceptual mechanisms that govern the semiotic space. In this sphere we find everything that is accessible to human thought “includ-ing the physical world with all its features and regularities and constrains on human action as well as beliefs and counterfactual activities” (Brandt & Brandt 2005: 226). For Puccini’s opera, the pheno-world is not limited to the plot, but it transgresses its boundaries pointing to more universal values, attitudes and ‘realia’ generally understood and accepted in the human world.

Establishing the semiotic space is merely a starting point for the re-construction of the ultimate message conveyed in the aria. Drawing on the insights made by Brandt & Brandt (2005), meaning construction as well as its further reception requires activating two basic input spaces, the Presen-tation Space and the Reference Space. The Reference Space, which is always de-rived from the Semiotic Space, relates to the actual content which is the very event occurring in the aria, viz. the blackmail Floria Tosca experiences from Baron Scarpia, and its tragic consequences. It seems that the basis for the Reference Space is the text being the verbal mode responsible for the so-called propositional content of the aria (cf. Zbikowski 2009b). Its detailed analysis is presented below.

First, to fully account for the contribution of the text to an overall un-derstanding of the main heroine’s despair as depicted in the analyzed aria, a few elements have to be taken into consideration: the content of the text and its semiotic resources (e.g. syntax, punctuation, repetitions of words, cf. Bezemer & Kress 2008; see also Lerdahl & Jackendoff 1983) which addi-tionally strengthen the message conveyed.

The lyrics of Tosca, written prior to Puccini’s composition by two libret-tists, Luigi Illica and Giuseppe Giacosa, are based on the original 1887 story La Tosca by the French playwright Victorien Sardou (www.britannica.com). Both the original version as well as English translation of the analyzed frag-ment is presented in the table below:

Vissi d’Arte in Italian English translation of Vissi d’Arte

Vissi d’arte, vissi d’amore,non feci mai male ad anima viva!

Con man furtivaquante miserie connobi aiutai.

I lived for my art, I lived for love,I never did harm to a living soul!

With a secret handI relieved as many misfortunes as I knew of.

Table 1. The lyrics of Vissi d’Arte aria


Sempre con fè sincerala mia preghiera

ai santi tabernacoli salì.Sempre con fè sincera

diedi fiori ag laltar.

Always with true faithmy prayer

rose to the holy shrines.Always with true faith

I gave f lowers to the altar.

Nell’ora del doloreperchè, perchè, Signore,

perchè me ne rimuneri così?

In the hour of griefwhy, why, o Lord,

why do you reward me thus?

Diedi gioiell idella Madonna al manto,e diedi il canto agli astri, al ciel,

che ne ridean più belli.

I gave jewels for the Madonna’s mantle,and I gave my song to the stars, to heaven,

which smiled with more beauty.

Nell’ora del doloreperchè, perchè, Signore,

ah, perchè me ne rimuneri così?

In the hour of griefwhy, why, o Lord,

ah, why do you reward me thus?

Vissi d’Arte in which the main heroine Floria Tosca laments over her unde-served predicament is considered one of the most moving female arias of all times. The content of the text is divided into four verses, arranged in three thematic blocks. In the first part Tosca’s positive deeds as a person are pre-sented (I never did harm to a living soul!; With a secret hand/ I relieved as many misfortunes as I knew of). The second part describes Tosca’s spiritual di-mension and portrays the heroine as a religious person (Always with true faith/ my prayer/ rose to the holy shrines./ Always with true faith/ I gave f lowers to the altar; I gave jewels for the Madonna’s mantle./ and I gave my song to the stars to heaven, /which smiled with more beauty). The third part is main-tained in the form of religious rhetoric. It resembles a prayer to God recited by an anguished woman who cannot understand why she has to deserve the fate she experiences: if God is to reward according to human actions, as all those in Tosca’s life, the main heroine should not experience the anguish she undergoes. Tosca’s feeling of injustice and disbelief manifests itself in a recursive rhetoric question to God (In the hour of grief why, why, o Lord, /ah, why do you reward me thus?) and illustrates the inevitable fate she is to experience.

The primary objective of the verbal layer of Vissi d’Arte is to construct and communicate meaning by “directing the attention of another person to objects or concepts within the same referential frame” (Tomasello 1999, after Zbikowski 2009b: 363). The role of text in the examined aria is very much in line with the above mentioned statement since, in comparison with other modes of communication involved, it remains the most precise form of mak-ing meaning which allows us to plunge into the literary content of the aria and capture the predicament the main heroine experiences.

Much as it is possible to examine the content of the textual layer of Vissi d’Arte in detail, little is to be said about its metrics and the way it can contribute to the overall reception of the aria. Known for its uneven pros-ody (cf. McKee 1998), the aria is virtually bereft of elements responsible for


establishing rhythmic patterns that could additionally boost its final inter-pretation. Instead, Vissi d’Arte conforms to the style of the delivery where it is music rather than text that assumes the role of a prosodic factor. Maeder (2013: 280) makes a similar observation about arias in general:

[In an aria] (…) the whole orchestra participates at every level. Melody, not prosody, typifies the voice’s utterances. An aria is characterized by repeti-tion: the orchestra quotes the voice’s melody; violins or any other instrument dialogue with the voice; in the opening section the orchestra announces the main themes and melodies; an aria consists of couplets; several sections are repeated; the musical constraints cause the voice to repeat certain words, sentences, and single vowels are prolonged. In an aria, voices tend to cover the whole range of notes, from the very low to the very high.

Still, the communication in opera should not be narrowed down to the se-mantics of every word or phrase sung, or their prosodic patterns. The au-dience (i.e. the hearer of the communicative act) craves for something that exceeds the boundaries of everyday communication, e.g. the emotional input involved. In this sense, as observed by Machin & Richardson (2012: 331), “our ears and brains are finely tuned not just to listen for the meanings of words and grammar, but also to the manner in which these are delivered – to voice quality and to rhythms of speech.” The above-mentioned statement imme-diately directs us to music understood here as an indispensable component of the text in opera, music, without which an in-depth interpretation and understanding of any operatic performance remains incomplete and futile. Adopting the Brandt and Brandt model (2005), it seems that music can suc-cessfully serve as the Presentation Space: it becomes the “affective content” (Jackendoff 2008), also described as “music-as-felt” (Brandt 2009) that adds an emotional component to an already existing content expressed in the verbal mode.5 Let us brief ly scrutinize its elements.

5 The notion of MUSIC-AS-FELT as proposed by P.A. Brandt (2009) is yet another stage in a complex process of conceptualizing music which manifests itself via conceptual integration. In Brandt’s opinion, the first blending operation in music can be observed on a formal level and al-ways results in the auditive perception of music, both while adopting the position of the listener as well as the performer (here we talk about the so-called MUSIC-AS-HEARD phenomenon which becomes the essence of the speaker-hearer interaction in the Semiotic Space). Adopting Brandt’s line of thought, rhythmic patterns that form the presentation space (indicated by such elements as pulse, tempo, or meter and usually expressed as ‘beats’) are mentally mapped onto melodic patterns being the elements of the reference space (expressed by duration, pitch, intensity, or timbre and dubbed as ‘tones’), thus giving rise to a blended melo-rhythmic phrase, or ‘tonal beats’ (Brandt 2009: 10), known more universally as the phrase pattern. To appropriately interpret the obtained ‘tonal beats’ expressed in the phrase pattern, it is essential that both players of music as well as its listeners activate the harmonic pattern as manifested by orchestration, allowing for such necessary elements as, e.g. gamut or chorus, thus giving rise to the ultimate stage of blend creation, in which the phrase pattern becomes a meaningful musical utterance. Due to the fact that the proposed paper is not musicology- but linguistics- oriented study, the deconstruction of the MUSIC-AS-HEARD stage will be omitted here.


The first two lines of Vissi d’Arte aria are maintained in a very rare key, E f lat minor, considered to be the darkest and most somber key of all (Orawski 2014). Compositions written in this key usually evoke extreme feelings: maj-esty, openness and love on the one hand and grief, fatalism, hopelessness, and resignation on the other (Orawski 2014). This character of E f lat minor key is also heard in the melodic line which accompanies the enumeration of Tosca’s deeds as found in the text (I lived for my art, I lived for love, I never did harm to a living soul! With a secret hand I relieved as many misfortunes as I knew of ). Tosca’s image as a good and faithful person presented in the text corresponds with the way it is sung: the aria starts in time signature 2/4 as pianissimo, with the added markings: dolcissimo (very soft), appassionato (im-passioned) and con grande sentimento (with great feeling). Therefore, the first bars of the aria must be sung very piano and very slowly, however with the intense feeling, not only to illustrate Tosca’s honesty and love for life, but also to start marking her anguish which revolves throughout the whole aria. Next, starting with the word ‘sempre’, the key changes into E f lat major to remain so till the end of the aria, with a new time signature 4/4. Still, the dynamics is maintained with lento appassionato, this time however, to indicate Tosca’s slow intense sorrowful prayer. E f lat major is considered to be the key of de-votion as well as intimate conversation with God (www.wmich.edu), which very well corresponds with appropriate verses of the text (In the hour of grief why, why, o Lord, ah, why do you reward me thus?). The phrase, repeated twice in the aria, is also sung with the dynamicity changing into crescendo molto (growing louder) which is to imitate the cry of despair of the main heroine.6

The modes representing the Presentation Space and the Reference Space are “functionally distinct” (Brandt & Brandt 2005: 228), which means that they have the potential to structure our thought in two different ways. The simultaneous use of the two modes in operatic performances proves that the cognitive processing of both language (text) and music stems from the same cognitive resources (Zbikowski 2009b; Jackendoff 2008) and is bound up with activating similar cognitive capacities, one of which is the cogni-tive capacity “to integrate combinatorially stored representations in working memory” (Jackendoff 2008: 196, emphasis mine). Therefore, they undergo the blending process in the sense of Fauconnier & Turner (2002).6 Communicating meaning via music is a complex task, mainly due to the fact that it is not ex-plicitly stated whether this mode has the potential to communicate meaning in its broader sense, i.e. the one that, from a musicological standpoint, is not “relegated to the status of soundtrack” (Frith 1993: 4). For this reason, the notion of musical meaning is frequently recalled (Kühl 2007; also Laske 1973). Obviously, musical meaning may be interpreted in its simple form as “a conse-quence of formal interactions, e.g. of successions of fully intramusical structures” (Bernstein 1976, after Antović 2011) that consist in joining music-related phenomena (e.g. timbre, form, phrase) together. Upon scrutiny, however, we discover that there is much more to musical meaning than only technical decoding of the score, and that it is primarily humanistic meaning that remains a driving force behind the semantics of music (Kramer 2004).


The integration of two distinct modes in the analysed aria does take place and is based on one of the most fundamental vital relations of con-ceptual blending, analogy, which facilitates the perception of correlations between elements from two distinct domains in the process of human com-munication (cf. Zbikowski 2018).7 Thus, we arrive at the very first stage of blending, known as the Virtual Space in the Brandt & Brandt model, where the audio-verbal manifestation of meaning occurs for the first time (cf. Pérez-So-brino 2014).8 This meaning, however, is “not inherent in music but neither is it in language”(McClary 1991: 21), being rather the “affective significance of the (…) event” (Brandt 2009: 12), as if subjectively felt by an individual recipient.9

Upon scrutiny, it seems that the first stage of blending process cor-responds, to a great extent, with text painting, the term used by musicians and also successfully put forward in a music-cum-text cognitive analysis (Zbikowski 2008, 2009a, 2009b; also Górska 2018). Text painting consists in

“writing music that reinforces the literal meaning” (Pérez-Sobrino 2014: 154). Allowing for the fact that many operas use extensively the libretto written prior to the instrumental version, as it happens in the case of Puccini’s Tosca, we can observe how music can “suggest or ‘paint’ an image that is referred to in the text itself “ (Górska 2018: 15), which gives us a fascinating image of the contribution of the non-verbal to capture and elaborate on the meaning of the verbal.10

Obviously, the meaning of Vissi d’Arte aria understood as a multimodal blend cannot be successfully decoded at the stage of Virtual Space since of-fering music as if it were to ‘paint’ the text does not provide any satisfactory clue as to how the ultimate meaning of the aria should be properly decoded. Therefore, it is necessary to establish the Relevance Space, responsible for re-calling appropriate schemas that frame the blend from the Virtual Space in a further elaboration process. The schemas pertain to both text and music, and are usually based on our “shared cultural conventions” (Machin & Richardson 2012: 330; also Tagg 1982; Cooke 1959). This allows us to recognize the text as a kind of lament due to mental ‘decoding’ of culturally shared conventions 7 Analogies between music and text are also stressed by Ludwig Wittgenstein. In his Philosphical Investigations, he makes the following observation: “Understanding a sentence is much more akin to understanding a theme in music than one may think” (2001: 527, after Zbikowski 2009b: 359).8 On the importance of merging music and text for an operatic performance see, e.g. Campbell (1994) and Simonton (2000).9 The notion of ‘as-if-ness’ for the Virtual Space is broadly discussed in Brandt & Brandt (2005: 230-232).10 In one of his papers Zbikowski poses an intriguing question with reference to the notion of text painting (2018: 4). How, he asks, “a sequence of musical sounds can represent a non-musical phenomenon, especially when that phenomenon has no significant sonic component?” And he makes the following concluding statement: “My argument has been that such representations rely on humans’ ability to make analogical connections between disparate domains.” What is also vital to know about analogies found between music in text is that they are “always framed by context” (Zbikowski 2018: 4). This means that every time we deal with the phenomenon of text painting,

“a larger interpretive framework” (Zbikowski 2018: 4) has to be recalled.


(e.g. the presence of repetitions, rhetoric questions directed to God, intimate conversation with the Almighty, all components being associated with such a genre). Music is also embedded in cultural conventions: we may not possess the ability to read musical notation, but we perfectly recognize the mood in which music is written since we subconsciously activate culturally agreed-upon patterns for music that evoke sadness, emotional pain, or despair (e.g. the composition written in minor key, slow tempo, appropriately chosen pitch ranges, articulation, or the dynamics of the melodic line).

At this point we arrive at the Meaning Space, the last stage of blend elab-oration where a properly framed representation of music-cum-text from the Virtual Space becomes the meaning which emerges “in the interpretive pro-cess” (Brandt & Brandt 2005: 238). Thus we obtain the utterance of Vissi d’Arte which not only illustrates Tosca’s despair, feeling of injustice and helplessness as expressed by the two modes but also allows the integration of music and text to be experienced intersubjectively (cf. Brandt 2009), by each and every potential hearer. This multimodal blend as “finally experienced” (Brandt 2009: 12) is what makes Puccini’s aria so moving and, at the same time, so successful whenever heard by the audience.

A schematic illustration of the cognitive processes responsible for the conceptualization and appropriate meaning decoding of Vissi d’Arte aria is presented in the form of Brandt and Brandt’s six-space model below:

ConclusionsThe proposed analysis is a modest contribution to a broad discussion upon operatic performances viewed as meaning-making events from the perspec-tive of multimodal as well as cognitive linguistic studies. A multimodal char-acter of operatic performances is undisputable since, as observed by Rossi & Sindoni “ (…) meanings in opera can only be understood by reintegrating all the systems, resources, and components that work in combination” (2017:

Figure 2. Analysis of Vissid’Arte using Brandt & Brandt’s model


81). Given that a pivotal assumption of multimodality is the efficiency of communication obtained by integrating various modes, and, what follows, successful meaning decoding in the speaker-hearer interaction, it seems that the revised six-space model of conceptual integration as delineated by Brandt & Brandt (2005) proves a useful tool here as it conveniently captures the essence of one of the basic mental operations – blending – and success-fully frames it within a communicative context.

Two basic modes used in Vissi d’Arte co-create a multimodal message of the aria, each however is to portray the predicament of the main heroine, Floria Tosca, in a different way. The objective of text is mainly limited to factual data referring to Tosca’s life (i.a. her readiness to help the needy, her piety, and, finally her disagreement with the fate she is to endure), the emo-tional component being reduced here to only a few minor semiotic elements (e.g. a recurrent rhetoric question perchè?/why?). The element responsible for the ‘affective’ content of the message (cf. Jackendoff 2008: 197) is the second mode, viz. music with all the breadth of shared schemas of melodic patterns used in the Western culture to account for sadness and despair (cf. Zbikowski 2009b, Lerdahl 2003, Machin & Richardson 2012).

Lastly, the question arises what makes Vissi d’Arte one of the most suc-cessful and moving arias. It seems that the answer lies in its intersubjec-tive reception by the audience which, as pointed out by Zlatev et al. (2008: 1), inheres in “sharing of experiential content (e.g., feelings, perceptions, thoughts, and linguistic meanings) among a plurality of subjects.” This seems to confirm the observation that when properly framed as a multi-modal message, the meaning of the aria gradually develops in the course of interpretation, being simultaneously ‘guided’ by relevant schemas of shared cultural conventions.

ReferencesAntović, Mihailo 2011: Musical metaphor revisited: Primitives, universals

and conceptual blending. URL: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1763503. ED: 14 May 2015.

Bernstein, Leonard. 1976: The Unanswered Question. Place: Harvard Univer-sity Press.

Bezemer, Jeff, Gunther Kress 2008: Writing in multimodal texts: A social semiotic account of designs for learning. Written Communication 25(2), 166-195.

Brandt, Per Aage 2005: Mental spaces and cognitive semantics: A critical comment. Journal of Pragmatics 37, 1578-1594.


Brandt, Line, Per Aage Brandt 2005: Making sense of a blend: A cognitive-se-miotic approach to metaphor. Annual Review of Cognitive Linguistics 3, 216-249.

Brandt, Per Aage 2009: Music and the abstract mind. The Journal of Music and Meaning vol. 7.3, 1-15.URL: http://www.musicandmeaning.net/issues/showArticle.php?artID=7.3. ED: 20 June 2018.

Campbell, Stuart (ed.) 1994: Russians on Russian Music: 1830-1888. An Anthology. Cambridge: Cambridge University Press.

Cienki, Alan 2008: Looking at analysis of mental spaces and blending. / Look-ing at and experiencing discourse in interaction. In: Todd Oakley and Anders Hougaard (eds.) 2008: Mental Spaces in Discourse and Interaction. Amsterdam-Philadelphia: John Benjamins, 235-245.

Cooke, Deryck 1959: The Language of Music. Oxford: Clarendon.Di Giovanni, Elena 2018: Audio description for live performances and audi-

ence participation. The Journal of Specialised Translation 29, 189- 211.Drabkin, William 2014: Analysis. In: Helen M. Greenwald (ed.) 2014: The Ox-

ford Handbook of Opera. Oxford: Oxford University Press, 245-271.Evans, Vyvyan, Melanie Green 2006: Cognitive Linguistics: An Introduction.

New York and London: L. Erlbaum.Fauconnier, Gilles, Mark Turner 2002: The Way We Think: Conceptual Blending

and the Mind’s Hidden Complexities. New York: Basic Books.Frith, Simon 1993: Youth, music, television. In: Goodwin Frith and Lawrence

Grossberg (eds.) 1993: Sound and Vision. New York: Routledge, 67-84.Górska, Elżbieta 2018: From Music to Language and Back. Language, Culture,

Mind and Society 2: 82-103. Harris-Warrick, Rebecca 2016: Dance and Drama in French Baroque Opera.

A History. Cambridge: Cambridge University Press.Heylen, Dirk 2009: Understanding speaker-listener interactions. Inter-

speech-2009, 2151-2154. URL:https://core.ac.uk/download/pdf/11473605.pdf. ED: 16 Nov 2019.

Hutcheon, Michael, Linda Hutcheon 2010: Opera. Forever and always mul-timodal. In: Ruth Page (ed.) 2010: New Perspectives on Narrative and Mul-timodality. London and New York: Routledge, 65-77.

Jackendoff, Ray 2008: Parallels and nonparallels between language and mu-sic. Music Perception 26(3), 195-204.

Jewitt, Carey (ed.) 2009: The Routledge Handbook of Multimodal Analysis. Lon-don: Routledge.

Jewitt, Carey, Jeff Bezemer, Kay O’Halloran 2016: Introducing Multimodality. London: Routledge.

Kardela, Henryk 1992: Gramatyka kognitywna jako globalna teoria języka [Cognitive grammar as a global theory of language]. In: Iwona


Nowakowska-Kępna (ed.) 1992: Język a Kultura. Podstawy metodologiczne semantyki współczesnej [Language and Culture. Methodological Foundations of Contemporary Semantics], 9-22.

Kramer, Lawrence 2004: Musical Meaning. University of California Press.Kress, Gunther 2009: Multimodality: A Social Semiotic Approach to Contemporary

Communication. London: Taylor & Francis.Kühl, Ole 2007: Musical Semantics: Cognitive Musicology and the Challenge of

Musical Meaning. Bern: Peter Lang.Langacker, Ronald W. 2009: Investigations in Cognitive Grammar. Berlin, New

York: Mouton de Gruyter.Lerdahl, Fred 2003: The sounds of poetry viewed as music. In: Johan Sund-

berg, Lennart Nord and Rolf Carlson (eds.) 2003: Music, Language, Speech and Brain. London: Macmillan, 34-47.

Lerdahl, Fred, Ray Jackendoff 1983: A Generative Theory of Tonal Music. Cam-bridge, MA: MIT Press.

Machin, David, John E. Richardson 2012: Discourses of Unity and Purpose in the Sounds of Fascist Music: A Multimodal Approach. Critical Discourse Studies 9(4), 329-345.

Maeder, Constantino 2011: Argumentative, iconic and indexical structures in Schubert’s “Die schöne Müllerin”. In: Pascal Michelucci, Olga Fisher and Christina Ljungberg (eds.) 2011: Semblance and Signification. Amsterdam and New York: John Benjamins, 389-404.

Maeder, Constantino 2013: Opera, oratorio and iconic strategies. In: Lars Elleström, Olga Fisher and Christina Ljungberg (eds.) 2013: Iconic In-vestigations. Amsterdam and Philadelphia: John Benjamins, 275-294.

McClary, Susan 1991: Feminine Endings. Music, Gender, and Sexuality. Minnesota and Oxford: University of Minnesota Press.

McKee, David 1998: Opera in the vernacular. The Opera Quarterly 14(4), 149-152. URL: https://doi.org/10.1093/oq/14.4.149. ED: 18 November 2019.

Nordquist, Richard 2019: Definition and examples of conceptual blending. Glossary of grammatical and rhetorical terms. URL: https://www.thoughtco.com/what-is-conceptual-blending-cb-1689780?print. ED: 21 September 2019.

Oakley, Todd, Seana Coulson 2008: Mental spaces and metaphoric lan-guagein discourse. In: Todd Oakley and Anders Hougaard (eds.) 2008: Mental Spaces in Discourse and Interaction. Amsterdam and Philadelphia: John Benjamins, 27-50.

Orawski, Piotr 2014: O odchodzeniu i dźwięku Es [On leaving and E-f lat minor key] Piękno muzyki [The beauty of music]18. URL: https://wszyst-koconajwazniejsze.pl/piotr-orawski-piekno-muzyki-18-o-odchodzeniu-i-dzwieku-e/ED: 25 May 2018.

100 L aMiCuS 2019 no. 3

Pérez-Sobrino, Paula 2014: Multimodal cognitive operations in classical mu-sic. Vigo International Journal of Applied Linguistics 11, 137-168.

Rabb, Theodore K. 2006: Opera, musicology and history. Journal of Interdis-ciplinary History 36(1), 321-330.

Rossi, Fabio, Maria Grazia Sindoni 2017: The phantoms of the opera: Toward a multidimensional interpretative framework of analysis. In: Maria Grazia Sindoni, Janina Wildfeuer and Kay L. O’Halloran (eds.) 2017: Mapping Multimodal Performance Studies. New York and London: Rout-ledge, 61-84.

Simonton, Dean K. 2000: The music or the words? Or how important is the libretto for an opera’s aesthetic success? Empirical Studies of the Arts 18(2), 105-118.

Strohm, Reinhard 1985: Essays on Handel and Italian Opera. Cambridge: Cam-bridge University Press.

Tagg, Philip 1982: Nature as a musical mood category. IASPM Norden’s Work-ing Paper Series, 1-25.

URL: https://www.tagg.org/articles/xpdfs/nature.pdf . ED: 14 April 2019. Tomasello, Michael 1999: The Cultural Origins of Human Cognition. Cambridge,

MA: Harvard University Press.Veale, Tony, Diarmuid O’Donoghue 2000: Computation and blending. Cog-

nitive Linguistics 11(3/4), 253-281.Wittgenstein, Ludwig. 2001 /1953/ Philosophical Investigations 3rd ed. Gertrude

E.M. Anscombe (transl.). Oxford: Blackwell.Zbikowski, Lawrence 2008: Metaphor and music. In: Raymond W. Gibbs

(ed.) 2008:The Cambridge Handbook of Metaphor and Thought. Cambridge: Cambridge University Press, 502-523.

Zbikowski, Lawrence 2009a. Review of Ole Kühl’s Musical Semantics: Cog-nitive Musicology and the Challenge of Musical Meaning. Musicae Scientiae 13(2): 441-473.

Zbikowski, L. 2009b: Music, language, and multimodal metaphor. In: Charles Ch. Forceville, Eduardo Urios-Aparisi (eds.) 2009: Multimodal Metaphor. Berlin: Mouton de Gruyter, 359-381.

Zbikowski, Lawrence 2015: Words, music and meaning. Signata 6, 143-164.Zbikowski, Lawrence 2018: Conceptual blending, creativity, and music. Mu-

sicae Scientiae 22(1), 6-23.Zlatev, Jordan, Timothy P. Racine, Chris Sinha,Esa Itkonen 2008: Intersub-

jectivity: What Makes us Human? In: Jordan Zlatev, Timothy P. Racine, Chris Sinha, Esa Itkonen (eds.) 2008 The Shared Mind: Perspectives of In-tersubjectivity. Amsterdam: Benjamins, 1-14.


Streszczenie

Gdzie tekst łączy się z muzyką: analiza multimodalna arii Vissi d’Arte

Artykuł ten prezentuje studium przypadku – multimodalną analizę arii sopranowej Vissi d’Arte z II. aktu opery Giacomo Pucciniego pt. Tosca. Od-wołując się do głównych założeń teorii integracji pojęciowej w ujęciu Gillesa Fauconniera i Marka Turnera (2002), w szczególności zaś do zrewidowanego sześcioprzestrzennego modelu zaproponowanego przez duńskich badaczy, Line Brandt i Per Aage Brandta (2005), autorka pragnie ukazać, do jakiego stopnia łączenie dwóch głównych kanałów komunikacyjnych, muzyki oraz tekstu, może być postrzegane jako operacja integracji pojęciowej, będąca odbiciem „zdolności kognitywnych człowieka” (Zbikowski 2009a: 448, tłum. AMH, zob. także Kühl 2007), a także w jaki sposób integracja różnych modal-ności może kształtować ostateczną recepcję przedstawienia operowego, bio-rąc pod uwagę interakcję zachodzącą pomiędzy nadawcą – wykonawcą arii i odbiorcą – publicznością.

Opera, wraz z całym swym bogactwem i różnorodnością elementów, takich jak muzyka, libretto, gesty, scenografia, kostiumy czy taniec, uzna-wana jest za jeden z najbardziej złożonych i jednocześnie najbardziej wy-magających przykładów komunikowania określonej treści (zob. Rossi i Sin-doni 2017). Choć stanowi ona interesujący obiekt badań dla językoznawców, pozostaje wciąż materiałem analizowanym głównie z muzykologicznego punktu widzenia. Niewielkie zainteresowanie operą w badaniach języko-znawczych (także kognitywnych), jak również wśród badaczy zajmujących się komunikacją multimodalną wynika z dwóch powodów. Po pierwsze, opera, ze względu na swoją złożoną i różnorodną strukturę, stanowi zbyt wielkie wyzwanie dla analizy językoznawczej. Drugim powodem, dla którego ję-zykoznawcy rezygnują z uznania jej za wartościowy materiał empiryczny, jest fakt, iż będąc „niestabilnym medium” (Drabkin 2014: 245, tłum. AMH), posiadającym zbyt wiele zmiennych, opera postrzegana jest jako mało wia-rygodne, a tym samym mało reprezentatywne źródło badawcze.

Będąc dziełem złożonym, opera pozostaje interesującym materiałem do badań nad multimodalnością, wpisuje się bowiem w jej trzy zasadnicze kryteria. Po pierwsze, znaczenie danego dzieła operowego konstruowane jest przy udziale użytych w niej różnych filtrów komunikacyjnych, z których każdy ma nie tylko „różnorodne możliwości, ale także ograniczenia” (Jewitt, Bezemeer & O’Halloran 2016: 3, tłum. AMH). Po drugie, proces powstawa-nia określonych znaczeń w operze skutkuje powstaniem tzw. multimodal-nych całości, których odkodowywanie ścisłe łączy się z trzecim założeniem,

102 L aMiCuS 2019 no. 3

a mianowicie skrupulatnym badaniem wszystkich modalności biorących udział w kształtowaniu się ostatecznego znaczenia (zob. Jewitt, Bezemer i O’Halloran 2016: 3).

Dokonując analizy multimodalnej przedstawienia operowego, należy pamiętać, że nie istnieje „jeden zdefiniowany model mogący posłużyć ba-daniu złożonej multimodalnie struktury” (Rossi i Sindoni 2017: 65, tłum. AMH). Stąd też świadome przywoływanie pojęć i teorii rozwijanych również w innych dyscyplinach (np. na gruncie językoznawstwa kognitywnego), które mogą okazać się przydatne w analizie dzieła multimodalnego.

Istotą wszelkich badań multimodalnych pozostają niezmiennie dwa kryteria. Są to: (i) komunikacja multimodalna oraz (ii) integracja multimodalna (Jewitt, Bezemer i O’Halloran 2016). Poprzez komunikację multimodalną rozu-miemy sposób, w jaki dana wiadomość zostaje przekazana przez nadawcę komunikatu jego odbiorcy (wraz z uwzględnieniem wszelkich modalno-ści biorących udział w tym komunikacie), oraz umiejętność poprawnego odkodowania przez odbiorcę zawartej w komunikacie multimodalnej tre-ści. Integracja multimodalna bada z kolei proces stapiania różnych modal-ności, mający na celu uzyskanie pełniejszego i, niejednokrotnie, bardziej intrygującego znaczenia. Obydwa wyżej wymienione kryteria, multimo-dalna komunikacja i multimodalna integracja, mają wiele punktów stycznych z postulatami, które odnajdujemy w założeniach językoznawstwa kogni-tywnego, głównie w odniesieniu do uwarunkowanego kontekstowo dy-namicznego konstruowania znaczenia, jakie zachodzi podczas interakcji między nadawcą i odbiorcą. Z tego też powodu przywołanie teorii integra-cji pojęciowej, a ściślej modelu integracji pojęciowej w jego zrewidowanej formie proponowanej przez Brandt i Brandta (2005), i zastosowanie go w proponowanym w artykule studium przypadku wydaje się uzasadnione.

Kognitywno-multimodalna analiza słynnej arii Vissi d’Arte z opery Tosca Pucciniego koncentruje się wokół dwóch kluczowych dla opery modalności, tzn. tekstu oraz muzyki. Analiza nie ma zatem charakteru holistycznego, a jedynie cząstkowy, ukazujący procesy konstruowania znaczenia, zawężone do działania dwóch wyżej wymienionych modalności.

W toku rozważań nad wybranym materiałem empirycznym poruszono dwa istotne problemy: (i) roli, jaką odgrywają muzyka i tekst w procesie kształtowania znaczenia, oraz (ii) zjawiska integracji na poziomie koncep-tualnym, obejmującego wybrane elementy wspomnianych powyżej modal-ności, często postrzeganych jako dwa odrębne choć kognitywnie powiązane ze sobą kanały komunikacyjne (zob. Jackendoff 2008).

W schemacie Brandtów odnajdujemy sześć przestrzeni niezbędnych, zdaniem jego autorów, do prawidłowego rozumienia wyłaniającego się w komunikacie znaczenia. Są to kolejno: przestrzeń semiotyczna (ang. se-miotic base space), przestrzeń referencyjna (ang. reference space), przestrzeń


prezentacyjna (ang. presentation space), przestrzeń relewancyjna (ang, rele-vance space), a także przestrzeń amalgamatu o charakterze dwuetapowym, w którym wyróżniamy przestrzeń wirtualną (ang. virtual space) oraz prze-strzeń znaczeniową (ang. meaning space). Istotą modelu pozostaje bazowa przestrzeń semiotyczna, kluczowa dla samego aktu komunikacji, zarówno z perspektywy nadawcy komunikatu (tu: sopranistki, która wciela się w po-stać głównej bohaterki, Florii Toski), jak i jego odbiorcy, będącego w anali-zowanym przypadku tzw. odbiorcą zbiorowym, czyli publicznością (zob. Di Giovanni 2018).

Nie mniej znaczący wpływ na kształtowanie komunikatu zawartego w arii Vissi d’Arte oraz jego recepcję mają dwie inne przestrzenie: przestrzeń referencyjna, której podstawę stanowi tekst arii, oraz przestrzeń prezen-tacyjna, gdzie odnajdujemy warstwę muzyczną. Łatwo zauważyć, iż oby-dwie modalności są zasadniczo odmienne, co oznacza również, że są one w stanie strukturyzować nasze myśli na dwa odrębne sposoby. Dzięki war-stwie werbalnej, zawartej w przestrzeni referencyjnej, dowiadujemy się o losach głównej bohaterki, Florii Toski, która pada ofiarą przewrotnego szantażu zainicjowanego przed barona Scarpię, prefekta rzymskiej policji, Aby uchronić swojego ukochanego przed egzekucją, Tosca zgadza się spę-dzić noc z baronem. Poczucie wstydu, rozpaczy i udręczenia manifestuje się w warstwie werbalnej, m.in. za pomocą licznych pytań retorycznych sta-wianych Bogu przez Toskę, oraz powtórzeń mających na celu mocniejsze wyrażenie niesprawiedliwości, jakie odczuwa główna bohaterka. Wydaje się jednak, iż tekst informuje odbiorcę głównie o faktach dotyczących roz-paczliwej sytuacji Toski, natomiast nie wykazuje już takiego potencjału, jeśli chodzi o komunikowanie emocji. Dlatego też modalnością odpowiedzialną za tzw. „treść afektywną” (ang. af fective content, Jackendoff 2008) analizowa-nej arii, określaną również terminem „muzyka odczuwana” (ang. music-as-felt, Brandt 2009), jest właśnie warstwa muzyczna, pełniąca funkcję przestrzeni prezentacyjnej. Aria Vissi d’Arte, utrzymana na początku w tonacji es-moll, uważanej za jedną z rzadziej używanych w muzyce ze względu na ewokowa-nie mrocznego i posępnego nastroju (Orawski 2014), ilustruje skrajne uczucia: od otwartości, podniosłego nastroju i miłości do świata i ludzi, po rozpacz, fatalizm, poczucie beznadziei i ostatecznej rezygnacji. Warstwa muzyczna arii doskonale koresponduje z nastrojem głównej bohaterki ukazanym w tek-ście, o czym mogą świadczyć zawarte w partyturze dodatkowe informacje, takie jak np. sposób, w jaki w danym momencie powinna zostać zaśpiewana aria, zmiana tempa utworu czy wreszcie zmiana tonacji z es-moll na Es-dur.

Jednoczesne użycie dwóch modalności w analizowanej arii wskazuje, iż mentalne przetwarzanie treści werbalnej oraz warstwy muzycznej wyrasta z tych samych źródeł kognitywnych (Zbikowski 2009b; Jackendoff 2008) i wiąże się bezpośrednio z uruchamianiem podobnych zdolności kognityw-

104 L aMiCuS 2019 no. 3

nych, jedną z których jest „integrowanie w sposób ‘kombinatoryczny’ repre-zentacji przechowywanych w pamięci roboczej” (Jackendoff 2008: 196, tłum. AMH). Daje to nam podstawę sądzić, iż obie modalności podlegają procesowi amalgamacji w rozumieniu Fauconniera i Turnera (2002).

Integracja tekstu oraz warstwy muzycznej arii opiera się na jednej z istotnych relacji obserwowanych w procesie integracji pojęciowej, ana-logii, dzięki której możemy obserwować korelacje występujące pomiędzy elementami należącymi do dwóch różnych domen (zob. Zbikowski 2018). Jest to pierwsze stadium amalgamatu w rozumieniu Brandt i Brandta (2005), czyli przestrzeń wirtualna, w której po raz pierwszy mamy do czynienia z audio-werbalną manifestacją znaczenia zakodowanego w arii. Znaczenie to jednak nie tkwi na tym etapie ani w muzyce, ani w tekście, pozostając

„sensem afektywnym arii, odczuwanym jedynie pozornie przez indywidu-alnego odbiorcę komunikatu” (Brandt 2009: 12).

Nie sposób nie zauważyć, iż integracja tekstu i muzyki opisana powyżej w dużej mierze pokrywa się z terminem „malowanie tekstu” (ang. text pain-ting), używanym głównie przez muzyków i muzykologów, a ostatnio zapo-życzonym także na potrzeby analizy kognitywnej (Zbikowski 2008, 2009a, 2009b; Pérez-Sobrino 2014; Górska 2018). Analizując arię Vissi d’Arte, możemy bowiem zaobserwować, w jaki sposób muzyka dodatkowo ozdabia obraz, jaki widzimy w samym tekście, co daje oczywiście możliwość głębszej inter-pretacji, a tym samym pełniejszego odczytania ostatecznego znaczenia arii.

Oczywistym pozostaje fakt, iż znaczenie analizowanej Vissi d’Arte, ro-zumianej od tego momentu jako amalgamat multimodalny, nie może zo-stać skutecznie odkodowane na etapie pierwszego stadium amalgamatu w rozumieniu Brandt i Brandta (2005). W tym celu konieczne jest utworzenie przestrzeni relewancyjnej, dzięki której przywołane zostaną odpowiednie schematy powstałe w oparciu o „wspólne konwencje kulturowe” (Machin i Richardson 2012: 330; także Tagg 1982; Cooke 1959), odnoszące się zarówno do warstwy werbalnej, jak i muzycznej arii. To dzięki owym wspólnie po-strzeganym konwencjom, typowym dla określonego kręgu kulturowego (w badanym przypadku: kultury europejskiej), odbiorca komunikatu jest w stanie rozpoznać tekst jako lament, z uwagi na pojawiające się w nim typowe cechy tego gatunku (np. intymna rozmowa Toski z Bogiem, pyta-nia retoryczne kierowane przez nią do Najwyższego, powtórzenia). War-stwa muzyczna arii również podporządkowuje się konwencjom przyjętym w muzyce. Dzięki temu odbiorca jest w stanie rozpoznać nastrój utworu, po-nieważ mentalnie przywołuje uzgodnione wzorce odpowiedzialne w muzyce za odczuwanie smutku, bólu emocjonalnego czy rozpaczy (objawiające się np. obecnością tonacji molowych, powolnym tempem, odpowiednią dyna-miką i artykulacją), nie musząc jednocześnie być biegłym w notacji czy też jej odczytywaniu.


Kulminacyjny punkt wyłaniania się znaczenia w arii ma miejsce w ostatnim stadium amalgamatu, w przestrzeni znaczeniowej. To właśnie tutaj odpowiednio sformułowany w przestrzeni wirtualnej amalgamat, na który składają się muzyka i tekst, urasta do rangi znaczenia dynamicznie wyłaniającego się podczas interpretacji. Dzięki uruchomieniu przestrzeni znaczeniowej oraz odwołaniu do wcześniejszych przestrzeni, zwłaszcza przestrzeni semiotycznej i relewancyjnej, aria Vissi d’Arte ukazuje nie tylko rozpacz głównej bohaterki, jej poczucie krzywdy czy bezradności, wyrażo-nych za pomocą tekstu i muzyki, ale przede wszystkim ewokuje intersubiek-tywne odczucie rozpaczy. Tak skonstruowany amalgamat multimodalny, określany mianem „ostatecznie doświadczanego” (ang. finally experienced, Brandt 2009: 12), jest tym, co czyni Vissi d’Arte jedną z bardziej poruszających i jednocześnie jedną z popularniejszych arii.

Tłum. Agnieszka Mierzwińska-Hajnos

BibliografiaBrandt, Line, Per Aage Brandt 2005: Making sense of a blend: A cognitive-se-

miotic approach to metaphor. Annual Review of Cognitive Linguistics 3, 216-249.

Brandt, Per Aage 2009: Music and the abstract mind. The Journal of Music and Meaning 7(3), 1-15. URL: http://www.musicandmeaning.net/issues/showArticle.php?artID=7.3. DW: 20 June 2018.

Di Giovanni, Elena 2018: Audio description for live performances and audi-ence participation. The Journal of Specialised Translation 29, 189-211.

Fauconnier, Gilles, Mark Turner 2002: The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books.

Górska, Elżbieta 2018: From music to language and back. Language, Culture, Mind and Society 2: 82-103.

Jackendoff, Ray 2008: Parallels and nonparallels between language and mu-sic. Music Perception 26(3), 195-204.

Jewitt, Carey, Jeff Bezemer, Kay O’Halloran 2016: Introducing Multimodality. London: Routledge.

Kühl, Ole 2007: Musical semantics: Cognitive Musicology and the Challenge of Musical Meaning. Bern: Peter Lang.

Machin, David, John E. Richardson 2012: Discourses of unity and purpose in the sounds of fascist music: A multimodal approach. Critical Discourse Studies 9(4), 329-345.

Orawski, Piotr 2014: O odchodzeniu i dźwięku Es 18. URL: https://wszystko-conajwazniejsze.pl/piotr-orawski-piekno-muzyki-18-o-odchodzeniu-i-dzwieku-e/DW: 25 May 2018.

106 L aMiCuS 2019 no. 3

Pérez-Sobrino, Paula 2014: Multimodal cognitive operations in classical mu-sic. Vigo International Journal of Applied Linguistics 11, 137-168.

Rossi, Fabio, Maria Grazia Sindoni 2017: The phantoms of the opera: Toward a multidimensional interpretative framework of analysis. W: Maria Grazia Sindoni, Janina Wildfeuer and Kay L. O’Halloran (red.) 2017: Mapping Multimodal Performance Studies. New York and London: Rout-ledge, 61-84.

Tagg, Philip 1982: Nature as a musical mood category. IASPM Norden’s Work-ing Paper Series, 1-25.URL: https://www.tagg.org/articles/xpdfs/nature.pdf. DW: 14 April 2019.

Zbikowski, Lawrence 2008: Metaphor and music. W: Raymond W. Gibbs (red.) 2008: The Cambridge Handbook of Metaphor and Thought. Cambridge: Cambridge University Press, 502-523.

Zbikowski, Lawrence. 2009a. Review of Ole Kühl’s Musical semantics: Cog-nitive Musicology and the Challenge of Musical Meaning. Musicae Scientiae 13(2): 441-473.

Zbikowski, L. 2009b: Music, language, and multimodal metaphor. W: Charles Ch. Forceville, Eduardo Urios-Aparisi (red.) 2009: Multimodal Metaphor. Berlin: Mouton de Gruyter, 359-381.

Zbikowski, Lawrence 2015: Words, music and meaning. Signata 6, 143-164.Zbikowski, Lawrence 2018: Conceptual blending, creativity, and music. Mu-

sicae Scientiae 22(1), 6-23.

108 L aMiCuS 2019 no. 3

Francesca CarboneUniversità degli studi di Napoli “L’Orientale”

Identification of emotions by prosody and semantics in French spontaneous speech: A pilot study

When we talk about emotions, we refer to a complex and structured phenomenon that can in-f luence and be inf luenced by external and internal elements involved in communication such as context, behavior and feelings of other individuals. During the expression of emotions, these in-ternal and external factors can have effects on both verbal and non-verbal cues which can change according to the context and the intentions of the speakers. This complexity of the emotional phenomenon has induced researchers to focus the attention on individual cues (e.g., intonation, verbal content) in isolation from the others. Although it is possible that the transmission of emo-tional information is entrusted to a single type of cues, this is certainly the least frequent case in the spontaneous production of emotional speech, in which, usually, the emotional phonological information co-occurs with the other segmental features (e.g. syllables, words). Consequently, studies on individual emotional cues have not increased our understanding of how people inte-grate nonverbal and verbal cues in the expression of emotions. The aim of this study is to face this topic by assessing the contribution of prosodic parameters (i.e., fundamental frequency, intensity, speech rate) versus verbal content to the decoding of emotions by a perception test in which spon-taneous speech is manipulated through the application of filters, validated in previous research. French listeners are asked to identify sentences that were annotated and selected from a French spontaneous corpus (EmoTV). Stimuli are presented in three versions: integral (non-manipulated), verbal (the prosodic information is hidden by applying filters) and nonverbal (the verbal content is filtered and masked). Results display a higher accuracy for integral stimuli suggesting that emo-tions are decoded by the interaction of the two different channels. This confirms the adequacy of a cognitive model in which verbal and nonverbal cues to emotions are theoretically integrated.


ABST

RACT

Keywordsemotions, spontane-ous speech, semantics, prosody


109Identification of emotions by prosody and semantics...

1. Introduction

During the spontaneous production of emotional speech, multiple channels (e.g., intonation, verbal content) are involved in conveying emotions and, although it is possible, it is quite rare that the transmission of emotional information is entrusted to a single type of cues in everyday life communi-cation (Planalp 1999). To precisely capture the meaning, listeners must inte-grate information derived from segmental signals of language, like syllables, words, phrases (semantic cues), with suprasegmental information, such as voice tone and acoustic properties of the speech (e.g., prosodic cues, such as fundamental frequency, speech rate, intensity) (Planalp 1999; Nygaard & Queen 2008). Both of these aspects of spoken language, semantics and vocal cues, constitute essential components of successful interpersonal commu-nication. Despite their joint importance, however, they have been studied separately by scientists and, consequently, research does not clarify how exactly they interact during communication and how they contribute to the recognition of emotions.

On the one hand, previous studies on emotional prosody have demon-strated that basic emotions can be quite accurately recognized through a single channel in a forced-choice response format (Pell et al. 2009, Paulmann & Kotz 2008; Juslin & Laukka 2003; Borod et al., 2000; Banse & Scherer 1996; Murray & Arnott 1993; Ekman 1992; Scherer et al. 1991). For instance, Banse and Scherer (1996) explored how accurately emotions can be conveyed only by prosody in sentences that did not contain any meaningful lexical-seman-tic information (pseudo-sentences). Their results showed that listeners are able to detect the emotional state of their communicative partner only by the acoustic modulation of the voice. On the other hand, studies on emotional semantics have pointed out the power of emotional words, suggesting that emotionally connotated words are processed faster and more easily than neutral ones, even when they are presented without any prosodic cue (Kousta et al. 2009; Schacht & Sommer 2009; Scott et al. 2009; Estes & Verges 2008; Kanske & Kotz 2007; Bradley & Lang 1999).

Studies dealing with the transmission of emotions from dif ferent channels have mostly presented multi-modal emotional stimuli (generally individual words) in paradigms that include mismatching or incongruent information between channels, frequently using a Stroop task methodology (Filippi et al. 2017; Kreifelts et al. 2007, De Gelder & Bertelson 2003; De Gelder & Vroomen 2000). When prosody and semantics are compared, outcomes are varied and do not reveal a unique trend. Electrophysiological data have shown that the semantics of an utterance is processed before the prosodic information during emotion processing (e.g., Kotz & Paulmann 2007). In contrast, behavioral studies have illustrated how prosody can give rise to

110 L aMiCuS 2019 no. 3

potential biases in the processing of emotional verbal content, even when conflicting with verbal content and faces simultaneously (Filippi et al. 2017). Furthermore, the selection of emotional meaning is significantly better when the voice tone is congruent with affective meaning, indicating that emotional prosody may be used to process the linguistic content of lexical items (Paul-mann & Pell 2011; Nygaard & Lunders 2002). Moreover, the dominance of the two channels (i.e. semantics and prosody) could be related to cultural differ-ences. For instance, Kitayama and Ishii (2002; Ishii et al. 2003), who applied an auditory Stroop task, have found that interference effects between word meaning and vocal emotion could be mediated by culture and language. In their task, comparing judgments of Japanese and American English listeners on the recognition of emotional words, it was found that for the first group (Japanese) vocal emotion interfered with judgments of word meaning to the greatest extent, while for the second (American English), word meaning in-terfered with judgments of vocal emotion to the greatest extent.

Although the interrelation of prosody and semantics in the transmis-sion of the emotions is still not clear, however, there is some evidence that emotional expressions encoded by more than one information channel are recognized with greater accuracy, consistent with the idea of a multi-modal advantage (Frühholz et al. 2016; Milesi et al. 2014; Paulmann & Pell 2011). Paulmann & Pell (2011) ran an identification task using different kinds of stimuli (i.e., facial expressions, semantic information conveyed by texts, emotional prosody). Their results confirmed that emotion recognition is sig-nificantly better in response to multi-modal versus uni-modal stimuli. From a cognitive point of view, this advantage suggests an automatic integration of emotional information across channels, which enhances the accessibility of emotion-related knowledge in memory (cf. emotion nodes or concepts, Russell & Lemay 2000; Halberstadt et al. 1995).

Studies using spontaneous emotional speech samples to compare mul-ti-modal versus uni-modal versions are even more sporadic (Huang et al. 2017; Kim & Provost 2016; Jürgens et al. 2013), which could be attributed to a number of factors, including background noise and ethical reasons (Johnstone & Scherer 2000). In fact, basic emotions are still rare events in everyday conversations and it is dif ficult to observe them being expressed in uni-modal manner (Cowie & Cornelius 2003). In order to overcome these issues, a number of researchers have used stimuli produced by actors in laboratories. This procedure, however, has often led to more stereotyped voices, which do not always ref lect the spontaneous expression of emotions (Drolet et al. 2012, Laukka et al. 2012; Bänziger & Scherer 2010; Scherer 2003). Since natural stimuli are considered more ecological (Scherer 2003), a possible solution to obtain spontaneous speech in uni-modal manner is to apply filters to hide the verbal content (e.g. Low Pass filters block the high


components of a sound and pass the low-frequency components hiding linguistic cues; Scherer & Oshinsky 1977; Van Bezooijen & Gooskens 1999 Knoll & Costall 2009) or to monotonize the prosodic contour (e.g. Pitch Syn-chronous Overlap and Add works by dividing the speech waveform in small overlapping segments; the pitch of a signal is changed by moving the seg-ments further apart to decrease the pitch or closer together to increase the pitch, Moulines & Verhelst 1995; Van Bezooijen & Gooskens 1999).

This pilot study is a part of a broader research whose primary objec-tive is to analyze the relationship of prosody and semantics in conveying emotions. More specifically, the goal of the present work is to test the con-tribution of semantics and prosody, separately and in combination, to the identification of emotions. In order to pursue this aim, in the present in-vestigation natural speech samples were carefully selected from a French corpus of spontaneous speech, consisting of emotionally laden interviews. To separate the dif ferent channels and to create uni-modal stimuli, fil-ters approved by previous literature (see above) were applied. I attempted to investigate how the recognition of spontaneous emotional sentences changes with the application of filters, resulting in the formation of uni-modal stimuli – i.e., conveyed by the prosodic or the semantic channel only – to express emotions. To reach this goal, listeners were presented with an identification task in a forced-choice response format, while re-action times were collected. Moreover, I aimed to test whether perceptual dif ferences in intensity (i.e. how intense an emotion is felt) exist between multi-modal (not manipulated) and uni-modal emotional stimuli. In line with the literature promoting an enhanced recognition in a multi-modal condition (Kotz & Paulmann 2007), I expected to find a clear trend showing higher recognition rates for stimuli in which prosody and semantics are preserved. In light of the data which show that emotions are often explic-itly categorized more easily from the semantic content of an utterance or from facial expressions (Borod et al., 2000; Johnstone & Scherer 2000; Pell 2002; Paulmann & Kotz 2008), I hypothesized to find lowest recognition rates when only prosodic cues were present. Concerning intensity, I ex-pected that the integral stimuli (i.e. non-manipulated) would be evaluated as more intense.

2. MethodsThis study consisted of two phases: a pre-experiment to annotate and validate the stimuli, followed by the main experiment, which tested the contribution of semantics and prosody to emotion identification in spontaneous speech.

112 L aMiCuS 2019 no. 3

2.1. Norming study

The annotation and the selection were based on the results of an identification test in which listeners had to indicate, in a forced choice task, the emotion expressed by positive and negative sentences, which were emotionally laden.

2.1.1. Material

The corpus consisted of 48 sentences with a duration between 2 and 5 seconds (5.2 syllables per second) taken from two different corpora: 30 sentences were extracted from EmoTV (Abrilian et al. 2005), while the other 18 were selected from the Interactional Data Corpus (CID, Bertrand et al. 2008). EmoTV is a French corpus of 100 videos taken from television interviews on emotionally intense topics; while the CID is composed of 8 hours of semi-spontaneous dialogues recorded in a soundproof room at the Laboratoire Parole et Langage (LPL-CNRS, Aix en Provence, France) in 8 dialogues in which conversations revolve around two pre-established central themes, namely the narration of unusual events and conf lict situations.

Firstly, the 30 extracts of 2 to 5 seconds produced only by female speak-ers (18 different interviewees) were cropped from 20 videos of EmoTV with discrete acoustic quality and converted into audio files using Praat (Boersma 2001). Following the same criteria, the remaining 18 stimuli were selected from the CID (5 female speakers). Given that in the spontaneous production of the emotional discourse, the expression of emotional states is rarely con-veyed through a single channel of communication (Planalp 1999), the verbal content of the selected sentences was variable and emotionally laden (e.g. EmoTV: j’étais profondément en colère avec elle d’avoir menti ‘I was deeply angry with her for lying’, CID: il a pas du tout eu le temps pour moi ‘he had no time for me at all’). Totally, 48 short sentences conveying emotions both by prosody and semantics were collected.

2.1.2 Procedure

The perceptual validation of the stimuli was carried out in two steps. Firstly, two French native speakers (PhD students of linguistics, aged be-tween 25 and 30) listened to the 48 stimuli and classified the sounds fol-lowing the valence as criterion (namely the positivity or negativity of an emotion, Barrett 1998). They were instructed to classify the stimuli consid-ering both their prosody and semantics. They found 24 utterances express-ing positive emotions and 24 expressing negative ones. Secondly, eighteen French native speakers (12 F and 6 M), aged between 20 and 35, were re-cruited for a perception task. The experiment took place in a silent room at the Laboratoire Parole et Langage (LPL-CNRS, Aix en Provence). Listeners were instructed to sit comfortably and wore professional headphones. A


short practice session preceded the experiment consisting of two tasks. Firstly, they were asked to identify the emotion conveyed by the stimuli by pressing one of the six buttons on the keyboard, each corresponding to one of six basic emotions (i.e. joy, anger, fear, sadness, neutral state, surprise) respec-tively. The name of the emotions appeared on the computer screen. Once they identified the emotion, the listeners were asked to assess how well the chosen emotion was represented by the sounds on a 5-point Likert scale. The range of values was between 1 and 5 (with 1 corresponding to “not well represented” and 5 to “very well represented”). The stimuli were reproduced by a laptop using PERCEVAL (André & Ghio 2003). Each stimulus was heard twice and presented in 3 blocks; the presentation of the stimuli was randomized.

2.1.3. Results

Statistical analyses were performed on 864 observations (18 listeners * 48 stimuli) on the R software (version 3.5.1). To test the agreement between lis-teners in the categorization of stimuli, Kappa of Fleiss (1971) was applied (k = 0.0293, z = 3.21, p <0.0001), showing a limited agreement among the listeners. Also, the internal agreement was equally poor for both corpora (CID: k = 0.11, z = 8.84, p < 0.01; EmoTV: k = 0.0516, z= 4.27, p = 0.0000199). Concerning repre-sentativity, a repeated measure ANOVA with emotions (5) as within-subjects factor showed that utterances expressing joy and sadness were evaluated as more representative (F (5.4) = 5.32, p < 0.0001). Moreover, a Student t-test was performed to verify if one of the two corpora (EmoTV/CID) was judged to be more significant. No significant differences emerged (t = 2.55, df = 862, p = 0.798). However, EmoTV collected a tiny advantage in the mean value of representativity (see Figure 1).

Since for both corpora the agreement was scarce and there were no signifi-cant differences between them, I decided to select the corpus presenting the more “spontaneous” sentences. Utterances were taken from EmoTV because they were spontaneously spoken in a natural setting, while sentences from the CID were task-orientated and recorded in a laboratory. This choice is also

Figure 1. Box-plot of represen-tativity value (y-axis) for the factor corpus (i.e. EmoTV, CID; x-axis)

114 L aMiCuS 2019 no. 3

supported by the evidence of the small advantage in representativity scores for EmoTV. Indeed, I selected from EmoTV stimuli judged in the same way by 80% to 100% of the listeners and obtained high scores in terms of repre-sentativity (see Table 1).

Verbal content Emotion Agreement Representa-tivity score

Forte odeur /âcre/ âcre/ je pus plus re-spirer dans ma maison / c’était infernal‘Strong smell / acrid / acrid / I could not breathe in my house / it was hellish’

Colère ‘Anger’ 89% 3.9

J'étais profondément en colère avec elle d'avoir menti‘I was deeply angry with her for lying’


Ça fait quand même plus de deux mois /c’est très longue / très longue /les gens en ont peut-être marre‘It’s still over two months / it’s very long / very long / people may be fed up’


Plus les enfants grandissent/ s’ils sont pas suivi /c’est/ c’est très idiot‘Children grow up more and more/ if they are not followed / it’s / it’s very silly’

Tristesse‘Sadness’ 94% 3.8

Il ne dit pas du tout la vérité‘He does not tell the truth at all’


Pour une maman/ c’est / il n’y a pas pire de voir ses enfants souffrir de cette façon-là‘For a mom / there’s/ nothing is worse than to see her children suffer in this way’


On essaie d’avoir une ville jolie /agréable / accueillante‘We try to have a nice city / pleasant / welcoming’

Neutre‘Neuter’ 89% 3.9

Pour le tourisme d'Andorre, Pas de Calais, et voyez ce qu'on peut avoir‘For tourism in Andorra, Pas de Calais, and see what we can have’

NeutreNeuter 83% 3.8

Bon / ça se passe bien/ ça fait un petit salut/ c'est intéressant‘Well/ it’s okay /it was a short greeting/ it’s interesting’

Joie‘Joy’ 94% 4.5

Là aujourd’hui / tout me plait‘There, today / I like everything’

Joie‘Joy’ 100% 4.5

Et le matin/ quand on se lève/ on se dit/ la vie est belleAnd in the morning/ when we get up/ we tell ourselves/ life is beautiful

Joie‘Joy’ 94% 3.7

Anglement /c'était une famille très apa-thique / forcément ça choqueAnglement /it was a very apathetic family /this is necessarily surprising

Surprise‘Surprise’ 94% 4.1

Table 1. Verbal content, agree-ment percentage and represen-tativity score (mean) of the sentences chosen for the main experiment


To sum up, for the main experiment, 12 stimuli (pronounced by 10 female speakers) were selected from EmoTV. They consisted in short sentences with a duration from 2 to 5 seconds. They always contained a congruent set of cues to express one of the following emotions: 3 stimuli expressed joy, 3 sadness, 3 anger, 1 surprise, 2 neutral state. Utterances conveying surprise were inserted as fillers to balance the number of positive and negative emotions. Sentences judged as neutral were used as control states.

2.2. Main experiment

2.2.1. Corpus

The 12 speech fragments selected from the norming study were presented to the listeners in three different versions:1. Nonverbal. By lowpass filtering the signal at 350 Hz (standard value) the speech was rendered unintelligible. In this condition, the only cue provided to the listeners is prosodic information. 2. Verbal. By means of electronic monotonization (at a fixed value of 220 Hz, i.e., the mean pitch over all listeners) the intonation was removed from the signal. Monotonization was affected through Pitch Synchronous Overlap and Add (PSOLA) analysis and resynthesis (Moulines & Verhelst 1995). Verbal in-formation is maintained, and so are temporal and loudness variation. The fragments are completely intelligible, but perfectly monotonous. 3. Integral. In this version all prosodic and verbal information is present.As a result, the corpus included 34 stimuli, 24 of which were uni-modal (i.e. 12 preserving only verbal content, 12 presenting only prosody), expressing 4 emotions (i.e. joy, anger, sadness, surprise) and neutral state.

2.2.2. Procedure

Eighteen French native speakers (10 F and 8 M), aged 20 to 35, were selected for the test. The experiment took place in a silent room at the University of Naples “L’Orientale”. The listeners were volunteers, recruited through social networking sites. They were asked to sit in a comfortable chair and wore professional headphones. Concerning the task, they were firstly asked to identify the emotion conveyed by the stimuli by pressing one of the six but-tons on the keyboard corresponding to the list of five basic emotions (i.e. joy, anger, sadness, neutral, surprise) appearing on the computer screen. Secondly, after identifying the emotions, listeners were asked to rate the intensity of the emotion expressed by the sounds on a 5-point Likert scale. The range of the values was from 1 to 5 (with 1 meaning “not intense emotion” and 5 “very intense emotion”). Stimuli were played on a laptop by PERCEVAL software

116 L aMiCuS 2019 no. 3

(André & Ghio 2003). The fragments were presented in separate blocks of in-tegral, verbal, and nonverbal fragments; in two sequences and in two random counterbalanced orders. Prior to starting the experiment, instructions were displayed on the screen and, in order to get familiarized with the experi-mental procedure, listeners ran three practice trials.

3. Results

3.1. Correct identification

Statistical analyses were performed on 648 observations (18 listeners * 36 sen-tences). A logistic regression (logit model) was used to describe the relationship between the correct identification and the different kind of emotions and cues. This model is particularly appropriate for categorization tasks because it is specifically designed for analyzing binary (e.g. yes/no) and categorical response variables (Hosmer et al. 1989). In this model, the dummy variable was correct/incorrect, while the factors emotions (5 levels: neutral state, joy, anger, sadness, surprise), type (3 levels: integral, verbal, nonverbal) and intensity (5 points Likert scale) were the predictors, allowing for interaction effects. Final model according to AIC stepwise variable selection is reported in Table 2, describing the coefficients associated to each predictor variable (first column).

Β Odds Ratio Std. Error z value Pr(>|z|)

(Intercept) 0.4243 1.5285 0.3843 1.104 0.269649

Colère ‘Anger’ -0.5816 0.5590 0.2694 -2.159 0.030882

Joie ‘Joy’ -0.9623 0.3820 0.2979 -3.231 0.001236

Surprise ‘Surprise’ 0.7069 2.0278 0.3624 1.951 0.051109

Tristesse ‘Sadness’ -0.9996 0.3680 0.2720 -3.676 0.000237

Nonverbal 1.6170 5.0380 0.2368 6.829 0.0001

Verbal 0.3480 1.4163 0.2280 1.526 0.126892

Intensity -0.2798 0.7559 0.0824 -3.396 0.000684

In line with the hypotheses and the previous literature, as shown by the bar plot (Figure 1), the emotions were better identified in the integral version (M: 75.2 %) followed by verbal type (M: 66.2 %). The lowest recognition rates are found when only prosodic cues were present in utterances (M: 36.8 %). Sig-nificant effects are found for nonverbal stimuli expressing sadness (p<0.0001) neutral state (p<0.0001), anger (p<0.0001), surprise (p<0.0001). For these stim-

Table 2. Coef f i-cients of LOGIT model estimated for the factor emotions (an-ger, joy, sur-prise, sadness), type (nonver-bal, verbal), intensity


uli the recognition rates are lower (sadness: 49 %; anger: 38%; neutral: 17%, surprise: 14%).

3.2. Intensity

Concerning intensity, a repeated measures ANOVA revealed a significant main effect for the emotion (F (5.37) = 38.06, p<0.0001) and type (F (9.74) = 37.61, p<0.0001) factors. As shown in Figures 3-4, the more intensely judged stimuli were globally better recognized (t-test: t=6.490, df=645, p<0.0001). More im-portantly, the integral stimuli were considered the most intense (M=3.725581, SD=1.043022), followed by the verbal (M=3.50463, SD=1.153348), while the nonverbal stimuli (M=2.916667, SD=1.138134) were judged the least intense (p<0.0001) confirming the predictions made about intensity.

3.3. Reaction time

Before setting up the statistical analyses, because of the positive asymmetry of data (skewness=3.899778), a log transformation was applied in order to normalize them (skewness of logarithm of reaction time =0.04336222). A re-peated measures ANOVA with type as the within-factor (three different levels: i.e. integral, verbal, nonverbal) was performed on reaction time (log trans-

Figure 2. Lis-tener’s correct identif ication percentage (y-axis) for the f ive emotions (x-axis), split by type (i.e. inte-gral, nonver-bal, verbal)

Figure 3. Boxplot of intensity value (x-axis) for the factor correct/incorrect (y-axis)

Figure 4. Boxplot of intensity value (x-axis) for the factor type (i.e. integral, non-verbal, verbal; y-axis)

118 L aMiCuS 2019 no. 3

formed). Analyses revealed significant main effect (F (7.4) =38.69, p<0.0001) with important differences between integral-nonverbal stimuli (p<0.0001) and integral-verbal (p<0.0001); while a not-so-significant effect is found by comparing verbal-nonverbal (p=0.327). Boxplot (Figure 5) shows how listeners identified integral stimuli faster than the other ones. The reaction time was slowest for stimuli presenting only prosodic cues.

4. Discussion The aim of the study was to investigate the contribution of semantic and pro-sodic cues separately and when they are both present, in the identification of emotions in spontaneous speech. In particular, the goal was to test whether emotion recognition is facilitated by multi-modal versus uni-modal stimuli. This was attempted by filtering utterances from spontaneous speech. In ad-dition, I explored whether the perception of an emotion’s intensity changed depending on the channel of transmission. In contrast to most of the meth-odological approaches adopted by previous studies, the issue was investi-gated by using spontaneous congruent stimuli, which always conveyed the same emotions as represented by both prosody and semantics. In line with Paulmann and Pell (2011), my results showed that, in spontaneous speech, the recognition of emotions increases in accuracy when multiple channels cooperate to convey them in a congruent manner, i.e. integral stimuli were recognized significantly better than uni-modal stimuli. This aligns with pre-vious studies which show that emotional judgments tend to improve when more than one source of congruent information about the intended emo-tion is available (Collignon et al. 2008; Pell 2005; De Gelder & Vroomen 2000; Massaro & Egan 1996). This assumption is also confirmed by the reaction times, showing that the access to emotions is faster when they are conveyed by both prosodic and semantic cues. From a cognitive point of view, since information in each emotional channel activates shared conceptual knowl-edge about an emotion (emotion concepts, cf. Bower 1981; Russell & Lemay 2000; Halberstadt et al. 1995), it is possible that, during emotion identifica-tion tasks, knowledge about emotions is more readily accessible when more congruent cues cooperate to convey the emotional states (Paulmann & Pell 2011). In line with this hypothesis, Borod et al. (2000) has suggested two steps

Figure 5. Boxplot of log reaction time data (x-axis) for the factor type (i.e. integral, nonverbal, ver-bal; y-axis)


in the processing of emotions: first, emotional channels are independently processed by separate sensory modality systems and, after that, they pass through a general af fective processor. Following this, it could be assumed that prosodic and verbal cues are incorporated during emotional recognition processes, leading to systematically higher accuracy rates, as observed here. Interestingly, confirming results of the study of Paulmann and Pell (2011), the described process seems to also involve the recognition of spontaneous neu-tral states that were better identified when presented in the integral version with both prosodic and semantic cues. This result supports those received by Cornew et al. (2009). In two gating experiments, they showed that par-ticipants identified neutral prosody more rapidly and accurately than happy or angry prosody, suggesting an actual prominent advantage for processing neutral over emotional content.

Concerning intensity, in line with the literature (Wells et al. 2016), the results illustrated that the integral stimuli were perceived as more intense, meaning that the interaction between the prosodic and semantic channel could co-occur to strengthen the perception of emotional intensity. Not sur-prisingly, the stimuli judged as more intense were identified better, signify-ing that perceived emotional intensity might be a strong signal for discrim-ination when usable (Audibert et al. 2008).

Moreover, this study showed that spontaneous emotions conveyed by semantics were recognized more accurately and faster than through pro-sodic information. A possible reason may be that there are some differences in how prosody and semantics trigger emotion nodes during the expression of emotion (Bower 1981; Paulmann & Pell 2011; Halberstadt et al. 1995; Russell & Lemay 2000). Previous studies have argued that emotions are conveyed through prosody in a categorical manner over a protracted time period (Juslin & Laukka 2003; Rigoulot et al. 2013). In contrast, the semantic content of stimuli ref lected, in a prototypical manner, the selected emotions, therefore activating the underlying emotional conceptual knowledge more strongly and faster (Paulmann & Pell 2011). Furthermore, the low recognition rates collected for stimuli presenting only prosodic parameters (M: 36%) could be explained in light of recent research on the prosody of authentic emotions. In fact, the acoustic analyses on spontaneous speech presented weaker acoustic differences between the emotions for the authentic expressions compared to studies on portrayal (Laukka et al. 2012), compromising the capacity of listeners to detect emotions in absence of other speech signals and the con-text. Future studies are necessary to verify whether this semantic dominance is culture/language-dependent, as suggested by studies on Japanese and American English listeners (Ishii et al. 2003; Kitayama & Ishii 2002). Another important issue is the gender variation in the perception of emotional pros-ody and semantics. Schirmer and Kotz (2003) in an ERP study have shown a

120 L aMiCuS 2019 no. 3

better integration of emotional cues for women than men. This study, as a pilot, did not address this matter but it will be investigated in the wider study by balancing the number of male and female participants and by presenting to the listeners stimuli pronounced both by men and women.

In conclusion, this study showed that the concept of multimodal ad-vantage could be extended also for the recognition of spontaneous emotions. Although the identification of emotions could be possible by separate chan-nels of transmission, congruent and simultaneous information conveyed by prosodic and semantic patterns at the same time increase the accuracy and speed in perceiving emotions. As underlined, this could confirm the adequacy of applying a cognitive model of emotional processing in which verbal and nonverbal cues to emotions are integrated.

AcknowledgmentsThis study was set up with a PhD grant of the author from University of Naples “L’Orientale” (Italy) in a context of a Doctoral Thesis whose Director is Prof. Alberto Manco. The Perceptual test (norming) was conducted at the Laboratoire Parole et Langage (LPL) in Aix en Provence (France). Thanks to Ca-terina Petrone for helping with the stimuli selection and Francesco Santelli for some comments on statistical analyses.

References Abrilian, Sarkis, Jean-Claude Martin, Laurence Devillers 2005: A cor-

pus-based approach for the modeling of multimodal emotional be-haviors for the specification of embodied agents. Proceeding from HCI International Las Vegas, USA.

André, Carine, Alain Ghio 2003: PERCEVAL: une station automatisée de tests de PERCeption et d’EVALuation auditive et visuelle. Travaux Interdis-ciplinaires du Laboratoire parole et langage d’Aix-en-Provence 2003 (TIPA) 22, 115-133.

Audibert, Nicolas, Véronique Aubergé, Albert Rilliard 2008: Acted vs. sponta-neous expressive speech: perception with inter-individual variability. Pro-gramme of the Workshop on Corpora for Research on Emotion and Affect, 23-26.

Banse Rainer, Scherer Klaus 1996: Acoustic profiles in vocal emotion expres-sion, Journal of Personality and Social Psychology 70(3), 614-636.

Bänziger, Tanja, Klaus R. Scherer 2010: Introducing the geneva multimodal emotion portrayal (gemep) corpus. Blueprint for Af fective Computing: A Sourcebook, 271-294.

Barrett, Feldman Lisa 1998: Discrete emotions or dimensions? The role of valence focus and arousal focus. Cognition & Emotion 12, 579-599.


Bertrand, Roxane, Philippe Blache, Robert Espesser, Ferré, Christine Meu-nier, Béatrice Priego-Valverde, Stéphane Rauzy 2008: Le CID-Corpus of Interactional Data-Annotation et exploitation multimodale de parole conversationnelle. Traitement Automatique des Langues 49(3), 105-134.

Boersma, Paul 2001: Praat, a system for doing phonetics by computer. Glot International 5(9/10), 341-345.

Borod, Joan C., Lawrence H. Pick, Susan Hall, Martin Sliwinski, Nancy Ma-digan, Loraine K. Obler, Joan Welkowitz, Elisabeth Canino, Hulya M. Erhan, Mira Goral, Chris Morrison, Matthias Tabert 2000: Relationships among facial, prosodic, and lexical channels of emotional perceptual processing. Cognition & Emotion 14, 193-211.

Bower, Gordon H. 1981: Mood and memory. American Psychologist 36, 2, 129-148.Bradley, Margaret M., Peter J. Lang, 1999: Affective norms for English words

(ANEW): Instruction manual and affective ratings. Technical Report C-1, the Center for Research in Psychophysiology, 30(1), 25-36.

Collignon, Oliver, Frédéric Gosselin, Sanjoy Roy, Dave Saint-Amour, Maryse Lassonde, Franco Lepore 2008: Audio-visual integration of emotion expression. Brain Research 1242, 126-135.

Cornew, Lauren, Leslie Carver, Tracy Love 2009: There’s more to emotion than meets the eye: A processing bias for neutral content in the domain of emotional prosody. Cognition & Emotion 24(7), 1133-1152.

Cowie, Roddy, Randolph Cornelius 2003: Describing the emotional states that are expressed in speech. Speech Communication 40(1/2), 5-32.

De Gelder, Beatrice, Jean Vroomen 2000: The perception of emotions by ear and by eye. Cognition & Emotion 14(3), 289-311.

De Gelder, Beatrice, Paul Bertelson 2003: Multisensory integration, percep-tion and ecological validity. Trends in Cognitive Sciences 7(10), 460-467.

Drolet, Matthis, Ricarda I. Schubotz, Julia Fischer 2012: Authenticity affects the recognition of emotions in speech: behavioral and fMRI evidence. Cognitive, Af fective, & Behavioral Neuroscience 12(1), 140-150.

Ekman, Paul 1992: An argument for basic emotions. Cognition & Emotion 6(3/4), 169-200.

Estes, Zachary, Michelle Verges 2008. Freeze or f lee? Negative stimuli elicit selective responding. Cognition 108(2), 557-565.

Filippi, Piera, Sebastian Ocklenburg, Daniel L Bowling, Long Heege, Onur Güntürkün, Newen Albert, Bart de Boer 2017: More than words (and faces): evidence for a Stroop effect of prosody in emotion word pro-cessing. Cognition & Emotion 31(5), 879-891.

Fleiss, Joseph L. 1971: Measuring nominal scale agreement among many raters. Psychological Bulletin 76(5), 378-382.

122 L aMiCuS 2019 no. 3

Frühholz, Sascha, Wiebke Trost, Sonja A Kotz. 2016: The sound of emotions—Towards a unifying neural network perspective of affective sound pro-cessing. Neuroscience & Biobehavioral Reviews 68, 96-110.

Halberstadt, Jamin B., Paula M. Niedenthal, Julia Kushner 1995: Resolution of lexical ambiguity by emotional state. Psychological Science 6(5), 278-282.

Hosmer, David W., Stanley Lemeshow, Rodney X. Sturdivant 1989: The mul-tiple logistic regression model. Applied Logistic Regression 1, 25-37.

Huang, Gao, Zhuang Liu, Laurens van der Maaten, Kilian Q. Weinberger, July 2017: Densely connected convolutional networks. Proceedings from the IEEE conference on computer vision and pattern recognition, 4700-4708.

Ishii, Keiko, Alberto Jose Reyes, Shinobu Kitayama 2003: Spontaneous atten-tion to word content versus emotional tone: Differences among three cultures. Psychological Science 14(1), 39-46.

Johnstone, Tom, Klaus R. Scherer 2000: Vocal communication of emotion. Handbook of Emotions 2, 220-235.

Jürgens, Rebecca, Matthis Drolet, Ralph Pirow, Elisabeth Scheiner, Julia Fis-cher 2013: Encoding conditions affect recognition of vocally expressed emotions across cultures. Frontiers in Psychology 4, 111.

Juslin, Patrik N., Petri Laukka, 2003: Communication of emotions in vocal expression and music performance: Different channels, same code? Psychological Bulletin 129(5), 770.

Kanske, Philipp, Sonja A. Kotz 2007: Concreteness in emotional words: ERP evidence from a hemifield study. Brain Research 1148, 138-148.

Kim, Yelin, Emily Mower Provost 2016: Emotion spotting: Discovering re-gions of evidence in audio-visual emotion expressions. Proceedings from the 18th ACM International Conference on Multimodal Interaction, October 2016, 92-99.

Kitayama, Shinobu, Keiko Ishii 2002: Word and voice: Spontaneous atten-tion to emotional utterances in two languages. Cognition & Emotion 16, 1, 29-59.

Knoll, Monja A., Maria Uther, Alan Costall 2009: Effects of low-pass filtering on the judgment of vocal affect in speech directed to infants, adults and foreigners. Speech Communication 51(3), 210-216.

Kotz, Sonja A., Silke Paulmann 2007: When emotional prosody and seman-tics dance cheek to cheek: ERP evidence, Brain Research 1151, 107-118.

Kousta, Stavroula-Thaleia, David P. Vinson, Gabriella Vigliocco 2009: Emo-tion words, regardless of polarity, have a processing advantage over neutral words. Cognition 112(3), 473-481.

Kreifelts, Benjamin, Thomas Ethofer, Wolfgang Grodd, Matthias Erb, Wild-gruber Dirk 2007: Audiovisual integration of emotional signals in voice and face: an event-related fMRI study. Neuroimage 37(4), 1445-1456.


Laukka, Petri, Nicolas Audibert, Véronique Aubergé 2012: Exploring the determinants of the graded structure of vocal emotion expressions. Cognition & Emotion 26(4), 710-719.

Massaro, Dominic W., Peter B. Egan 1996: Perceiving affect from the voice and the face. Psychonomic Bulletin & Review 3(2), 215-221.

Milesi, Valérie, Sezen Cekic, Julie Péron, Sascha Frühholz, Chiara Cristin-zio, Margitta Seeck, Didier Grandjean 2014: Multimodal emotion per-ception after anterior temporal lobectomy (ATL). Frontiers in Human Neuroscience 8, 275.

Moulines, Eric, Werner Verhelst 1995: Time-domain and frequency-domain techniques for prosodic modification of speech. In: Kleijn, Bastiaan, W., Kuldip K. Paliwal, (eds.) Speech Coding and Synthesis, Amsterdam: Elsevier, 519-555.

Murray, Iain R., John L. Arnott 1993: Toward the simulation of emotion in synthetic speech: A review of the literature on human vocal emotion. The Journal of the Acoustical Society of America 93(2), 1097-1108.

Nygaard, Lynne C., Erin R. Lunders 2002: Resolution of lexical ambiguity by emotional tone of voice. Memory & Cognition 30(4), 583-593.

Nygaard, Lynne C., Jennifer S. Queen 2008: Communicating emotion: linking affective prosody and word meaning. Journal of Experimental Psychology: Human Perception & Performance 34(4), 1017.

Paulmann, Silke, Sonja A. Kotz 2008: Early emotional prosody perception based on different speaker voices. Neuroreport 19(2), 209-213.

Paulmann Silke, Mark David Pell 2011. Is there an advantage for recognizing multi-modal emotional stimuli? Motivation & Emotion 35, 192-201.

Pell, Marc David 2002: Evaluation of nonverbal emotion in face and voice: Some preliminary findings on a new battery of tests. Brain & Cognition 48(2/3), 499-504.

Pell, Marc David 2005: Prosody–face interactions in emotional processing as revealed by the facial affect decision task. Journal of Nonverbal Behavior 29(4), 193-215.

Pell, Marc David, Laura Monetta, Sally Paulmann, Sonja A. Kotz 2009: Rec-ognizing emotions in a foreign language. Journal of Nonverbal Behavior 33(2), 107-120.

Planalp, Sally 1999: Communicating Emotion: Social, Moral, and Cultural Processes. Cambridge: Cambridge University Press.

Rigoulot, Simon, Eugen Wassiliwizky, Marc David Pell 2013: Feeling back-wards? How temporal order in speech affects the time course of vocal emotion recognition. Frontiers in Psychology 4, 367.

Russell, James A., Ghyslaine Lemay 2000: Emotion concepts. In: M. Lewis and J.M. Haviland-Jones (Eds.), Handbook of Emotions (2nd ed.). New York: Guilford Press.

124 L aMiCuS 2019 no. 3

Schacht, Annekathrin, Werner Sommer 2009: Emotions in word and face pro-cessing: early and late cortical responses. Brain & Cognition 69(3), 538-550.

Scherer, Klaus R. 2003: Vocal communication of emotion: A review of re-search paradigms. Speech Communication 40(1-2), 227-256.

Scherer, Klaus R., James S. Oshinsky 1977: Cue utilization in emotion attri-bution from auditory stimuli. Motivation & Emotion 1(4), 331-346.

Scherer, Klaus R., Rainer Banse, Harald G. Wallbott, Thomas Goldbeck 1991: Vocal cues in emotion encoding and decoding. Motivation & Emotion 15(2), 123-148.

Schirmer, Annett, Sonja A. Kotz 2003: ERP Evidence for a sex-specific Stroop effect in emotional speech. Journal of Cognitive Neuroscience 15(8), 1135-1148.

Scott, Graham G., Patrick J. O’Donnell, Hartmut Leuthold, Sara C. Sereno 2009: Early emotion word processing: evidence from event-related po-tentials. Biological Psychology 80, 95-104.

Van Bezooijen, Renée, Charlotte Gooskens 1999: Identification of language varieties: The contribution of different linguistic levels. Journal of Lan-guage and Social Psychology 18(1), 31-48.

Wells, Laura Jean, Steven Mark Gillespie, Pia Rotshtein 2016: Identification of emotional facial expressions: effects of expression, intensity, and sex on eye gaze. PloS One 11, 12, e0168307.

Streszczenie

Rozpoznawanie emocji na podstawie prozodii i semantyki w języku francuskim w mowie spontanicznej: Badanie pilotażowe

Celem tego artykułu jest zbadanie wpływu sygnałów prozodycznych i se-mantycznych – osobno oraz współwystępujących ze sobą – na rozpoznawa-nia emocji w mowie spontanicznej. Istotą eksperymentu było zbadanie, czy rozpoznawanie emocji jest łatwiejsze w przypadku bodźców monomodal-nych czy multimodalnych. Aby to osiągnąć, zastosowano filtrowanie wypo-wiedzi w mowie spontanicznej. Ponadto zbadano również, czy rozpozna-wanie intensywności emocji zmieniało się w zależności od kanału przekazu. W przeciwieństwie do metodologii używanych w większości wcześniej-szych badań, zastosowano bodźce zgodne, które zawsze przekazywały tę samą emocję w obu modalnościach: prozodii i semantyce. Tak jak w badaniu Paulmann i Pella (2011), wyniki obecnego badania wskazują, że dokładność w rozpoznawaniu emocji w mowie spontanicznej wzrasta w reakcji na bo-


dziec zgodny. To znaczy, że bodźce zintegrowane były rozpoznawane zna-cząco lepiej niż bodziec jednomodalny. Wyniki te potwierdzają wcześniejsze badania, które wskazują, że rozpoznanie emocji wzrasta, jeśli dostępne jest więcej niż jedno źródło informacji zgodnych na temat danej emocji (Col-lignon et al. 2008; Pell 2005; De Gelder & Vroomen 2000; Massaro & Egan 1996). Czasy reakcji również wskazują, że dostęp do emocji jest szybszy, kiedy przekazywane są one przy pomocy obu modalności: semantycznej i prozo-dycznej. Z kognitywnego punktu widzenia, w związku z tym, że informacja w każdym z kanałów komunikacji emocji aktywuje wspólną wiedzę o danej emocji (por. Bower 1981; Russell & Lemay 2000; Halberstadt et al. 1995), uza-sadnione wydaje się twierdzenie, że w czasie zadań polegających na rozpo-znawaniu emocji wiedza na temat danej emocji jest łatwiej dostępna, jeśli więcej bodźców zgodnych współdziała w przekazywaniu stanów emocjo-nalnych (Paulmann & Pell 2011). W zgodzie z tą hipotezą, Borod et al. (2000) zasugerowali dwa etapy w przetwarzaniu emocji. Najpierw kanały emocjo-nalne są przetwarzane niezależnie, poprzez oddzielne systemy sensoryczne, a następnie przechodzą przez ogólny procesor afektywny (general af fective procesor). Z tego wynika, że bodźce prozodyczne i werbalne są integrowane w procesie rozpoznawania emocji, co prowadzi do systematycznie coraz lepszej dokładności, jak wykazano w obecnym badaniu. Co ciekawe, zgodnie z wynikami Paulmann i Pella (2011), opisany proces dotyczy również roz-poznawania spontanicznych stanów neutralnych, które lepiej odczytywano w obecności bodźców zintegrowanych: prozodycznych i semantycznych. Potwierdzają to badania Cornew et al. (2009). W dwóch eksperymentach wy-korzystujących zadania stopniowego odkrywania/udostępniania wyrazów (gating experiment)pokazali oni, że uczestnicy eksperymentu rozpoznawali prozodię neutralną szybciej i dokładniej niż prozodię wyrażającą złość lub radość, co wskazuje na znaczącą przewagę przetwarzania bodźców neu-tralnych nad emocjonalnymi.

Jeśli chodzi o intensywność emocji, jak pokazują wyniki Wells et al. (2016), bodźce zintegrowane odczytywano jako bardziej intensywne, co wskazuje, że interakcja pomiędzy kanałem prozodycznym i semantycznym wzmacnia rozpoznawanie intensywności emocjonalnej. Nie dziwi fakt, że bodźce oceniane jako intensywniejsze były lepiej rozpoznawane, co oznacza, że postrzegana intensywność emocji może stanowić silny sygnał do rozróż-niania emocji, o ile jest dostępna (Audibert et al. 2008).

Co więcej, obecne badanie pokazało, że emocje spontaniczne przekazy-wane w modalności semantycznej rozpoznawano dokładniej i szybciej niż te, które przekazywano w modalności prozodycznej. Zjawisko to można wyja-śnić na gruncie różnic tym, jak bodźce prozodyczne i semantyczne pobudzają węzły emocjonalne (emotion nodes) w trakcie wyrażania emocji (Bower 1981; Paulmann & Pell 2011; Halberstadt et al. 1995; Russell & Lemay 2000). Wcze-

126 L aMiCuS 2019 no. 3

śniejsze badania pokazują, że emocje wyrażone przez prozodię mają cha-rakter kategoryczny w przedłużonym okresie. Natomiast emocje wyrażone semantycznie mają charakter prototypowy, co sprawia, że wybrane emocje aktywują wiedzę na temat pojęcia danej emocji mocniej i szybciej (Paulmann & Pell 2011). Niskie wskaźniki rozpoznania w reakcji na bodźce prozodyczne (M: 36%) można wyjaśnić w świetle najnowszych badań dotyczących pro-zodii emocji wyrażanych spontanicznie. Analizy akustyczne wskazały na mniejsze różnice akustyczne pomiędzy emocjami w mowie spontanicznej niż w przypadku emocji wyrażanych aktorsko (Laukka et al. 2012), co utrudniło słuchaczom rozpoznanie emocji pod nieobecność innych elementów mowy i kontekstu. Konieczne są dalsze badania, które sprawdziłyby, czy ta domi-nacja modalności semantycznej jest charakterystyczna dla danego języka/kultury, na co wskazują badania na Japończykach i Amerykanach (Ishii et al. 2003; Kitayama & Ishii 2002). Inną ważną zmienną w percepcji prozodii i semantyki emocji jest płeć. Schirmer i Kotz (2003) w badaniu ERP pokazali lepszą integrację bodźców emocjonalnych u kobiet niż u mężczyzn. Obecne badanie pilotażowe pominęło ten aspekt, ale w badaniu głównym liczba ko-biet i mężczyzn zostanie wyrównana zarówno w grupie eksperymentalnej, jak i w nagraniach bodźców.

Podsumowując, obecne badanie pokazuje, że przewaga multimodalna jest również zauważalna w rozpoznawaniu spontanicznych emocji. Chociaż rozpoznawanie emocji na podstawie tylko jednej modalności jest możliwe, bodźce zgodne i równoczesne przekazane zarówno w warstwie prozodycznej, jak i semantycznej prowadzą do wzrostu dokładności i tempa rozpoznawa-nia emocji. Te wyniki można traktować jako potwierdzenie kognitywnego modelu przetwarzania emocji, w którym bodźce werbalne i niewerbalne są integrowane.

Tłum. Małgorzata Fabiszak

BibliografiaAudibert, Nicolas, Véronique Aubergé, Albert Rilliard 2008: Acted vs. spon-

taneous expressive speech: perception with inter-individual variabil-ity. Programme of the Workshop on Corpora for Research on Emotion & Af fect, 23-26.

Borod, Joan C., Lawrence H. Pick, Susan Hall, Martin Sliwinski, Nancy Ma-digan, Loraine K. Obler, Joan Welkowitz, Elisabeth Canino, Hulya M. Erhan, Mira Goral, Chris Morrison, Matthias Tabert 2000: Relationships among facial, prosodic, and lexical channels of emotional perceptual processing. Cognition and Emotion 14, 193–211.


Bower, Gordon H. 1981: Mood and memory. American Psychologist 36(2), 129-148.

Collignon, Oliver, Frédéric Gosselin, Sanjoy Roy, Dave Saint-Amour, Maryse Lassonde, Franco Lepore 2008: Audio-visual integration of emotion expression. Brain Research 1242, 126-135.

Cornew, Lauren, Leslie Carver, Tracy Love 2009: There’s more to emotion than meets the eye: A processing bias for neutral content in the domain of emotional prosody. Cognition & Emotion 24(7), 1133–1152.

De Gelder, Beatrice, Jean Vroomen 2000: The perception of emotions by ear and by eye. Cognition & Emotion 14(3), 289-311.

Halberstadt, Jamin B., Paula M. Niedenthal, Julia Kushner 1995: Resolution of lexical ambiguity by emotional state. Psychological Science 6(5), 278-282.

Ishii, Keiko, Alberto Jose Reyes, Shinobu Kitayama 2003: Spontaneous atten-tion to word content versus emotional tone: Differences among three cultures. Psychological Science 14(1), 39-46.

Kitayama, Shinobu, Keiko Ishii 2002: Word and voice: Spontaneous atten-tion to emotional utterances in two languages. Cognition & Emotion 16(1), 29-59.

Laukka, Petri, Nicolas Audibert, Véronique Aubergé 2012: Exploring the determinants of the graded structure of vocal emotion expressions. Cognition & Emotion 26(4), 710-719.

Massaro, Dominic W., Peter B. Egan 1996: Perceiving affect from the voice and the face. Psychonomic Bulletin & Review 3(2), 215-221.

Paulmann Silke, Mark David Pell 2011. Is there an advantage for recognizing multi-modal emotional stimuli? Motivation & Emotion 35, 192-201.

Pell, Marc David 2005: Prosody–face interactions in emotional processing as revealed by the facial affect decision task. Journal of Nonverbal Behavior 29(4), 193-215.

Russell, James A., Ghyslaine Lemay 2000: Emotion concepts. W: M. Lewis and J.M. Haviland-Jones (Red.), Handbook of emotions (Wyd. 2). New York: Guilford Press.

Schirmer, Annett, Sonja A. Kotz 2003: ERP Evidence for a sex-specific Stroop effect in emotional speech. Journal of Cognitive Neuroscience 15(8), 1135–1148.

Wells, Laura Jean, Steven Mark Gillespie, Pia Rotshtein 2016: Identification of emotional facial expressions: effects of expression, intensity, and sex on eye gaze. PloS One 11, 12, e0168307.

128 L aMiCuS 2019 no. 3

Małgorzata Fabiszak1

Karolina Krawczak2

1,2Adam Mickiewicz University in Poznań

The status of multimodal constructions in developing models and empirical testing- introduction to the debate

Language can be viewed as a system of signs shared between community members, a communicative behaviour, a mental phenomenon, “hard-wired” in the brain. But what does this mean for our understanding of language, of methods of its investigation, and modelling?

For de Saussure (1916), language is a formal system of signs consisting of the signifier (roughly – the form) and the signified (i.e., the meaning). It is one of a variety of sign systems used for communication and is a part of social psychology. Its description rests on identifying how signs differ from each other in terms of their form and meaning. In this sense, the study of language is self-contained, being based on an introspective analysis of the observed products of communication. It does not need to cross disciplines, a view which today may well seem rather obsolete.

Cognitive Linguistics, at its beginnings (Langacker 1987, 1991) also de-fined the linguistic sign as a bi-polar symbol with the phonological pole (i.e., the form) and the semantic pole (i.e., the meaning). Similarly, in Construction Grammar (Fillmore 1988, Goldberg 1995), the unit of grammar is a construc-tion understood as a form and function pairing. On this account, the formal pole includes both syntactic and phonological information (such as prosody and intonation), while the function pole encodes semantic and pragmatic

129The status of multimodal constructions...

content. Constructions thus understood form a taxonomic network, which constitutes a language grammar and which is entirely usage-based (Hopper 1987, Barlow & Kemmer 1999, Bybee 2006), i.e. it emerges from language use. The emergent mental constructions may have different levels of schematiza-tion and can be dynamically transformed, as novel examples are encountered by language users. This inherent dynamicity and resultant variation means that grammar should be viewed as probabilistic, rather than rule-based. What follows is that it should be described in terms of usage patterns re-trievable through generalization from actual language use, as attested in corpora (e.g., Stefanowitsch & Gries 2006, Gries & Stefanowitsch 2006, Glynn & Fischer 2010, Glynn & Robinson 2014).

Beyond corpus-based investigations into the structure of language, Cognitive Linguistics also makes a strong claim about the psychological plausibility of the linguistic models, seeing psycholinguistic experiments as an important part of testing hypotheses about language (Lakoff 2008). Recently, the language/emotion interface has generated a lot of experimen-tal research, underlying linguistic-semantic approach to the exploration of emotional content encoded, and communicated via language. Numerous studies show that emotional content, verbal and nonverbal, exerts strong impact on the perception and processing of linguistic content – facilitating linguistic content processing at lexical and sentence level (Kißler et al. 2007, 2013). This new paradigm brings up a range of questions concerning the language/emotion interface – do linguistic and emotional contents belong to different systems of representation of meaning? Should we come up with new semantics to incorporate emotional, experientially acquired meanings into language systems? Do we acquire emotional meanings and linguistic meanings separately?

In recent years, the multimodal character of language has received much attention from cognitive linguists (e.g., Cienki & Muller 2008, Cienki 2016). Accordingly, rather than being seen as a purely verbal system with prosodic features, language is also claimed to incorporate visual clues or indices, such as body posture, gestures or facial expressions. In that under-standing, co-speech gestures are of utmost importance for interpreting the meaning of a communicative act. But what exactly constitutes a gesture? Are gestures part of constructions? Is there a grammar of gestures? And how should gestures be represented in the multimodal model of language? How does sign language expresses itself multimodally?

Finally, multimodality is not only a feature of spoken communication. Written language and other non-spoken communicative modes such as in-ternet discourse, adverts, cartoons or comics can also be described as mul-timodal, so that both images and the (verbal) text contribute to the meaning making process, and even the shape of the fonts may inf luence our evaluation

130 L aMiCuS 2019 no. 3

of the message (e.g., Kress and van Leuveen 2001, Dancygier & Sweetser 2012, Dancygier & Vandelanotte 2017, Vandelanotte & Dancygier 2017).

This brings us to a series of questions we would like to ask of our debaters:1. What does this mean for our understanding, investigating and modelling

of language and communication? 2. Are multimodal interactions or texts understood by language users

through reference to mental representations? If so, are these represen-tations mono- or multimodal? (Representations need not be understood as stable patterns but as dynamic co-activation of neuronal assemblies.)

3. Are linguistic/semiotic signs useful ways of describing the linguistic be-haviour of communication participants used by the analysts? (After all, a map is not a territory and yet it helps navigate through space.)

4. Finally, is language as a verbal system of communication independent from other communicative systems such as gesture?

These questions are the focus of the debate between our experts: Dylan Glynn, Barbara Dancygier and Johanna Kißler and published below. Cornelia Müller promised to address these issues in the 2020 volume of LAMiCUS.

ReferencesBarlow, Michael, Suzanne Kemmer (eds.) 1999: Usage-Based Models of Language.

Stanford: CSLI PublicationsBybee, Joan. 2006: From usage to grammar: The mind’s response to repeti-

tion. Language 82, 711-733.Cienki, Alan 2016: Cognitive Linguistics, gesture studies, and multimodal

communication. Cognitive Linguistics 27, 603-618.Cienki, Alan Cornelia Müller 2008: Metaphor and Gesture. Amsterdam: John

Benjamins.Dancygier, Barbara Eve Sweetser 2012: Viewpoint in Language. A multimodal

perspective. Cambridge: Cambridge University Press.Dancygier, Barbara Lieven Vandelanotte (eds.) 2017: Viewpoint phenomena

in multimodal communication, special issue. Cognitive Linguistics 28(3).de Saussure, Ferdinand 1916: Cours de linguistique générale. Paris: Payot. Fillmore, Charles J. 1988: The mechanisms of ‘Construction Grammar’. Berke-

ley Linguistic Society 14, 35-55.Glynn, Dylan, Kerstin Fischer (eds.) 2010: Quantitative Methods in Cognitive

Semantics: Corpus-driven approaches. Berlin: Mouton de Gruyter.Glynn, Dylan, Justyna Robinson (eds.) 2014: Corpus Methods for Seman-

tics: Quantitative studies in polysemy and synonymy. Amsterdam: John Benjamins.


Goldberg, Adele 1995: Constructions: A Construction Grammar approach to argu-ment structure. Chicago: University of Chicago Press.

Gries, Stefan T., Anatol Stefanowitsch (eds.) 2006: Corpora in Cognitive Linguis-tics: Corpus-based approaches to syntax and lexis. Berlin: Mouton de Gruyter.

Hopper, Paul 1987: Emergent grammar. Berkeley Linguistics Society 13, 139-157.Kißler, Johanna, Cornelia Herbert, Peter Peyk, Markus Junghöfer 2007: Buzz-

words: early cortical responses to emotional words during reading. Psy-chological Science 18(6), 475-480.

Kißler, Johanna 2013: Love letters and hate mails. Cerebral processing of emotional language content. In: Jorge Armony, Patrik Vuilleumier (eds.) 2013: The Cambridge Handbook of Human Af fective Neuroscience. Cambridge: Cambridge University Press, 304-330

Kress, Gunther, Teo van Leeuwen 2001: Multimodal Discourse: The modes and media of contemporary communication. London: Arnold.

Lakoff, George 2008: The neural theory of metaphor. In: Raymond Gibbs (ed.), The Metaphor Handbook Cambridge: Cambridge University Press, 17-38.

Langacker, Ronald 1987: Foundations of Cognitive Grammar. Theoretical prereq-uisites. Stanford Stanford University Press.

Langacker, Ronald 1991: Foundations of Cognitive Grammar. Vol. 2 Descriptive application. Stanford Stanford University Press.

Stefanowitsch, Anatol, Stefan T. Gries (eds.) 2006: Corpus-based Approaches to Metaphor and Metonymy. Berlin: Mouton de Gruyter.

Vandelanotte, Lieuven, Barbara Dancygier (eds.) 2017: Multimodal artefacts and the texture of viewpoint, special issue. Journal of Pragmatics 122, 1-9.

132 L aMiCuS 2019 no. 3


Karolina Krawczak2

1,2Uniwersytet im. Adama Mickiewicza w Poznaniu

Status konstrukcji multimodalnych w tworzeniu modeli i badaniach empirycznych- wprowadzenie do debaty

Język można rozumieć na wiele sposobów: jako system znaków wspólny dla członków danej społeczności, zachowanie komunikacyjne, zjawisko umy-słowe lub wrodzoną sprawność mózgu. Co to jednak oznacza dla naszego rozumienia języka, metod jego badania i modelowania?

De Saussure (1916) uznawał język za formalny system znaków złożony z par: element znaczący (forma) – element znaczony (treść). W psychologii społecznej język jest postrzegany jako jeden z wielu systemów znaków uży-wanych do komunikacji. Jego opis opiera się na określeniu, w jaki sposób znaki różnią się między sobą pod względem formy i znaczenia. W tym sensie naukę języka rozumie się jako odrębny proces, oparty na introspektywnej analizie zaobserwowanych produktów komunikacji. Z tej perspektywy nie ma potrzeby łączenia dyscyplin, co dziś może wydawać się dość przesta-rzałym podejściem.

Językoznawstwo kognitywne od początku (Langacker 1987, 1991) rów-nież uznawało znak językowy za dwubiegunowy symbol: z biegunem fono-logicznym (tj. formą) i biegunem semantycznym (tj. znaczeniem). Podobnie w gramatyce konstrukcji (Fillmore 1988, Goldberg 1995) jednostką gramatyki jest konstrukcja rozumiana jako połączenie formy i funkcji. W tym podejściu

133Status konstrukcji multimodalnych...

biegun formalny zawiera zarówno informacje składniowe, jak i fonologiczne (takie jak prozodia i intonacja), podczas gdy biegun funkcyjny koduje tre-ści semantyczne i pragmatyczne. Tak rozumiane konstrukcje tworzą sieć taksonomiczną, która stanowi gramatykę języka i jest w całości oparta na uzusie (Hopper 1987, Barlow & Kemmer 1999, Bybee 2006), tzn. wyłania się z użycia języka. Konstrukcje mentalne mają różne poziomy schematyzacji i podlegają dynamicznym przekształceniom, w miarę jak użytkownik ję-zyka napotyka nowe przykłady. Nieodłączna dynamiczność i wynikająca z niej zmienność systemu sprawia, że gramatyka postrzegana jest raczej jako probabilistyczna niż oparta na regułach. Oznacza to, że należy ją opisywać jako serię wzorców, które wynikają z użycia i powstają poprzez uogólnienie napotkanych przykładów rzeczywistego użycia, co potwierdzają badania korpusowe (np. Stefanowitsch & Gries 2006, Gries & Stefanowitsch 2006, Glynn & Fischer 2010, Glynn & Robinson 2014).

Oprócz wagi badań struktury języka opartych na korpusach, języ-koznawstwo kognitywne mocno podkreśla psychologiczną wiarygodność modeli językowych, postrzegając eksperymenty psycholingwistyczne jako ważną część testowania hipotez (Lakoff 2008). Wiele badań empirycznych poświęcono ostatnio zagadnieniom dotyczącym języka i emocji, kładąc pod-waliny pod badania treści emocjonalnych zakodowanych i przekazywanych za pośrednictwem języka. Liczne publikacje pokazują, że emocje zawarte w przekazie werbalnym i niewerbalnym wywierają silny wpływ na postrzega-nie i przetwarzanie treści językowych, ułatwiając zwłaszcza ich przetwa-rzanie na poziomie leksykalnym i zdaniowym (Kißler i in. 2007, 2013). Ten nowy paradygmat stawia szereg pytań dotyczących interfejsu język/emocje. Czy treści językowe i emocjonalne należą do różnych systemów reprezenta-cji? Czy należy opracować nową semantykę w celu włączenia empirycznie nabytych znaczeń emocjonalnych do systemu językowego? Czy znaczenia emocjonalne i językowe nabywamy razem czy osobno?

W ostatnich latach multimodalny charakter języka wzbudził duże za-interesowanie językoznawców i językoznawczyń kognitywnych (np. Cienki i Müller 2008, Cienki 2016). W rezultacie odchodzi się od postrzegania ję-zyka jako systemu czysto werbalnego z cechami prozodycznymi, kierując się jednocześnie ku takiej jego interpretacji, która pozwala wziąć pod uwagę sygnały wizualne, m.in. postawę ciała, gesty lub wyraz twarzy. W tym rozu-mieniu gesty współwystępujące z mową są niezwykle ważne dla interpretacji aktu komunikacyjnego. Ale czym dokładnie jest gest? Czy gesty są częścią konstrukcji? Czy istnieje gramatyka gestów? Jakie miejsce powinien zająć gest w multimodalnym modelu języka? W jaki sposób język migowy wyraża się multimodalnie?

Wreszcie multimodalność jest nie tylko cechą komunikacji mówionej. Zarówno język pisany, jak i pewne nieliterackie rodzaje komunikacji, w tym

134 L aMiCuS 2019 no. 3

dyskurs internetowy, reklamy, kreskówki czy komiksy, można uznać za mul-timodalne. Obraz i przekaz werbalny biorą udział w procesie tworzenia zna-czeń, przy czym nawet kształt liter może wpłynąć na ocenę wiadomości (np. Kress & van Leuveen 2001, Dancygier & Sweetser 2012, Dancygier & Vandelanotte 2017, Vandelanotte & Dancygier 2017).

To prowadzi nas do serii pytań, które zadano uczestnikom debaty:1. Co multimodalność oznacza dla zrozumienia, badania i modelowania

języka i komunikacji? 2. Czy interakcje multimodalne lub teksty są rozumiane przez użytkowni-

ków języka dzięki odniesieniu do reprezentacji umysłowych? Jeśli tak, to czy te reprezentacje są mono- czy multimodalne? (Reprezentacje nieko-niecznie rozumiane jako niezmienne wzory, ale jako dynamiczna akty-wacja układów neuronowych).

3. Czy znaki językowe/semiotyczne, z których korzystają badacze, są przy-datne w opisie zachowań językowych uczestników komunikacji? (Przecież mapa nie jest tym samym co terytorium, a mimo to pomaga odnaleźć się w przestrzeni).

4. Wreszcie czy język jako system komunikacji werbalnej jest niezależny od innych systemów komunikacyjnych, takich jak gest?

Te pytania są przedmiotem debaty między naszymi ekspertami: Dylanem Glynnem, Barbarą Dancygier i Johanną Kißler. Podsumowanie debaty opu-blikowane jest poniżej. Cornelia Müller obiecała odnieść się do tych kwestii w kolejnym tomie naszego pisma, który ukaże się w roku 2020.

Tłum. Anna Jelec

BibliografiaBarlow, Michael, Suzanne Kemmer (red.) 1999: Usage-Based Models of Language.

Stanford: CSLI PublicationsBybee, Joan. 2006: From usage to grammar: The mind’s response to repeti-

tion. Language 82, 711–733.Cienki, Alan 2016: Cognitive Linguistics, gesture studies, and multimodal

communication. Cognitive Linguistics 27, 603-618.Cienki, Alan Cornelia Müller 2008: Metaphor and Gesture. Amsterdam: John

Benjamins.Dancygier, Barbara Eve Sweetser 2012: Viewpoint in Language. A Multimodal

Perspective. Cambridge: Cambridge University Press.Dancygier, Barbara Lieven Vandelanotte (red.) 2017: Viewpoint phenomena in

multimodal communication, wyd. specjalne. Cognitive Linguistics 28(3).de Saussure, Ferdinand 1916: Cours de linguistique générale. Paryż: Payot. Fillmore, Charles J. 1988: The mechanisms of ‘Construction Grammar’. Ber-

keley Linguistic Society 14, 35–55.

135Status konstrukcji multimodalnych...

Glynn, Dylan, Kerstin Fischer (eds.) 2010: Quantitative Methods in Cognitive Semantics: Corpus-driven Approaches. Berlin: Mouton de Gruyter.

Glynn, Dylan, Justyna Robinson (eds.) 2014: Corpus Methods for Semantics: Qu-antitative Studies in Polysemy and Synonymy. Amsterdam: John Benjamins.

Goldberg, Adele 1995: Constructions: A Construction Grammar Approach to Ar-gument Structure. Chicago: University of Chicago Press.

Gries, Stefan T., Anatol Stefanowitsch (red.) 2006: Corpora in Cognitive Lin-guistics: Corpus-based Approaches to Syntax and Lexis. Berlin: Mouton de Gruyter.

Hopper, Paul 1987: Emergent grammar. Berkeley Linguistics Society 13, 139–157.Kißler, Johanna, Cornelia Herbert, Peter Peyk, Markus Junghöfer 2007: Buz-

zwords: Early cortical responses to emotional words during reading. Psychological Science 18, 6, 475-480.

Kißler, Johanna 2013: Love letters and hate mails. Cerebral processing of emotional language content. W: Jorge Armony, Patrik Vuilleumier (red.) 2013: The Cambridge Handbook of Human Af fective Neuroscience. Cambridge: Cambridge University Press, 304-330.

Kress, Gunther, Teo van Leeuwen 2001: Multimodal Discourse: The Modes and Media of Contemporary Communication. Londyn: Arnold.

Lakoff, George 2008: The neural theory of metaphor. W: Raymond Gibbs (red.), The Metaphor Handbook. Cambridge: Cambridge University Press, 17-38.

Langacker, Ronald 1987: Foundations of Cognitive Grammar. Theoretical Prere-quisites. Stanford: Stanford University Press.

Langacker, Ronald 1991: Foundations of Cognitive Grammar. Tom. 2 Descriptive Application. Stanford: Stanford University Press.

Stefanowitsch, Anatol, Stefan T. Gries (red.) 2006: Corpus-based Approaches to Metaphor and Metonymy. Berlin: Mouton de Gruyter.

Vandelanotte, Lieuven, Barbara Dancygier (red.) 2017: Multimodal artefacts and the texture of viewpoint, wyd. specjalne. Journal of Pragmatics 122, 1-9.

136 L aMiCuS 2019 no. 3

Barbara DancygierThe University of British Columbia

The status of multimodal constructions in developing models and empirical testing - a multimodal semiotic perspective

The primary focus of this conference has been multimodality. The organizers also proposed some questions for discussion, helping us think about how the recent focus on multimodal aspects of communication relates to or changes our approach to language.

Like other debaters, I would like to tweak the questions a bit, to be clear about where my expertise and my convictions lie. For example, we were asked to comment about the existence of mental representations. Not being an expert in this area, I am cautious about sounding too confident, but I would still argue that some form of representation of concepts needs to be there somewhere, so that we can reach some stable information about elements of reality. The way I typically think about such issues is by refer-ring to frames, in the Fillmorean sense – knowledge structures we use and access all the time; importantly, accessing an aspect of the frame (through an action such as reaching for the phone) equals accessing the entire frame (understanding the phone to be a communication device, acknowledging that phone communication happens across different locations and without face-to-face contact, etc.).

I also think consideration of multimodality adds some useful angles to the concept. Frames can be accessed via language, but also via embodied be-



havior, images, and aspects of sensory perception. Like language, an image evokes a frame, even when it is not a complete representation – so a child might draw a yellow blob with lines sticking out of it, and we do recognize it as the sun. In fact, sounds and onomatopoeic expressions are equally ef-ficient in evoking concepts, which is why children are happy to call a train ‘a choo-choo’. Yes, such representations are schematic and partly conventional, and they can be further described as iconic or indexical, but they connect to some shared idea or concept – the frame. And they make the connection without relying on language only, even if a linguistic item is available.

We were also asked to comment whether representations are mono- or multi-modal. I am convinced that we have access to concepts that come with multiple aspects of representation – linguistic, but also visual, audi-tory, tactile, or related to a motor program – in other words, concepts are represented multimodally. This naturally means that frame structures are multifaceted and this complexity of representation is why frame metonymy is such a common trope – you do not need to represent a full frame in com-munication, but you may access the frame structure by being given a token, and the tokens can be of various kinds. Looking at various multimodal ar-tifacts which rely at least partially on visual representation, such as ads or street art, I see a broad range of creative forms which evoke a frame visually, by showing an aspect of it (for example, McDonald’s ‘golden arches’ for the food served and business model used). But such artifacts also distort various aspects of the frame, in various modalities. They change size, color, shape or texture, to evoke familiar frames and yet communicate something new. They change form enough to draw attention to it, but not too much, so that the viewer or listener can still recognize it and construct meaning starting from a familiar frame. Importantly, those multimodal representations are so salient and stable that manipulating them does not cause lack of com-prehension. Rather, what happens is construction of new meanings out of the existing ones.

Let me brief ly clarify what I mean by frames being ‘stable’. The invita-tion to the debate suggests that representations need not be understood as stable patterns but as dynamic co-activation of neuronal assemblies. From my perspective, as a linguist rather than a neuroscientist, I view stability as the existence of frames which can be reliably shared by various speakers over a certain time span, even if the information in the frame changes. To return to the simple example of the ‘phone’ frame, contemporary phones are not rotary and do not have separate receivers – the material form and the attendant motor programs have changed. And yet, when one considers phone icons, they usually look like rotary phones or like receivers (check the on/off icon on your mobile phone, which typically looks like a receiver). So what is stable and what is dynamic, given how quickly the technology

138 L aMiCuS 2019 no. 3

changes? Much of the frame, such as the idea of remote communication, remains stable, and while the actual material or visual aspects change, the ‘left-over’ iconicity remains stable.

Construction of meaning thus requires that frame structure is largely stable, but dynamically used, and that at least some visual and enactive as-pects are reliable. This is true about a number of communicative forms, but especially true about the forms which metonymically appeal to aspects of experience. One interesting example is color. For example, most Lego ads, however creative, will maintain the primary Lego colors; also, a McDonald’s ad will somehow use bright red and bright yellow – both use color to evoke rich frames of the products the companies sell (toy plastic blocks a child can use to build things, or fast food). However, when Lego advertises the artistic potential of playing with the blocks, they show a picture imitating the Mona Lisa, made of Lego bricks, but in colors that Lego does not offer – closely resembling the tones of the Leonardo painting. So in one case, color evokes the familiar Lego set, in the other, it evokes the artistic potential. In both cases rich frames are evoked, with one aspect of the representation (color) highlighted in one case and tweaked in the other.

To continue with the Lego example, the various modalities of the frame have to work together to create meaning. Constructing things out of plastic blocks requires a mental representation of what the final product will look like, but it also requires a sense of how manipulation of smaller structures contributes to the emergence of larger structures. In a sense, the experience correlates with meaning construction – well defined building blocks can be brought together through embodied experience into a new, richer concept, complete with material, tactile, visual, and other multimodal features. The underlying multimodal nature of concepts is necessary for meaning to emerge.

Another question we were asked is about the connection between signs as discussed in semiotics and linguistic forms. The example of a map is very useful here – it is a visual form, and yet makes navigating space possible, without using language. I think the question is tapping into further aspects of multimodality, especially visual representation. Before I respond, I’d like to clarify my assumptions.

There is a bit of a theoretical and terminological overlap that is often taken for granted, but requires clarification. For a semiotician, language, as a symbolic system, exists naturally in the broader context of various signs, including icons and indexes. The difference between language and other signs matters in this context. But it seems to me that the need to make such distinctions is less obvious if we consider the role of conceptualiza-tion, and the scope of the terms used. Recently, multimodality has become a term that covers a lot of ground, as it refers to many aspects of cognition and communication: there is ‘multimodality in interaction’, that is, gesture,


eye-gaze, etc., and there is interaction between language and other semiotic channels, such as image or sound; the latter trend is particularly interesting to semioticians, and that’s where the issue of whether semiotic or cognitive linguistic interpretations are similar or different is the most pronounced. The way the question has been formulated (especially via the map example) seems to suggest that the division between multimodal embodied interaction and multisensory artifacts is worth maintaining. I actually disagree. The reason why any such division has to be misleading is conceptualization – the crucial aspect of any form of expression. Cognitive linguistics gives priority to conceptual structures, over specific forms of expression, and approaching both ‘multimodalities’ from that perspective obviates the need for ad hoc distinctions. Multimodality can be approached from the perspective of com-municative load-sharing, and the various embodied or semiotic modalities can interact with language in a number of complex ways.

I will return here to the examples I discussed in my talk, which I de-scribed as ‘street deixis’. In my neighborhood, a f lower shop has recently closed down, leaving a large sign in the window: “Thank you, Kitsilano!”. Language is used here in a rather standard form of giving thanks, but that’s the only typical aspect of this utterance. The addressee, ostensibly the whole neighborhood, is any inhabitant, any passer-by, any person walking the street. In fact, walking is not enough, the passer-by has to stop, use eye-gaze to focus on the window (rather than another person), its message, and the sign. The ‘speaker’ here is the business, the f lower shop, which is still, in the mind of the passer-by, metonymically associated with the particular location, while in fact no longer occupying it. And so on, and so forth. Devi-ations from ordinary speech acts of giving thanks are numerous here, and so profound that they should invalidate the act altogether, and yet all works fine. I would argue that it does because all the multiple channels of commu-nication, all the multimodalities involved, linguistic, deictic, visual, and em-bodied, work together on the basis of familiar concepts. It is quite clear how we need stable frames and conceptualizations to interact with such a sign, especially those pertaining to ordinary language interaction and the prag-matics of speech acts, but also to doing business in a neighborhood, moving along the street, turning to look at a shop window, etc. Conceptualization holds this communicative event together (though only for the brief moment of someone stopping and reading), while the nature of specific expressive modalities is adjusted to the situation. Meaning construction mechanisms take over, regardless of the modalities used. What is necessary is the mental representations in the mind of the passer-by, who will be transformed for a brief moment from just a person walking by to a participant in a communi-cative act. Interaction with the modalities is what holds such acts together, regardless of their semiotic nature.

140 L aMiCuS 2019 no. 3

So, to conclude, the distinctions between linguistics and semiotics are only important if we look at the form of expression, but they disappear when we look at the paths of conceptualization. Conceptualization patterns are the great equalizer here. And we need to add that cognitive linguistics has developed an impressive toolkit to describe conceptualization, so it can now apply it to multimodal artifacts. In fact, as I argued in my talk, language use adjusts to the context of multimodal artifacts, and so even if we focus on language alone, we can learn much from multimodal usage. Some observa-tions would simply escape us if we insisted on dividing lines based on form.

Besides, talking about signs seems to assume a clear relationship be-tween form and function. But we have to consider that, to give just one ex-ample, the increased use of images, whether iconic or symbolic, is commu-nicatively a relatively new thing, and new forms are emerging all the time. Are these innovations signs? Or are they simply emerging conventions which are subject to change as they evolve to fit new communicative needs by using new available means? Internet memes are a good example here. New ones are emerging all the time, and old ones are being adjusted and changed (for example, it used to be one image macro and text, but more and more often the image macro is having other images superimposed or photoshopped onto it). It seems to be more productive to focus on conventional usage and its emerging constructional features than to search for semiotically satisfactory definitions. As I said earlier, cognitive linguistic tools are very well suited to capturing the emergence of new forms and meanings, and correlations between them.

The next question asked us to ref lect on whether language is indepen-dent of other modalities. I think the comments I made earlier make it pretty clear that my answer is “No, absolutely not.” Language is immersed in all other modalities, in embodied interaction, in visual, auditory and tactile perceptions. And, in action. This is because language participates not only in embodied cognition, but also in enactive cognition – and these areas of cognition pertain to how we rely on our bodies to communicate things, but also to do things.

The next question that matters a lot here, and which has been raised, is methodological – if language is so deeply multimodal, how are we going to test our hypotheses? How do we gather data? How do we process data when the data is “everything”? We have developed various methods of enquiry, but they are not so open-ended as to cover all the aspects of multimodality that I mentioned. Representative examples can be gathered, but what tests can we apply to them? Given that the answers are not easy to propose, what should we do? I can just feel the dreaded ghost of ‘introspection’ looming over us. Well, I’m actually not afraid, because I feel that ‘introspection’ is nothing more than a spectre some of us resurrected from the olden days of


armchair linguistics. What we do when we gather examples to support hy-potheses, even if we cannot provide sufficiently rich corpora, is we analyze. As long as our examples are attested and as long as there are sufficiently many of them that we can recognize patterns, we are being analytical, not introspective. Perhaps with time all forms of multimodal communication will lend themselves to corpus and quantitative methods. The technology is out there and it will support our efforts. But meanwhile, we have to go on looking for patterns, ideas, and hypotheses. There are presumably many ways to understand multimodal behavior and communication, and we have to try them all. We cannot limit our enquiry to things that the methodology we ourselves have designed allows us to do. The methodology will catch up, if we give it a chance.Thank you!

142 L aMiCuS 2019 no. 3

Johanna KißlerBielefeld University

The status of multimodal constructions in developing models and empirical testing - a psychological and neuro-cognitive perspective

The previous debaters have already introduced and highlighted important issues. Therefore, I will try to touch upon already articulated viewpoints in my own comments as I address the questions presented to us:

One question regards the breadth or inclusiveness of cognitive linguis-tics and language studies, asking should multimodality be considered as a part of language, should gesture be a part of language, should picture me-dia, and even artefacts seen in street communication be a part of language?

My intuitive answer is yes to all of these. Yes in that these are widespread, related, interesting, and important communicative media and phenomena worth studying scientifically. From my own point of view, I would also like to include emotion to that list, another phenomenon important in commu-nication that is typically treated as extra-linguistic and that is dear to my heart. In fact, it is hard to imagine any real-life language discourse devoid of emotion. So much so, that there have been heated debates as to whether emotion should not really be called communication. Obviously, allowing for such breadth brings up a definitional issue, where each field needs to decide and to a large extent also negotiate the scope of its research. In broadening language studies to include all kinds of old and new “multimodal” commu-



nication phenomena, one incurs the risk of definitional, and to some extent also methodological, vagueness and arbitrariness. What kinds of approaches, methods and tools should be used for this type of analysis?

In linguistics, I believe, a great leap forward has been made by the “Chomskyan revolution”, with its focus on syntax and combinatorial rules. As great as this leap has been, providing among other things a basis for now-adays widespread automatic language processing by machines, it has also resulted in many self-limitations. Opening up the field via cognitive linguis-tics and perhaps also more generally communication studies therefore has borne and still bears great promise. Combining the virtues and rigor of the Chomskyan approach with the open-mindedness of today’s Cognitive Lin-guistics, as I perceive it, and as I have been enjoying it during this conference, could greatly advance the field. In practice, this would mean specifying the combinatorial rules of multimodal communication, its syntax so to speak. A gargantuan task! To achieve it, new computational approaches, drawing on statistics, big data and machine learning techniques seem highly promising. Dylan Glynn has already talked about regression approaches, and also about classification of types and tokens which would be one basic requirement in this context. Computational linguistics and computer science approaches in general may be helpful here and may make the seemingly gargantuan task of generating a multimodal communication grammar considerably more man-ageable. On the commercial side, companies like Apple or Google have pro-vided an impressive, if somewhat terrifying, proof of concept in this regard.

Classification of types and tokens as addressed by Dylan Glynn pro-vides a good link to another topic that was brought up, namely the issue of representation. How, where, and in what format do we store knowledge and combinatorial rules? As a psychology student many years ago, I was overwhelmed and somewhat irritated by the concept of representation, since it seemed both highly abstract and at the same time suggested a physical reality that I could not wrap my mind around. Only much later – compara-tively recently in fact, in working with computer scientists – have I come to fully appreciate the usefulness and necessity of representation for modelling purposes. As a psychology student, however, I was deeply confused by it and started to look for something seemingly more real and tangible. I hoped to find it in studying the brain.

Today, many years later, I am of course aware that neuroscience is a different level of analysis and that reduction will not always buy you a lot. Whether brain science as a reductionist approach to cognitive or commu-nicative phenomena can provide answers will depend a lot on the questions asked. In general, we can reduce our sensory impression of, for instance, the color green to chemical phenomena on our retina and to the activity of light waves, fundamental physical phenomena. However, along this reductionist

144 L aMiCuS 2019 no. 3

pathway, one can easily lose track of the original question and its adequate level of analysis. After all, not all of us want to end up as astrophysicists. We may all lie awake at night and wonder about the origin of the universe, but we may do so in different ways, leading us to pursue different research questions and different career paths during the day.

So: Are there mental representations? I would say: not really in a ful-ly-understood physical sense. Still, my closest approximation of represen-tation from a biology-inspired point of view, is a view of representations as dynamic co-activations of neural systems on a millisecond scale that ulti-mately give rise to semantics and syntax. Here, I would fully agree with the organizers’ proposal. I also believe that neuroscience has made considerable progress in understanding how this happens. This is a level of analysis that has, over the years, come close and dear to my heart, not very surprisingly after all.

Because it falls broadly in my realm of expertise, I also want to bring up the concepts of multimodal neuronal assemblies and multimodal integra-tion and what these may or may not buy you regarding representation. Of course, understanding neuronal assemblies will not result in direct under-standing of interactive behavior of people in a room. However, it can reveal some principles that can be useful for studying language, communication and interaction. For instance, we know that there are initially separate pro-cessing pathways within our brains and that there are dedicated areas that integrate information coming from these pathways. We can study the inte-gration of the different pathways and see how they weigh auditory and visual input, how they weigh verbal input, how they weigh tactile input. In short, how they weigh different modalities against each other. For instance, the human brain tends to give a lot of weight to visual input over other modali-ties, this is a known fact. It also typically weighs pictorial input over verbal input. Or, as the saying goes “a picture is worth a thousand words”. This is something that Barbara Dancyger has also touched upon. In episodic mem-ory, there even is a “picture superiority” effect. Furthermore, in interaction and memory, there is also an “emotion superiority effect”– emotion weighs heavily on our minds and biases us in many different ways.

Still, we all come from a heavily verbal tradition. Our cultures have his-torically developed to be more verbal than anything else. This is for a good reason. Language allows us to communicate via a sparse, simplified code. Therefore, a word can also be worth a thousand pictures – or abstract from a thousand pictures. This is, in my view, because in the past non-verbal and non-language based representations were expensive. It was costly to pro-duce pictures and difficult to transmit them. Until recently one could not just make an audiotape of what people said, how they said it, or video-record their facial expression while they spoke. Instead, humans used a verbal code


when they sought to conserve. Initially via oral history, and since its inven-tion, mostly via written code, which we still rely on so much.

But these verbal and written codes are not what, I believe, we as hu-mans had originally evolved to process. Therefore, as soon as all these mul-timodal artefacts have become available and are thrown at us everywhere, we are led and potentially also mislead, by their seeming veracity and the fact that they feed into our biologically older processing systems whose input we weigh over the more recently developed linguistic, and particularly the written code. Oftentimes, if we have video recordings of speech or conver-sation available, we give more weight to the facial expressions than to what is said and we will in my experience and opinion typically weigh prosody over the content. Therefore, yesterday it surprised me to see a presentation that argued against this. In general, I believe that there are principles from neurobiology that we should consider and draw upon to explain phenomena that other people identify in the streets or in art history, or cultural history, or nonverbal communicative behavior.

Although I perceive neurobiology as a useful and recently reasonably successful level of analysis, it clearly need not be the only one. Representa-tions are useful concepts to think about, and they are necessities for compu-tational analysis, where their likely neurobiological substrate is still largely disregarded.

I am afraid I do not have much to say about the status of semiotic signs and their analysis-perhaps other than that to me semiosis in itself seems both too static and too broad to capture the dynamics of ongoing communication. However, I would like to remark on a different, but in my view related issue: So far, we are lacking a good mechanistic understanding of the many levels and dynamics of cognitive representation and I am not even sure whether the concept of “mechanistic knowledge” will seem desirable to the present audience. Still, many successful branches of (natural) science view mech-anistic understanding as an important goal, perhaps the ultimate goal of their research endeavors.

As with the question of multimodality, the questions of representa-tion and level of analysis are to a considerable extent questions of definition. Within the conference, but also with this debate, I thought there were many such questions of definition. I attended a presentation yesterday that was entirely devoted to definitional considerations, and I found this was inter-esting, because I come from a branch of academia where theoretical issues of definition have been largely replaced by operationalization and testability. Operationalization and testability loom large in my scientific world, and I believe that cognitive linguistics as I experienced it in the past three days could benefit from moving in that direction, ideally without losing its own elaborate theoretical foundations.

146 L aMiCuS 2019 no. 3

I fully advocate that there is also the issue of induction, of generating ideas. Here, strictly experimental approaches arguably may have thrown out the baby with the bathwater. Focusing only on testability and getting rid of induction, getting rid of generation of ideas, so to speak, results in stark and possibly harmful limitations. There is no need to shut out the “real world”, just walking down the street and looking around, looking at phenomena in the real world, can be an important part of induction.

Having said that, I think it is equally important to look at the status of data and decide: Is this a datum? Is it an example? In many cases in what I have seen in the past days, what I have seen called data, to me are examples. There is nothing wrong with these observations per se. Still, from my point of view, it is important to clarify their status and to determine to what extent an example really “transcends” to a different level – either because it points to something very common – which could be determined quantitatively or to something very powerful, a new mechanism, a new dynamic mapping between symbols, for instance. In the end, it is up to each discipline to decide what is to be treated as evidence.

On the other hand, personally, I very much regret that psychology, the discipline that I come from, in spite of being viewed by de Saussure as a natural point of departure for language studies, recently has not been inter-ested very much in the study of naturalistic language-use or communication. Exceptions exist regarding issues like the experimental processing of pitch, faces, words, and the likes, or how these are integrated in the brain. These are all important. Still, a more macroscopic or even “holistic” look would also be desirable and nowadays the methods are there to capture and ana-lyze larger, more realistic and therefore often multi-modal, units without losing control, without falling back to “armchair philosophy”. An almost exclusive focus on micro units seems odd for a science that claims to aim for an understanding of how people behave. Psychology is the “study of human experience and behavior”. Multimodal interaction clearly takes up a consid-erable part of human behavior. Therefore, a joint, truly interdisciplinary and multi-perspective effort may in the long run turn out to be most fruitful to understand the different aspects and levels of multi-modal communication.

148 L aMiCuS 2019 no. 3

Dylan GlynnThe University of Paris 8 Vincennes - Saint Denis

The status of multimodal constructions in developing models and empirical testing - a methodological perspective

I’m not really a specialist in multimodality, even if I’m extremely interested in it. I have doctoral students who work on it, but I wouldn’t call myself an expert by any means. If I were to consider myself an expert at all, it would be in methodology for linguistics and I expect my contribution will be from a methodological perspective. Professor Fabiszak kindly sent us four questions and last night, and before the dinner, I sat down and tried to think my way through them. I will try to answer the questions but in reverse order since I felt the first questions were the more difficult and important and the last questions were less challenging.

The last question is more or less: Is language independent from other com-municative systems such as gesture? I hope this is a reasonable summary of the question. As a cognitive linguist, the answer is obviously and emphatically no. In fact, I feel the question would best be turned around: why would lan-guage be independent from other communicative systems? If we accept Oc-cam’s Razor (the principle of simplicity in scientific explanation), one would need an empirical reason (an observational reason) why one would want to posit the existence of a theoretical construct such as an independent faculty of language. Now, I’m not saying there aren’t good reasons to posit this. There is research beyond my areas of expertise, in aphasia for example, that



may warrant such a proposal. However, clearly the onus should be on those who believe that we need such an understanding of language to argue for it. Therefore, the question should rather be, are there arguments for under-standing language as an independent communicative system?

I do wish to add a few caveats to my initial categorical response. Firstly, this position does not entail that various structures in semiosis, syntax, lexis, facial expressions and so forth, are not highly specialised. However, highly specialised is different from being independent simply because highly spe-cialised does not preclude interdependent structure with other cognitive capacities.

Secondly, I can think of various reasons for analytically treating language as an independent system, even if it is understood theoretically as part of a broader set of communicative competences and structures. The most obvious reason is the complexity of the system. In science, it is widely believed that it is better to say more about less than say less about more, a maxim with which I could not agree more. However, given the fact that previous approaches to language (that have treated it as an independent system) have yet to ade-quately account for it in those terms, despite the daunting complexity, the Cognitive Linguistic ‘holistic’ approach may well be necessary. As scientists, we need to limit our object of study but not at the expense of being able to explain it. Of course, that in turn places the onus on those who seek to explain language as part of a holistic semiotic-communicative system to demon-strate how that can be done rigorously, i.e. adhering to scientific principles such as falsifiability. It is precisely this challenge that I believe is the most important challenge for Cognitive Linguistics or any functional theory of language which holds that an explanatorily adequate approach cannot afford to delimit its research to language as an independent phenomenon.

To summarise my response, I can say that no, language should not be the-oretically understood as an independent system unless you can give me a good reason otherwise. Moreover, it should not be empirically treated as such despite the fact that this renders our object of study exponentially more complex.

I believe question three can be summarised as: Are linguistic semiotic signs useful ways of describing linguistic behaviour? However, I am concerned that I have misunderstood this question. In light of my concern, I’ll do my best to answer both the literal question and what may be implied by that question. To begin with the literal interpretation of the question, let me turn to something I discuss when teaching. Each year, I ask my first-year students to think of something that they cannot name. Of course, they are stumped and are never able to think of something that cannot be named. We conclude that to think of something, to conceptualize something, one categorises it and by doing so one creates or, indeed usually, assigns a pre-determined form-meaning pair, a semiotic sign. In our classes, we call this

150 L aMiCuS 2019 no. 3

cognising. It is from this point that we introduce the theory of “Cognitive Linguistics” – language understood as experience-based conceptualisation.

In class, to underline the extent that we cognise our experienced world, we consider two examples. Firstly, we take the example of a young child when he or she sneezes. I recount a story, where when I first moved to Hungary, I was surprised that the children would make the sound Haptsi when sneezing. I observed that even children in the relatively early stages of language acqui-sition would “say” Haptsi. No one in my classes “says” Haptsi when they sneeze. They say something like Achoo, as in English. Similarly, a second example is the verbal response to accidently burning oneself cooking in the kitchen, where a French speaker will “say” Aaii!, an English speaker will “say” Ow!. In these instances, we are observing behavioural responses, not lexemes in any traditional sense of the term. Yet we see that they are, in fact, form–meaning pairs. In other words, they are linguistic signs. In the microseconds between experiencing pain or the desire to sneeze and uttering the verbal response associated with those experiences, we cognise the experience and ascribe a learnt semiotic sign to it. From this, we conclude that it is very difficult for us to think or behave without cognising.

Given that I take this response to the literal question as self-evident, I expect the implicature in the question was about objectivity and the in-herent circularity that results from using the sign-system of language to describe language. Perhaps, the implicature is, therefore, that we should use logic or mathematics instead of signs to describe communicative events. If the question implies the use of logic, as in truth-based reason, then I have yet to see an adequate description of linguistic behaviour using such reasoning. Given that this is a Cognitive Linguistics conference, I find it unlikely that this was the intention of the question. However, the convenors know that I specialise in quantitative methods, especially the application of quantitative methods to the slippery world of qualitative data and linguistic behaviour. Therefore, perhaps the question implies the use of mathematics.

In response to this, I would say that statistics, or mathematics applied to describing the probability of events in the world, is also predicated on signs. In Fig. 1, you see the mathematical equation that is linear regression, a widely–used statistical method in linguistics.

Obviously, this is a mathematical formula, written using symbols. In Figure 1, the y signifies the dependent variable, the k signifies the intercept and the

Figure 1. Linear Regression


β signifies the predictor. There exists a well-worn debate in mathematics as to whether formulae are discovered or invented. Does the mathematician discover something in the universe and ascribe signs to it or are the formulae creations that represent (signify) how humans can explain the universe? Such debates, however, do not concern the use of applied statistics. In the formula presented here, the symbols are form-meaning pairs. In this instance they represent variables and, in any given application of this formula, these vari-ables are directly or indirectly quantifiable phenomena – things in the world.

In summary, other than perhaps logic which has been systematically shown to produce inadequate descriptions when applied to parole / perfor-mance, I can think of no way we could describe linguistic behaviour without using signs. Perhaps a more fruitful question would be: How can we describe linguistic behaviour in a way that produces falsifiable results? Taking this further, is it possible to use quantitative methods to help determine descriptive and predictive accuracy of those descriptions? Arguably, these questions are more pertinent than whether the use of linguistic signs to describe linguistic be-haviour is appropriate or “objective”.

Let us turn to the first two questions, which were, in my opinion, the most important.

I found that question two was full of many implications and entail-ments, and so I broke it down into what I thought were the three principal sub-questions:

Are multimodal interactions understood as mental representations?

Are these representations mono- or multimodal?

Are these representations best understood as stable patterns on neural assemblies?

Firstly, can we understand multimodal interactions as mental representations? My initial response is: What are mental representations? How are they oper-ationalised? Are we speaking about Chomsky’s trees or Langacker’s pic-tograms? Do we mean instantiations of schematic relations? I know that as a linguist, representations are constructs we use to help us talk about an object of study that is not observable. I think perhaps this last point is crucial. So leaving aside a meta-discussion on whether representations are scientific or the empirical question as to whether they exist as part of an individual’s linguistic competence, representations can be understood as analytical tools. I hope we can all agree that we don’t have pictures in the brain and that image schemata and grammatical pictograms don’t ex-ist anymore than X-bar patterns exist. If we do all accept that point, then representation is understood firstly as an analytical tool. In order to ap-preciate whether representations are applicable to multimodal interactions,

152 L aMiCuS 2019 no. 3

I would prefer to think about representations operationalised as generali-sations across usage. It should be evident that I am here evoking Langacker (1987) and Hopper’s (1987) Usage-Based Model where structure emerges from usage just as a path emerges across a frequently traversed field or forest. If one accepts this operationalisation of representation, then of course gen-eralisations can emerge from multimodal interactions and of course they can be described in terms of representations. Whether speakers possess these generalisations and employ them in semiosis is an empirical question. Per-sonally, I find Dąbrowska’s (2004, 2008) position and evidence in this regard the most convincing. This position can be summarised as: sometimes yes and sometimes no because it can vary from speaker to speaker and even from time to time for a given speaker.

As for the next question – Representations, whether they are monomodal or multimodal. We are cognitive linguists, everybody here, so we treat language structure in a non-modular holistic way. That is about as fundamental as one can get to the theory of Cognitive Linguistics. I expect the text-book response to this question would be: if, for the individual (and, by extension, the speech community), the extra-verbal sign (an emoji, a gesture, a raised eyebrow etc.), is an ad hoc or infrequent addition to a type of verbal component in communication (a morpheme, lexeme or construction), then it is probably treated as context information that we use in processing, combined with encyclopaedic semantics to help understand the communicative event. In such situations, one would speak of mono-modal representations (or gen-eralisations). However, if, for the individual (and, by extension, the speech community), the extra-verbal sign is frequently or systematically associated with a type of verbal communication, then it will become entrenched as such. In these situations, one would speak of multi-modal representations (gen-eralisations). As a cognitive linguist, I typically employ the term construction to refer to the Gestalt of any complex for form-meaning pair, mono-modal or multi-modal. I am confident that this would be unproblematic for any lin-guist that agrees with the tenets of the cognitive paradigm. Prosodic mark-ing, facial expressions or even memes, indeed any form-meaning pair, is a potential component of a construction. Where construction is understood as a representation of a complex form-meaning pair constituting a part of a speaker’s linguistic competence.

Finally, the last sub-question concerned the co-activation of neural assem-blies versus stable patterns. As an empirical quantitative linguist, my immedi-ate response is: to the best of my knowledge, our understanding of synaptic connectivity, plasticity and organization (neural assemblies) still offers no data that facilitate the description of conceptual structure – representations. For this reason, I will side with stable patterns. Indeed, the notions of pat-terns and stability are central to the usage-based model of language. With


stability, we are talking about the relative consensus between speakers on the language system at a given place and time. Put in other terms, this is the degree of conventionalisation in language. This can be operationalised in terms of the predictive accuracy derived from relative frequency or rel-ative salience or, ideally, a combination of the two (cf. Arppe et al. 2010 for discussion on this point). Pattern is a notion that we use to capture both the idea that the structures we identify are not discrete and that they are multidimensional (complex) in nature. Put simply, instead of rules, which are by definition discrete and which tend to lead people to think about struc-tures independently from one another, we speak of patterns. Patterns, even complex patterns, can of course be operationalised in terms of probabilities making the notion the perfect tool for speaking about language structure from a usage-based perspective.

This response deserves a simple caveat. I do not wish to suggest that the study of neural phenomena cannot inform linguistics. At some stage in the future, I believe it may be crucial in our efforts to fully explain conceptual structure. Nevertheless, from my perspective with a limited understanding of the field, there remains a large gap between the study of neural structure and anything that we would call conceptual structure.

The first question is perhaps the most difficult, but perhaps also the most pertinent: Multimodality is not only a feature of spoken communication. Writ-ten language and other non-spoken modes are multimodal. What does this mean for our understanding, investigation and modelling of language?

As somebody specializing in methodology, it was the last bit that caught my attention: what does this mean for investigating and modelling language? If we believe that language has a multimodal potential, which I assume we all do as cognitive linguists, then we must extend our model to include it. Given that we believe that in order to explain language it must be treated holistically but also that the language faculty is part of a general cognitive capacity, this much should be reasonably obvious in Cognitive Linguistics. Nevertheless, with respect to this last point, I do have some opinions that may be more contentious and therefore more useful to discuss.

Let us begin by considering emojis. Language is a systematic, struc-tured, semiotic system, but it is not closed nor is it static. Via the iconic symbols that made up early writing systems, the use of images in language dates back to the origin of writing. Similarly, we all know that using an ex-clamation mark or a question mark in written text or an emoji in an SMS is highly structured, is part of the grammar of language. Interestingly, one must assume emojis and punctuation originated primarily as a means for expressing information canonically encoded by prosodic patterns and facial expressions (and perhaps some other gestures). If there were ever doubts that gesture and prosody were beyond the realm of grammar, such an observa-

154 L aMiCuS 2019 no. 3

tion should put such doubts to rest. Given this, why not simply extend this principle to memes or any other semiotic production, simplex or complex? What I believe is at stake here is the question of the systematicity of the form–meaning pairing (what cognitive linguists would call entrenchment and functional linguists would call automatisation) versus ad hoc form-mean-ing pairs produced in context and dependent on that context.

In other words, the problem should not be whether language is mul-timodal or whether emojis have entered language as a structured part of that language, but determining which of these non-verbal types belong to speakers’ competences and which do not. The problem is rendered all the more difficult because the distinction between entrenched and ad hoc signs is non-discrete, varied and dynamic, between individuals but also for the individual. Let us consider this problem more closely.

Although we speak of languages, assuming the usage-based model, we are, in effect, always making generalizations across individuals in a speech community. Variation between individuals is an inherent part of the system and the reason we speak in terms of patterns not rules. However, it is also possible that, for a given individual, a sign (form-meaning pair) may be en-trenched at one stage of his or her life, yet at another stage, it may not be en-trenched (cf. Dąbrowska 2008). Moreover, the notion of entrenchment, even for the individual at a given moment in time is typically understood as a matter of degree. This amount of systemic instability when compounded by the fact that individual types (signs) in non-verbal communication can be extremely difficult to isolate as well as the fact that we lack transcription traditions for most non-verbal types, makes the issue of scientifically integrating extra-ver-bal communicative structure, such as emojis, into language study a nontrivial methodological question. A few examples should make this point clear.

Anybody who has received an SMS from their grandmother will prob-ably have found that she sounds a little stiff, as if something unspoken is amiss. The communication event is failing because one is expecting visual representations of certain interpersonal things that we do with emojis and punctuation marks. That’s multimodal grammar at work. The grammatical acceptability of the SMS from one’s grandmother is low because she omitted the little smiley at the end or a wink after her sarcastic comment. One would expect that modelling this, just like we model the pragmatics or inter-per-sonal semantics of lexis, should be straightforward.

The problem is a result of a lack of stability. Although phonetic and even phonological variation is part of verbal communication, relatively speaking, both the form and the entrenchment of the signs are stable. This is not the case for emojis. Between different computing platforms and in different contexts from the old school smiley your primary school teacher added to her marks in pen, via the keyboard representations for that venerable tradition


(such as :) or :P) to the animated gifs in modern chat, the formal variation is immense (cf. Tabacaru & Van Der Mark 2019). The analytical tools we have developed for language handle this degree of formal variation poorly (cf. Glynn forthc. for a discussion how we may overcome this problem). Further-more, in the case of emojis at least, variation in the degree of entrenchment is substantial. The case of one’s grandmother writing an SMS is an exam-ple in hand. For many people who do not regularly use computer-mediated communication, an emoji may be simply an ad hoc sign, processed actively using the full gamut of cognitive competences at the speaker’s disposal. It must also be remembered that this dynamic complexity in stability of form and degree of conventionality impacts upon language’s two most complex phenomena: semasiological variation (polysemy associated with the form) and the combinatory effects upon that semasiological variation (grammatical constructions). Lastly, of course all this complexity and variation needs to be sensitive to stylistic and other sociolinguistic effects. None of this is to say that we should not attempt to integrate such communicative elements into our models, but it is to say that it is an extremely difficult task.

Of course these methodological hurdles are not restricted to emojis. Actual facial expressions for which the emojis stand as symbolic graphs pro-duce different challenges. Unlike emojis, facial expression is so systemically integrated and preconscious that, sufferers of autism aside, we can be con-fident that their role in communication is firmly entrenched in all speakers’ competences. Whether it is recognising irony or mirativity or many other basic functions of language, facial expressions are a frequent and essential part of language structure. In this regard, facial expressions are more like prosody than emojis. Just like prosody, the identification of types and then, in turn, the tokens (q.v. Johanna Kißler uses the term “datums”) of those types is an extremely difficult endeavour. With an emoji, the identification of type and token is relatively straightforward, where what constitutes a given facial expression or prosodic pattern (type) and then identifying occurrences (tokens) of that facial expression or prosodic pattern systematically is a fun-damental problem for research in gesture and prosody.

When we extend these questions to complex novel signs such as memes (q.v. Barbara Dancygier) we are faced with both the kind of complexity and instability we observe with emojis as well as the issues of type and token identification we face in the study of gesture and prosody. Is a given meme (type) widely entrenched in a language system? How do we deal with the many variations of that type – is it one type with many tokens or many dif-ferent types? Extrapolating for the discussion above should be self-evident.

Therefore, at least within cognitive linguistic circles, the question should not be if we should integrate multimodality in our descriptive, predictive and explanatory models of language, but how we can collect the data to attempt

156 L aMiCuS 2019 no. 3

this. For me, this is primarily a methodological challenge. We have a long and venerable tradition of identifying phonological, morphological and lexical types and tokens just as we have a reasonably well-developed tradition of identifying constructions, but multimodal types and tokens are still new. Put simply, until we have established analytical tools for identifying an occur-rence of the “throw gesture” as opposed to a non-semiotic physical response to some random stimulus, I believe we need to focus on the methodological challenges. For memes and emojis, we need to address issues of variation and stability, operationalised in terms of frequency (corpora) and salience (experimentation). For prosody and for gesture, we need to operationalise the notion of type to permit the systematic identification of tokens of that type. Without reliable ways of collecting data, we will ultimately not be able to integrate these communicative dimensions into our endeavours to explain language. These issues are non-trivial, but given our understanding of lan-guage, I believe it is essential that we confront them.

ReferencesArppe, Antti, Gaëtanelle Gilquin, Dylan Glynn, Martin Hilpert, M. and Arne

Zeschel 2010: Cognitive Corpus Linguistics: five points of debate on current theory and methodology. Corpora 5, 1-27.

Dąbrowska, Ewa 2004: Language, Mind and Brain: Some psychological and neu-rological constraints on theories of grammar. Edinburgh: Edinburgh Uni-versity Press.

Dąbrowska, Ewa 2008; The later development of an early-emerging system: The curious case of the Polish genitive. Linguistics 46, 629-650.

Glynn, Dylan Forthcoming: Emergent Categories: Quantifying semantic distinctiveness and similarity in usage. Contrast and Analogy in Lan-guage: Perspectives from cognitive linguistics. In: Marcin Grygiel, Barbara Lewandowska-Tomaszczyk (eds.) Amsterdam: John Benjamins.

Hopper, Paul 1987: Emergent grammar. In: Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society. 139-157.

Langacker, Ronald 1987: Foundations of Cognitive Grammar, Vol. 1. Theoretical prerequisites. Stanford: Stanford University Press.

Tabacaru, Sabina, Sheena van der Mark 2019: Crosslinguistic perspectives on the use and meaning of emoji in Asia and Europe. The Fifteenth Interna-tional Cognitive Linguistics Conference, 06-11 August 2019, Nishinomiya, Japan.

158 L aMiCuS 2019 no. 3


Karolina Krawczak2


The status of multimodal constructions in developing models and empirical testing- Overview of the debate

The debate on the status of multimodal constructions in developing models and empirical testing revolved around four general questions concerning (i) the relationship between language and other communicative systems such as bodily expression; (ii) the applicability of linguistic signs to describe lin-guistic behavior; (iii) the role of mental representations in understanding multimodal communicative events and the character of such representations; (iv) the impact of multimodality as a feature of any mode of communicative activity on researching and modeling language.

The debaters have approached these questions from three different per-spectives, specific to their immediate fields of interests, i.e., multimodality, psychology, and methodology for linguistics. This in itself affords a valuable take on the status of multimodal constructions in empirical research. Let us see what our three experts have to say on the matter, where their differences lie, and where their views converge.

Regarding the first question above, the debaters agree unanimously that language is by no means independent from other communicative systems. Dylan Glynn notes that even if the various structures in different semiotic systems may well be highly specialized, this does not exclude their inter-dependence. Similarly, he also observes that however complex the holistic


approach to language espoused in Cognitive Linguistics may seem, in sci-ence we should never lose sight of the true nature of our object of study; nor should we ever compromise on the adequacy of our descriptions and expla-nations. Barbara Dancygier points out that language inheres in embodied and enactive cognition as well as all forms of perceptions. Finally, Johanna Kißler proposes that emotion should also be seen as part and parcel of any linguistic behavior. In addition, she sees computational linguistics as playing an important role in solving some possible methodological or definitional problems that may arise from the holistic approach to language structure.

The second question referring to the efficacy of linguistic / semiotic signs in describing linguistic behavior was directly addressed by Dylan Glynn, who further specifies it to be about the inescapable circularity re-sulting from employing linguistic units for the description of language. He concludes that the use of signs is ultimately indispensable in any attempt to describe language structure, but if adequate quantitative methods are im-plemented, the findings thus obtained can be falsifiable. On a similar note, Johanna Kißler indicates that ensuring appropriate operationalization and testability of our research questions in investigating linguistic phenomena is of great importance, a standard that has long been adhered to in many data-driven approaches to language.

With respect to the question relating to mental representation, there is some general consensus between our experts. They all accept that we do need representations, though their understanding of them differs slightly. That said, Johanna Kißler and Dylan Glynn both agree that mental repre-sentations are not physically real in any sense. Rather, as noted by Dylan Glynn, they can be perceived as constituting analytical tools, while also being cognitive phenomena. More specifically, he proposes that in the spirit of the Usage-Based Model of language, as advanced by Hopper (1987) or Langacker (1987), representations are best understood as “generalizations across usage” or “complex form-meaning pairs”, which may well be mono- and multimodal. Johanna is more inclined to view representations as “dynamic coactivations of neural systems”. She admits that while studying such coactivations will not directly inform our research on interaction, it may shed light on some of the fundamental principles of human communicative behavior, be it verbal or bodily. On a slightly different note, Barbara Dancygier proposes to consider mental representations in terms of Fillmorean frames, accessible through and comprising verbal and nonverbal experience alike.

Finally, concerning the question of how to study linguistic behavior understood as a multifaceted and multimodal phenomenon empirically, the debaters have all recognized its relevance, particularly when it comes to such aspects as data. Barbara Dancygier states that while getting sufficiently rich data for the analysis of multimodal communication may indeed still be a

160 L aMiCuS 2019 no. 3

challenge, we should not let methodology that is available to us today limit our horizons. Instead we should let it “catch up” along the way. Dylan Glynn emphasizes the need to actively work on overcoming the methodological hurdles that researching multimodality brings. Among the most pressing problems to be considered here he sees the status of data in multimodal research and data collection. Addressing these questions is critical for the advancement of empirical research into multimodal communication.

162 L aMiCuS 2019 no. 3


Karolina Krawczak2


Status konstrukcji multimodalnych w tworzeniu modeli i badaniach empirycznych. Przegląd debaty

Debata na temat statusu konstrukcji multimodalnych w tworzeniu modeli i w badaniach empirycznych toczyła się wokół czterech ogólnych kwestii. Były to: (i) związek między językiem a innymi systemami komunikacji, ta-kimi jak mowa ciała; (ii) zakres zastosowania znaków językowych do opisu zachowań językowych; (iii) rola reprezentacji umysłowych w rozumieniu multimodalnych zdarzeń komunikacyjnych oraz charakter takich repre-zentacji; (iv) wpływ multimodalności jako cechy komunikacji na tworzenie modeli języka i badania empiryczne.

Debatujący podeszli do tych pytań z perspektywy właściwej dla ich za-interesowań naukowych, tj. multimodalności, psychologii i metodologii języ-koznawstwa. W ten sposób uzyskano wielowymiarowe spojrzenie na status konstrukcji multimodalnych w badaniach empirycznych. Przekonajmy się, co mają do powiedzenia nasi trzej eksperci; na czym polegają różnice ich stanowiskami, a gdzie ich poglądy są zbieżne.

W odniesieniu do pierwszego pytania debatujący jednogłośnie odrzucili twierdzenie, że język jest niezależny od innych systemów komunikacji. Dylan Glynn zauważył, że nie da się wykluczyć istnienia współzależności struktur w odmiennych systemach semiotycznych, nawet jeżeli są one wysoko wy-specjalizowane. Powiedział również, że choć może się wydawać, że języko-


znawstwo kognitywne ma charakter złożonego, holistycznego podejścia do języka, to jako językoznawcy nie powinniśmy tracić z oczu prawdziwej na-tury przedmiotu badań ani przeceniać adekwatności proponowanych opisów i wyjaśnień. Barbara Dancygier wskazała, że język jest nieodłączną częścią ucieleśnionego i enaktywnego poznania oraz wszelkich form postrzegania. Johanna Kißler stwierdziła, że emocje należy uznać za nieodłączną część zachowań językowych. Ponadto podkreśliła rolę lingwistyki komputero-wej w rozwiązywaniu problemów metodologicznych i definicyjnych, które mogą wynikać z zastosowania holistycznego podejścia do struktury języka.

Drugie pytanie dotyczyło zastosowania znaków językowych (semiotycz-nych) w opisie zachowań językowych. Dylan Glynn sprecyzował, że mowa tu o tzw. problemie błędnego koła w rozumowaniu wynikającym z zastosowa-nia znaków językowych do opisu języka. Doszedł on do wniosku, że użycie znaków jest niezbędne przy próbie opisania struktury języka. Jednocześnie przy zastosowaniu odpowiednich metod analizy ilościowej, uzyskane w ten sposób wnioski mogą zostać poddane falsyfikacji. Johanna Kißler zwróciła uwagę obecnych na to, że zapewnienie odpowiedniej operacjonalizacji i te-stowalności pytań badawczych postawionych w badaniu zjawisk językowych ma ogromne znaczenie. Co więcej, standard ten od dawna jest przestrze-gany w podejściach badawczych typu data-driven, czyli opartych na danych.

Jeśli chodzi o kwestię reprezentacji umysłowych, między naszymi eks-pertami zaistniał pewien konsensus. Wszyscy eksperci zgodzili się, że re-prezentacje umysłowe są niezbędne, choć nieco różnili się pod względem rozumienia, czym one są. Johanna Kißler i Dylan Glynn zgodzili się co do tego, że reprezentacje umysłowe nie są pod żadnym względem fizycznie realne. Dylan Glynn zaproponował, że powinny być postrzegane raczej jako pewne zjawiska poznawcze, będące jednocześnie narzędziem analitycznym. Mówiąc dokładniej, w modelu języka rozwiniętym przez Hoppera (1987) i Langackera (1987), reprezentacje rozumiane są jako „uogólnienia wysunięte na podstawie użycia” lub „złożone pary forma-znaczenie”, które mogą być mono- i multimodalne. Johanna Kißler uznała, że reprezentacje lepiej zde-finiować jako „dynamiczne współaktywacje w układzie nerwowym”. Przy-znała, że badanie takich współaktywacji nie ma bezpośredniego związku z badaniem komunikacji, może jednak rzucić światło na niektóre zasady rządzące ludzkim zachowaniem, w szczególności komunikacją werbalną i niewerbalną. W nieco innym tonie Barbara Dancygier zaproponowała, że reprezentacje umysłowe warto rozważyć w kategoriach ram Fillmore’a, do-stępnych poprzez doświadczenia werbalne i niewerbalne oraz obejmujących inne zachowania ludzi.

Ponadto debatujący zgodzili się, że szczególnie ważne jest pytanie dotyczące metodologii empirycznych badań zachowań językowych rozu-mianych jako zjawisko wieloaspektowe i multimodalne, zwłaszcza w kwe-

164 L aMiCuS 2019 no. 3

stii zbierania i analizy danych. Barbara Dancygier przypomniała, że choć zebranie wystarczającej liczby danych pozostaje wyzwaniem dla analizy komunikacji multimodalnej, nie powinniśmy pozwolić, aby dostępna meto-dologia ograniczała nasze horyzonty badawcze. Wprost przeciwnie, braki w tym zakresie powinna niejako nadrobić po drodze. Dylan Glynn podkreślił potrzebę nieustannej pracy nad pokonywaniem przeszkód metodologicz-nych związanych z badaniem multimodalności. Wśród najbardziej palących problemów wymienił status danych w badaniach multimodalnych i kwestię gromadzenia danych. Odpowiedź na te pytania ma kluczowe znaczenie dla postępu badań empirycznych w zakresie komunikacji multimodalnej.

Tłum. Anna Jelec

166 L aMiCuS 2019 no. 3 (3)

Reviews

Krystyna Waszakowa 2017: Kognitywno-komunikacyjne aspekty

słowotwórstwa. Wybrane zagadnienia opisu derywacji w języku

polskim. Warszawa: Wydział Polonistyki Uniwersytetu Warszawskiego.

230 s. ISBN: 978-83-65667-44-1.

Recenzent: Renata Przybylska

(Uniwersytet Jagielloński

w Krakowie)

Kognitywno-komunikacyjne aspekty słowotwórstwa. Wybrane zagadnienia opisu derywacji w języku polskim stanowi kolejne ważne dzieło w dorobku Krystyny Waszakowej, wybitnej badaczki współczesnej polszczyzny, specjalizującej się w badaniach nad jej podsystemem słowotwórczym. Autorka od lat udat-nie łączy strukturalistyczny paradygmat badawczy z wybranymi nurtami i metodami językoznawstwa kognitywnego. W najnowszym opracowaniu przyznaje się do inspiracji wypływających głównie z gramatyki kognityw-nej w ujęciu Ronalda Langackera, w mniejszym stopniu z teorii przestrzeni mentalnych Gillesa Fauconniera i Marka Turnera oraz koncepcji Jeleny Ku-briakowej. Dostrzega potrzebę „wzbogacenia dotychczasowych opisów sło-wotwórczych z jednej strony o wymiar pojęciowy, z drugiej zaś – o dyna-miczny ogląd słowotwórstwa w użyciu (w tekście, wypowiedzi), ogólniej: w komunikacji” (s. 15).

W tym celu nie porzuca instrumentarium badawczego strukturalizmu ani wypracowanej na jego gruncie terminologii, lecz uzupełnia je o metody i pojęcia wynikające z ujęć kognitywnych. Krystyna Waszakowa widzi bo-wiem w derywatach słowotwórczych struktury, w których „poprzez zwią-zek z innymi znakami językowymi i wpisanymi w ich znaczenie pojęciami


167Reviews

ujawnia się sposób konceptualizacji świata” (s. 35). I właśnie te wątki ko-gnitywnej teorii języka dotyczące konceptualizacji są dla badaczki najcen-niejsze – zakłada ona, że w derywatach są ukryte relacje między światem realnym a umysłem.

Materiałem językowym do swoich rozważań uczyniła przede wszystkim rozmaite innowacje słowotwórcze, gromadzone od lat dziewięćdziesiątych XX wieku, a pozyskiwane m.in. z Internetu. Dzięki temu omawiana praca może stanowić także wprowadzenie w problematykę związaną z nowymi jednostkami leksykalnymi współczesnej polszczyzny, ma też oczywisty wa-lor dokumentacyjny, tym bardziej, że poddawane interpretacji neologizmy zawsze są przytaczane wraz z całym kontekstem ich użycia.

W jednym z rozdziałów (rozdziale III) autorka omawia wybrane pro-cesy kognitywne odzwierciedlające się w strukturach słowotwórczych, takie jak: porównanie i obrazowanie obejmujące m.in. wyodrębnianie schema-tów, kategoryzowanie i ujmowanie danej sytuacji na różnych poziomach konkretyzacji, rozkładalność i kompozycjonalność derywatów wynikającą z rozpoznawania poszczególnych komponentów w wyrażeniu, pojmowanie jednej struktury w kategoriach innej (metaforyzacja i scalanie pojęć), ujmo-wanie sceny w kategoriach figury i tła, profilowanie ról archetypowych. Jak kognitywizm w ujęciu Langackera może oświetlić pewne znane skądinąd aspekty słowotwórstwa, Waszakowa przykładowo pokazuje, interpretując rodzinę wyrazów słowa komputer: komputerownia, komputerowiec, komputerowy, komputeryzować (się), komputeromania, komputerofobia, i uznając poszczególne derywaty za wynik profilowania podstawowych kategorii gramatycznych na tle przywoływanego jako baza bardzo ogólnego pojęcia „wszystko, co wiąże się z komputerem”, a zatem jako wynik tego aspektu obrazowania, który wprowadza rozróżnienie „figura/tło”. Z kolei zdolność kognitywna człowieka do wyodrębniania schematów poznawczych i rzutowania ich na rzeczywistość (w tym też na rzeczywistość językową) według Waszako-wej jest widoczna w tym, że używając języka, mówiący konstruują pewne schematy słowotwórcze i rzutują je na nowe struktury – dlatego rozumieją znaczenie neologizmów typu debatant, definicjant, reklamant, rekrutant, czyli ‘ten, kto debatuje, definiuje, reklamuje, rekrutuje’, bo mają one oparcie w schemacie wywiedzionym ze znanej serii słów: debiutant, dyskutant, pro-jektant itp. Oczywiście jedne schematy są bardzo silnie utrwalone, zako-rzenione w języku, inne słabiej; dotyczy to np. schematu wyłaniającego się z serii stosunkowo nowych zapożyczonych nazw osobowych aborter, happener, protester, surfer, zapper, które współcześnie dają się już interpretować jako derywaty słowotwórcze z nowym sufiksem -er.

Jedno z najważniejszych zagadnień opisu słowotwórczego języka, a mianowicie pojęcie pochodności słowotwórczej, Waszakowa ponownie poddaje krytycznej analizie w rozdziale IV. Najpierw dyskutuje z dotych-

168 L aMiCuS 2019 no. 3 (3)

czasowymi ujęciami strukturalistycznymi, po czym ukazuje na przykładach z jednej strony kognitywną funkcję motywacji słowotwórczej, z drugiej strony – funkcję komunikacyjną, zdeterminowaną przez rolę kontekstu użycia danego derywatu.

Bodajże najciekawsze partie książki (rozdział V) to interpretacje ko-gnitywne derywatów powstałych w akcie nominacji, m.in. takich: chirur-giczny w kontekście chirurgiczne cięcie, chirurgiczne naloty, znaczący ’wyko-nywany niezwykle precyzyjnie i skutecznie’; mongolizacja (życia, obyczajów); masakryczny (rok, tłok, obraz); konkurujące nominacje słowotwórcze: kremato-rium, spalarnia, spopielarnia, oparte na różnym profilowaniu obrazowanego pojęcia.

Część poświęconą słowotwórstwu w aspekcie komunikacyjnym stanowi rozdział VI. Tu z kolei teoretyczne założenia komunikatywnego podejścia Waszakowa stosuje do analizy wybranego przypadku: derywatów z czło-nami bio- i eko-.

Rozdział VII, poświęcony systematyzacji jednostek słowotwórczych w ujęciu kognitywnym, zawiera najpierw odwołanie do ujęć strukturalistycz-nych, a następnie omówienie głównych typów systematyzacji kognitywnych: (i) systematyzacji przez wyróżnienie prototypowych i peryferyjnych derywa-tów i formantów (np. prototypowe są w języku polskim derywaty sufiksalne, a inne są mniej lub bardziej oddalone od prototypu); (ii) systematyzacji przez umieszczenie derywatów w ramach kategorii radialnej; (iii) systematyzacji przez zastosowanie schematów słowotwórczych w duchu koncepcji Lan-gackera. Autorka zwraca uwagę, że systematyzacje kognitywne pozwalają rozwiązać niektóre dotychczasowe problemy językoznawstwa struktura-listycznego, np. problem granic między kategoriami słowotwórczymi czy kwestię hierarchii typów derywacji.

W końcowej partii swego dzieła Waszakowa analizuje zbieżności teorii Witolda Doroszewskiego z późniejszym modelem kognitywnym Langac-kera, a także porównuje klasyczną teorię onomazjologiczną Miloša Dokulila z ujęciem kognitywno-dyskursywnym. Poświęcone tym zagadnieniom oba rozdziały są bardzo cenne z punktu widzenia metanaukowego, dotykają bowiem rzadko podejmowanej kwestii zależności poznania naukowego nastawionego na poszukiwanie prawdy (obiektywnej?) od wyznawanego i tworzonego przez badacza modelu języka. Co bowiem jest realnością obiektywną, czy model języka także? Ponadto takie porównania, zesta-wienia koncepcji naukowych powstałych w różnym czasie i miejscu na ten sam temat pozwalają sprawiedliwie docenić wkład poszczególnych badaczy w poznanie natury języka, niezależnie od tego, czyja koncepcja została szerzej wypromowana i wprowadzona w obieg naukowy. Rodzi się też ref leksja, że zasiane kiedyś ziarna myśli mogą rozkwitnąć i wpłynąć na postęp w nauce czasem z dużym opóźnieniem, dopiero wtedy, gdy zaist-

169Reviews

nieją sprzyjające okoliczności. Nauka to wszak w pewnym stopniu sztuka wiecznych powrotów.

I tak wracając do teorii Dokulila, Waszakowa między innymi pokazuje, że w ujęciu kognitywnym brak np. podstaw do wyróżniania formacji trans-pozycyjnych wprowadzonych przez tego uczonego, ponieważ nominalizacja jest uznawana za zjawisko semantyczne, a nie tylko składniowe. Zarazem pojęcie bazy onomazjologicznej u Dokulila i różnych stopni oraz sposobów jej wyspecyfikowania zbliża się do pojęcia procesów profilowania w semantyce kognitywnej. Teoria derywacji Dokulila ma charakter czysto synchroniczny, proponuje statyczny opis słowotwórstwa w ramach systemu językowego bez wykraczania poza niego, bez rozważań o wpływie świadomości językowej mówiących, ich skojarzeń, działania kontekstu itp., natomiast słowotwór-stwo kognitywno-komunikatywne uwzględnia te czynniki, proponuje ujęcie dynamiczne, oparte na uzusie językowym, na odwołaniu się do konkretnego zdarzenia – użycia danego derywatu w danym kontekście. Autorce omawia-nej tu książki są równie bliskie oba podejścia. Jej analizy dowodzą możliwości twórczego pożenienia strukturalizmu z kognitywizmem.

Książka Waszakowej jest „gęsta” od nowych pomysłów interpretacyj-nych, z których każdy zasługiwałby na pogłębioną dyskusję, uznanie lub krytykę. Autorka doskonale panuje nad niezwykle szerokim spektrum za-gadnień, odkrywczo czyta literaturę dawniejszą, osadzając ją w kontekście najnowszych koncepcji naukowych. Jest też na polskim gruncie łączniczką tego, co cenne w językoznawstwie tworzonym na Zachodzie i na Wschodzie, w Rosji. Zwłaszcza znajomość badań rosyjskich (prace Jeleny Kubriakowej, Pietruchiny, Pozdniakowej, Ułuchanowa, Ziemskiej i innych) i przeszczepia-nie ich do polskiego językoznawstwa uważam za bardzo cenne.

Postawę naukową Krystyny Waszakowej wyróżnia otwartość na nowe koncepcje i chęć poważnej z nimi dyskusji, opartej na konkretnych badaniach i na merytorycznych argumentach. Pozostając strukturalistką, nie okopuje się ona w twierdzy jedynie słusznej dotychczas wyznawanej metodologii. Nie dołącza do obozu tych, którzy a priori odrzucają kognitywizm, nie zadając sobie trudu jego poznania. Przeciwnie – szuka podobieństw, szuka tego, co pozwoli jej spojrzeć na przedmiot badań z nowej, odświeżającej perspektywy. Dlatego warto polecić jej książkę zarówno kognitywistom, jak i badaczom o orientacji strukturalistycznej.

Kończący książkę rozdział przedstawia perspektywy badań porównaw-czych nad słowotwórstwem, prowadzonych w duchu kognitywistycznym. Rozdział ten można zadedykować młodemu pokoleniu lingwistów poszu-kujących swojego pola badań. Waszakowa jasno pokazuje, że wielką przy-szłość mają badania ukierunkowane na „ukazanie międzyjęzykowych różnic w zakresie konceptualizacji, sposobów obrazowania, zawartych w odpowiedni-kach leksykalnych, będących derywatami słowotwórczymi” (s. 199). Jej zdaniem

170 L aMiCuS 2019 no. 3 (3)

poprzez charakterystykę pojęć ujawniających się w strukturach słowotwórczych można bowiem odsłonić właściwy danemu językowi sposób widzenia świata.

Książka zawiera na końcu streszczenia w dwóch językach: rosyjskim i angielskim oraz cenny indeks przedmiotowy, w którym zebrano używane w pracy niektóre terminy oraz te wyrażenia opisowe, które odnoszą się do przedmiotu pracy. Nie jest to zatem indeks wszystkich terminów w pracy użytych. Co istotne, indeks ten poucza o istnieniu wciąż w polskiej termi-nologii kognitywnej licznych terminów synonimicznych, np. za tożsame uznano określenia: integracja, teoria integracji, stapiania, blending, teoria prze-strzeni mentalnych. Przy okazji chcę zwrócić uwagę na wysiłki Autorki, aby jak najtrafniej oddać terminologię językoznawstwa kognitywnego, jak wiadomo wywodzącą się z języka angielskiego. Jeden z pomysłów uważam za nietrafny, mianowicie skalkowanie ang. abstraction przez polski wyraz abstrakcja (s. 27), lepiej byłoby wprowadzić odpowiednik polski abstrahowanie, bo chodzi w tym terminie o czynność taką, że ktoś abstrahuje od czegoś, tworząc uogólniony schemat. Słowo abstrakcja w polszczyźnie nie ma znaczenia czynnościowego; zresztą abstrahowanie pojawia się na s. 45 pracy. Podobnie nietrafne jest od-danie terminu angielskiego elaboration (s. 37) przez polski wyraz rozwój – już lepsze byłoby rozwijanie, które ma oczekiwany tu wydźwięk czynnościowy, można bowiem powiedzieć rozwijanie tematu, ale nie np. *rozwój tematu.

Oczekiwałam także indeksu omówionych w pracy neologizmów słowo-twórczych, ale go nie znalazłam. Tymczasem taki indeks byłby pożyteczny zwłaszcza dla leksykografów rejestrujących innowacje słownikowe, a także dla czytelników zainteresowanych po prostu tym, co w języku nowe. W każ-dym razie to Krystyna Waszakowa utrwaliła w swoim dziele m.in. takie nowe słowa, jak: studniówka ‘sto dni nowego rządu’, gejrlin, haktywista, nikiforyzm, ekościema, podpisywactwo i wiele innych smakowitych rodzynków, odkrytych przez nią „w cieście” polskiego słowotwórstwa.

171Reviews

The core ideas in William Edmondson’s contribution The Sequential Imper-ative: General Cognitive Principles and the Structure of Behaviour have a long history. As a matter of fact, they go back to Edmondson’s (1986) paper in Language and Communication (or likely even further). Already in this paper, he suggested that one of the core functions of grammar – or linguistic cognition – is linearization of cognitive entities into behavior, and de-se-quencing of necessarily linear behavior into cognitive entities. More than twenty years later, these ideas were embedded into the more general theory of the “sequential imperative” put forth in Edmondson (2010). The book to be talked about here represents a more in-depth discussion of the theory introduced in the latter article and relates its propositions to several fields of cognitive science.

The book is structured into three main parts, rounded with an intro-duction and an afterword. The first part, “The sequential imperative and the functional specification of the brain”, is about the central propositions of the book referred to as generalized cognitive principles. Edmondson elabo-rates on these principles drawing on anecdotal evidence as well as evidence from the cognitive and linguistic literature. The main idea (the “sequential

William H. Edmondson 2017: The Sequential Imperative. General

Cognitive Principles and the Structure of Behaviour. (Value Inquiry

book series.) Leiden, Boston: Brill Rodopi. ISBN: 978-90-04-34289-7 (Pb),

978-90-04-34299-6 (E-book).

Reviewed by Andreas Baumann

(University of Vienna)


172 L aMiCuS 2019 no. 3 (3)

imperative”) is that the task of cognition is to transform (de-sequence) be-havior into atemporal entities (representations) during perception and vice versa during production of behavior. Moreover, sequentiality of behavior is presented as an inevitable consequence of physiology.

The main theme of the second part (“Serving the sequential imperative”) takes the concepts formulated in the first part to provide a structural account of language (with a focus on phonology). Edmondson presents a structural model which merges several linguistic domains (phonology, syntax, prag-matics at least). Each additional bit of structural (and atemporal) information inf luences the perception (and production) of sequential linguistic input (output) at the same time.

The third part is called “Behaviour and evolution – on and off planet” and collects several ideas that build on the proposed model but go into slightly different directions. On the one hand it addresses issues of cognitive management (the management of learning and attention and in which way this is related to the sequential imperative). On the other hand, it elaborates on evolutionary questions. Finally, Edmondson closes with a discussion of the (potential) universality of the sequential imperative and a thought ex-periment about the extent to which the proposed principles may apply to all known (or perhaps also unknown) species.

The style in which the book is written is unorthodox. The book is, as Edmondson points out already in the introduction, meant as “a set of notes for bigger projects to come” (p. IX), and this is also ref lected in its structure: sections and subsections within each chapter, often encompassing just a few paragraphs, are mostly not linked explicitly but numbered hierarchically, with a hierarchical depth of up to six levels. The intention of the author is to facilitate jumping from one thought to the next without grappling with laborious transitions and literature reviews.

While this may be true, the lack of subsection headings and signposting sometimes makes it challenging to follow the main argument in each larger section. The reader finds himself jumping back and forth to check which hierarchical level some paragraph one has just read belongs to. This is very much like organizing and reorganizing post-it notes on a board into clusters. So, reading this book is a comparably interactive task and demands a good amount of concentration. Perhaps this experience is intended by the author as well. A second consequence of this style is that the book almost reads like a mathematical text in which every single definition, proposition and example is numbered. Indeed, Edmondson often refers back to particular paragraphs while advancing his thoughts. Repetitions of the key propositions (labeled as GCP1 to GCP8) are very much appreciated.

The book covers many areas relevant to cognition and raises multiple questions. In what follows, I will focus on two main thoughts f leshed out in

173Reviews

the book. The first one is about the nature of cognitive representations. The second one is about language evolution.

As stated before, one of the central propositions in this book is that cognitive representations are strictly atemporal (this proposition is labeled GCP2: “Cognitive entities are [… i]nherently atemporal”, p. 23). This is moti-vated and discussed in depth in chapter 2 (mainly 1(2).5-6). The role of cogni-tion, according to Edmondson, is to de-sequence behavior (which is always sequential) into atemporal cognitive representations (essentially the second part of GCP3: “Perception is de-sequencing”, p. 23). Representations are not supposed to be static in the sense that they cannot change, however. Rather they are suggested to be atemporal entities which lack any inherent dynamic properties.

The proposition is particularly interesting as it finds a clear correspon-dence in artificial neural networks (ANNs; Schmidhuber 2015; LeCun et al. 2015) studied in computational science and mathematics. ANNs can be used to model mappings from sequential objects like speech, sequences of actions or time series, to static – albeit highly dimensional – vector representations. This is often done with recurrent neural networks employing particular types of cells that can store bits of information during processing a sequence (e.g. long short-term memory cells; Gers et al. 1999). The output of these cells is then mapped to multi-dimensional representations in deeper layers of the network (possibly to be further processed for doing specific tasks). Each dimension carries one (potentially condensed) aspect of information. Im-portantly, the ordering of dimensions does not matter at this point: they are, as Edmondson would probably put it, “atemporal” entities.

Thus, two of the central propositions (GCP2 and GCP3) are conceptu-ally supported by computational research. On the neurophysiological level, things are less clear. In section 1(2).6, Edmondson defends his propositions against potential neurophysiological criticism, mainly drawing on literature from the 1960s. The crucial question is this: is there neurological evidence for temporal patterns of cognitive representations? The author discusses related issues like temporal patterning in auditory studies (1(2).6.2.1.2), neu-ronal speed of response (1(2).6.2.1.3), and brain waves (1(2).6.2.1.4). It is then concluded that (with the exception of the latter) “it makes no sense to seek to align measures of temporal activity and patterns measured in the brain with temporal structure in behaviour and perception” (p. 37).

I would like to contest this conclusion by referring to hippocampal replay phenomena. In a couple of studies it was demonstrated that tem-poral patterns of brain activity are replayed in the hippocampus during subsequent rest or sleep (see e.g. Frankland and Bontempi 2005, for an overview). This was shown in humans (Huber et al. 2004) and other animals. More specifically, for example, Lee and Wilson (2002) have demonstrated

174 L aMiCuS 2019 no. 3 (3)

that neuronal activation patterns of rats associated with spatial movement are replayed during slow wave sleep, thereby retaining activation ordering. The behavioral sequence was compressed in time by a factor of 20. They take their finding as evidence for the formation of long-term memories of sequential events.

The crucial point here is that replay is broken down to the neural level and that there is a clear alignment of temporal neuronal activity (albeit com-pressed) and temporal patterns of behavior. This correspondence is too tight to be discarded as being not relevant. Do these temporal patterns on the neu-ronal level correspond to temporally structured representations? Or are they mapped – through the proposed de-sequencer perhaps – to atemporal cog-nitive representations conceptually similar to static vector representations in (computational) artificial neural networks discussed before? The answer is not clear; in any case, the book fulfils its purpose of initiating interesting discussions and thoughts.

Let us turn to the second topic I want to address. Chapter 8 (“Issues in evolution and language”) focuses on discussing the evolutionary conse-quences of GCP3 (“Behaviour is sequencing; Perception is de-sequencing”, p. 141). Interestingly, it seems that Edmondson conceptualizes cultural evo-lution (e.g. the evolution of languages) as epiphenomenon of the evolution of a cognitive device. For example, right in the introduction of Chapter 8 Edmondson states the following: “[o]ne particular misconception we need to sweep away is that ‘a behaviour’ evolves – it doesn’t; the behavioural ap-paratus evolves” and “[it] is behaviour which changes, because the potential for sequencing and de-sequencing evolves” (both p. 140; emphasis as in the original). Based on these assertions it seems that it is solely biological evo-lution driving the evolution of cognition (the cognitive apparatus) which in turn drives behavioral change (such as linguistic change). Referring to Jackendoff (2002), Edmondson continues to argue that a behavior “such as language does not evolve in the sense of interest to us here. [… T]hey are already languages – and it is how we got that far that is of interest here” (p. 142). I would like to question this argument.

To begin with, there is quite some research on cultural evolution not referred to in this book, starting with Dawkins’ (1976) and Hull’s (1988) work on Generalized Darwinism (see also Aldrich et al. 2008). Basically, they argue that it is the same set of evolutionary mechanisms which operate in biology and the cultural domain. The theory of Generalized Darwinism was extended later to the linguistic domain particularly by Croft (2000) and Ritt (2004), among others. When discussing the history of the design of screws (blade to cross-head), Edmondson, for example, argues that “[t]alk of evolution in the design of screws is loose talk [and t]alk of ‘evolution of language’ is loose talk in the same way [… because] the focus is on the superficial ‘observables’

175Reviews

and not on the underlying capability” (p. 145). This would not be supported by Generalized Darwinism, since in this paradigm evolution does not re-quire any underlying capability but just (cultural or biological) replicators.

Now, research on cultural and in particular linguistic evolution has shown that studying cultural change is indeed relevant to understanding cognition. This is because by investigating the history and change of cultural replicators one can identify cognitive biases and constraints, i.e. properties of cognition (Kirby, Dowman, and Griffiths 2007; Evans and Levinson 2009). So, studying behavioral change can shed light on how the cognitive appara-tus evolved, to begin with.

Finally, Edmondson seems to assume a unidirectional relationship be-tween biology and behavior when he says that “complexities in the processes of sequencing and de-sequencing […] change as brains evolve” (p. 161). This may not be exclusively so. For instance, Ferretti and Adornetti (2014: 316) argue that “[i]f it is true that external scaffoldings [e.g. language] are used to reduce the computational cost of the brain, it is also true that they represent a new commitment for the brain”. That is, linguistic behavior represents a selection pressure in the biological evolution of the brain.

As Edmondson states repeatedly, the book understands itself as a prole-gomenon to a discussion rather than a finalized theory (interestingly enough, the same term that Edmondson used in his 1986 kick-off paper). The fact that Edmondson manages to let the reader’s thoughts f low to yield new ideas and to provide novel views and angles on certainly well-studied phenomena is probably the biggest strength of this book.

ReferencesAldrich, Howard E, Geoffrey M Hodgson, David L Hull, Thorbjørn Knud-

sen, Joel Mokyr and Viktor J Vanberg 2008: In defence of generalized Darwinism. Journal of Evolutionary Economics 18(5), 577-596. doi:10.1007/s00191-008-0110-z.

Croft, William 2000: Explaining Language Change: An Evolutionary Approach. (Longman Linguistics Library). Harlow, England and New York: Longman.

Dawkins, Richard 1976: The Selfish Gene. New York and Oxford: Oxford Uni-versity Press.

Edmondson, William H. 1986: Issues in linearization: Prolegomena for a general theory of communication. Language and Communication 6(4), 225-266. doi:10.1016/0271-5309(86)90013-3.

176 L aMiCuS 2019 no. 3 (3)

Edmondson, William H. 2010: General cognitive principles: The structure of behaviour and the sequential imperative. International Journal of Mind, Brain & Cognition 1(1), 7-40.

Evans, Nicholas and Stephen C. Levinson 2009: The myth of language univer-sals: Language diversity and its importance for cognitive science. The Be-havioral and Brain Sciences 32(5), 429-494. doi:10.1017/S0140525X0999094X.

Ferretti, Francesco and Ines Adornetti 2014: Biology, culture and coevolution: Religion and language as case studies. Journal of Cognition and Culture 14(3-4), 305-330. doi:10.1163/15685373-12342127.

Frankland, Paul W. and Bruno Bontempi 2005: The organization of recent and remote memories. Nature Reviews Neuroscience 6(2), 119-130. doi:10.1038/nrn1607.

Gers, Felix A., Jürgen Schmidhuber and Fred Cummins 1999: Learning to forget: Continual prediction with LSTM. ICANN ‘99, 850-855.

Huber, Reto, M. Felice Ghilardi, Marcello Massimini, Giulio Tononi 2004: Lo-cal sleep and learning. Nature 430(6995), 78-81. doi:10.1038/nature02663.

Hull, David L. 1988: Science as a Process: An Evolutionary Account of the Social and Conceptual Development of Science. Chicago: The University of Chi-cago Press.

Jackendoff, Ray 2002: Foundations of Language. Brain, Memory, Grammar, Evo-lution. Oxford: Oxford University Press.

Kirby, Simon, Mike Dowman, Thomas L. Griffiths 2007: Innateness and cul-ture in the evolution of language. Proceedings of the National Academy of Sciences 104(12), 5241-5245. doi:10.1073/pnas.0608222104.

LeCun, Yann, Yoshua Bengio and Geoffrey Hinton 2015: Deep learning. Na-ture 521, 436-444.

Lee, Albert K., Matthew A. Wilson 2002: Memory of sequential experience in the hippocampus during slow wave sleep. Neuron 36(6), 1183-1194. doi:10.1016/S0896-6273(02)01096-6.

Ritt, Nikolaus 2004: Selfish Sounds and Linguistic Evolution: A Darwinian Ap-proach to Language Change. Cambridge: Cambridge University Press.

Schmidhuber, Jürgen 2015: Deep Learning in neural networks: an overview. Neural Networks 61, 85-117.

177Reviews

Goldberg, Adele E. 2019. Explain Me This. Creativity, Competition,

and the Partial Productivity of Constructions. Princeton: Princeton

University Press. 195 pp. ISBN-13: 978-0-691-17426-6

1. The explain-me-this puzzleAdele E. Goldberg is one of the leading linguists of our time. Her research on argument structure constructions (Goldberg 1995) has been a cornerstone reference in the development of Construction Grammar as a linguistic the-ory. Her later work on constructional generalizations (Goldberg 2006) has taken up issues such as the usage-based nature of linguistic knowledge, language learning, and the network-like organization of constructions, and has placed them squarely at the center of current research in Construction Grammar. Adopting a wide range of experimental and corpus-based meth-ods, her studies have addressed pertinent questions in language acquisition (Wonnacott et al. 2012), neurolinguistics (Allen et al. 2012), syntax (Ambridge and Goldberg 2008), and research on metaphor (Citron and Goldberg 2014), amongst other areas. It is thus fair to say that the bar is set high for Gold-berg’s new monograph, which synthesizes her recent work on constructional productivity, and which proposes a solution to a phenomenon that she labels the explain-me-this puzzle.

The central question that the book tries to answer is the following. How is it possible that speakers of a language accept certain utterances that they

Reviewed by Martin Hilpert

(Neuchâtel University)


178 L aMiCuS 2019 no. 3 (3)

have never heard before as completely idiomatic, while they reject other ut-terances as simply impossible and ungrammatical? To take a concrete exam-ple, an utterance such as Vernon tweeted to say she doesn’t like us shows the verb tweet in the syntactic environment of a following infinitive clause. Native speakers of English will find this usage of the verb fully acceptable, despite its low frequency of occurrence. By contrast, an utterance such as She con-sidered to say something, which features the verb consider in the same syntactic context, will be judged as decidedly unidiomatic. This example is no isolated case, but ref lects a more general phenomenon, which Goldberg calls partial productivity. Even for highly productive constructions, certain combinations of verbs and syntactic contexts simply do not match, and this includes the combination of the verb explain and the ditransitive construction that gives the explain-me-this puzzle its name.

The answer that Goldberg develops in the book is based on two ideas for which she uses the terms coverage and statistical preemption. The notion of coverage relates to the mutual similarity that different instantiations of a construction exhibit. Constructions with even coverage have instances that are highly similar to each other. Constructions with a more uneven coverage may consist of several clusters of instances that differ substantially from one cluster to another. Statistical preemption is an idea that relates to the competition between constructions. If a given form-meaning pair is strongly entrenched, it may prevent speakers from producing an alternative form with the same meaning. Coverage and statistical preemption conspire in the explain-me-this puzzle. Goldberg’s book discusses how the two are re-lated and how their respective effects can explain the partial productivity of grammatical constructions.

2. The chapters of the bookGoldberg develops her overall argument in eight chapters. An introduction lays out the explain-me-this puzzle and states the basic theoretical concepts that underlie the discussion. These concepts are labeled as the CENCE ME principles. The acronym stands for the following ideas: C – constructions, E – expressiveness, N – new information and old information, C – compe-tition, E – efficiency, M – memory, and E – error-driven learning. Taken together, the CENCE ME principles add up to a usage-based view of language that adopts Construction Grammar as its theoretical framework and that incorporates psychological insights on the nature of learning and memori-zation alongside functional linguistic principles. The chapter also presents the reader with an updated definition of constructions. Going beyond her earlier uses of the term, Goldberg characterizes constructions as clusters of

179Reviews

memory traces within a high-dimensional conceptual space, which ranges over formal, functional, and contextual variables.

Chapter 2 takes a step back from the main topic of constructions and instead focuses on words. The reason for this slight detour is that the ac-quisition of words and their meanings holds a number of implications for the explain-me-this puzzle. Specifically, words evoke an array of meanings that is abstracted away from individual contexts of use. They are routinely used for new ideas when no conventionalized alternatives are available. For example, the English word friend can be used to refer to acquaintances on social media, even when these are not strictly speaking friends. During the process of language acquisition, the use of words is characterized by initial overgeneralization and subsequent narrowing. Goldberg argues that all of these aspects show up in similar form in the acquisition and use of more abstract pairings of form and meaning.

To readers who are familiar with the main ideas of Construction Gram-mar, chapter 3 will serve as a basic review. Goldberg summarizes the concepts that form the bedrock of her research program, including the observations that syntactic forms carry meaning, that clashes between constructional and lexical meaning give rise to coercion effects, that constructions are organized in a vast network, and that their usage is constrained by factors that pertain to all levels of linguistic structure, from phonology to information packaging.

Chapters 4 and 5 address the two ideas that are central to the explain-me-this puzzle, namely coverage and statistical preemption. Chapter 4 fo-cuses on the issue of coverage. The discussion takes as its starting point the insight that speakers operate with highly detailed memories of language use that are continually updated (Bybee 2010). Following from this general idea is the notion that new instances of language use are related to existing clusters of linguistic memories, so that categories are formed and general-izations emerge. Utterances that are perceived as similar to utterances that were processed before are processed as members of a category. Since no two utterances are exactly identical, the memorized instances of language use form a category that exhibits variability across the dimensions of sound, morpho-syntactic structure, meaning, and situational context. Goldberg describes this variation in terms of a hyper-dimensional space. Any given construction is represented by memory traces of concrete usage events that form a contiguous cloud of points within such a hyper-dimensional space. What Goldberg calls coverage (Suttle and Goldberg 2011, Perek 2016, Barak and Goldberg 2017) relates to the distribution of these points in the hyper-dimensional space. Are the points distributed evenly, so that the distances from one point to the next are relatively similar, or are the points distributed more unevenly, so that some of the points cluster tightly together while others are further away from each other? It is useful to draw an analogy here. Even

180 L aMiCuS 2019 no. 3 (3)

coverage would correspond to a landscape that is dotted with small villages. From any point in that landscape, the next village is immediately in reach. Uneven coverage would correspond to a landscape that is completely empty, except for a metropolis and a handful of large cities. Goldberg argues that the productivity of a construction is related to coverage. New instances of a construction are deemed acceptable if they fall into an area that is evenly covered. For the landscape analogy, this means that new settlements are possible anywhere in an area with many small villages, while an empty land-scape halfway between two large cities will remain uninhabited.

Chapter 5 turns to the issue of statistical preemption. The basic prem-ise of statistical preemption is that speakers form generalizations over sets of constructions that are comparable with regard to their meanings, as for instance the ditransitive construction (John gave Mary the book) and the prep-ositional dative construction (John gave the book to Mary). Since the two con-structions express meanings that are broadly similar to one another, it makes sense to view them as competitors. If a speaker wants to express an event of giving, one of the two constructions can be chosen to verbalize that event. It is further assumed that speakers memorize the lexical elements that they hear in these constructions. For the verb give and any other verb that occurs in the two constructions, speakers keep track of how often the verb is used in each member of the construction pair. Some verbs show interesting asymme-tries. For example, the verb recommend regularly occurs in the prepositional dative construction (I recommended the hotel to all my friends), whereas its use in the ditransitive construction is rare to non-existent. Speakers thus per-ceive a statistical imbalance in the frequency of recommend across the two constructions, and they interpret that imbalance as meaningful. If a lexical item rarely or never appears where it would be expected with a certain base frequency, then speakers conclude that there is a grammatical constraint. In psychological terms, statistical preemption crucially draws on mismatches between speakers’ expectations and the language use that they actually hear. The surprise that speakers experience when they anticipate one construction and end up hearing another gives rise to what Goldberg terms error-driven learning. To give an example, a young child may expect to hear a transitive sentence such as The magician disappeared the rabbit as a description of an event it has just witnessed. If the speaker instead produces The magician made the rabbit disappear, that expectation is not fulfilled, and the surprisal value of the alternative expression that was actually experienced cognitively strengthens that construction, simultaneously weakening the associative link between the verb disappear and the transitive construction.

With the main aspects of the argument in place, chapter 6 rounds out the discussion by addressing a number of central insights from the us-age-based literature on language learning. Young children are known to

181Reviews

use argument structure constructions conservatively, which is an idea that Tomasello (2003) has called the verb island hypothesis. Young children also tend to simplify their language output in the face of variability along di-mensions that are hard to identify. The chapter further discusses transfer effects in L2 learners, who draw on generalizations that are present in their native languages. Goldberg argues that pre-existing linguistic knowledge will make it harder for L2 learners to draw on statistical preemption. A Ger-man-speaking L2 learner of English may not notice the conspicuous absence of utterances such as Explain me this because the L1 equivalent Erklär mir das is actually part of her active linguistic repertoire.

As the explain-me-this puzzle has been the subject of intensive research in different theoretical approaches to language, chapter 7 acknowledges some of the alternative proposals that have been made over the years. Goldberg discusses early work by Pinker (1989), who tried to model the combinatory restrictions of the ditransitive constructions and its verbs in terms of verb classes and rules that apply to those rules. Within the last thirty years, an extensive body of empirical research has provided insights into the proba-bilistic nature of linguistic knowledge, which is largely incompatible with accounts that draw on crisp, categorical rules. A more recent approach from a generative background is offered by Yang (2016), who proposes two statistical principles that aim to model how children learn to apply and to constrain productive schemas. These two principles are termed the Tolerance Principle, which sets a limit for the number of exceptions that a schema may have, and the Sufficiency Principle, which defines a minimum ratio of cases that must instantiate the schema, given a baseline of possible cases. Goldberg is highly critical of the implementation of these principles, arguing that they are partly ill-defined and that they require an unrealistically large number of cases for any schema formation to occur. Yet another approach that Goldberg reviews comes from within the cognitively oriented literature and draws on mech-anisms that are rather similar to the process of error-driven learning that Goldberg sees at work in statistical preemption. Stefanowitsch (2008) and Ambridge (2013), amongst others, have argued that the surprisal of hearing The magician made the rabbit disappear instead of The magician disappeared the rabbit does not need to depend on competition between functionally similar constructions. What they propose is that this effect can result from a mere frequency asymmetry between a verb in a given construction and that verb in language use in general. Given that disappear is a fairly frequent word, its non-use in the transitive construction is conspicuous, so that speakers inter-pret that non-use in terms of a grammatical constraint. Stefanowitsch uses the term negative entrenchment for this idea, Goldberg calls it conservatism via entrenchment. Direct behavioral comparisons of the two concepts yield mixed results (Ambridge et al. 2012, Ambridge and Blything 2015). Goldberg

182 L aMiCuS 2019 no. 3 (3)

ends her discussion by claiming that the traditional idea of a trade-off be-tween productive rules and memorized instances is actually a fallacy. The empirical evidence that has become available suggests that the two are in fact positively correlated. The more instances are memorized, the more likely is the formation of a schema. The more strongly entrenched a schema is, the more likely new instances will be formed and memorized.

Chapter 8 provides a brief summary of the previous chapters and out-lines several topics for further research. Goldberg notes the importance of typological studies (Croft 2001), encourages the exploration of conversational data for research in Construction Grammar, and she highlights the potential that computational modeling (van Trijp and Steels 2012) has for new insights.

3. Concluding remarksExplain Me This is an insightful synthesis of Adele Goldberg’s recent work on English argument structure constructions. The book develops its main theme by guiding the reader through a wealth of experimental and corpus-based studies that shed light on the partial productivity of constructions. Goldberg argues her case very carefully, and her discussion of coverage and statistical preemption presents these issues in their mutual contexts, which lets the reader fully appreciate the bigger picture and the important connections between these issues. The book succeeds beautifully in providing a bird’s-eye-view of the relevant questions, which makes it especially rewarding for readers who are already familiar with parts of Goldberg’s research.

There are, of course, also a few critical points. The first of these relates to the title of the book itself: Explain Me This. Creativity, Competition, and the Partial Productivity of Constructions. The two main issues of the book, cover-age and statistical preemption, do not feature as such in that title. Instead, they are rendered as creativity (= coverage) and competition (= statistical preemption). While this choice may help to make the book more marketable to a lay audience, it is confusing for readers with a background in linguistics.

There are other aspects that make it difficult to say who the targeted readership is. In the preface, the book is presented as an accessible introduc-tion for students, teachers, and researchers. The first three chapters can in fact be read as a short introduction to Construction Grammar. What follows, however, is a dense survey of empirical research, paired with the develop-ment of a theoretical argument. The discussion presupposes a fair amount of knowledge about linguistic theory and methodology, including Bayesian formulas. The chapters thus vary greatly in their accessibility, and students and teachers may find themselves struggling with the text.

With regard to the overall organization of the book, it might have been useful to arrange the sequence of chapters in a slightly different way, in

183Reviews

particular concerning chapters 6 and 7. Chapter 6, which discusses construc-tional research on first language acquisition, builds on the overview of core concepts in Construction Grammar in chapter 3, which would have provided a useful direct context. In chapter 7, Goldberg surveys alternative accounts of the explain-me-this puzzle that struggle with problems that chapters 4 and 5 have already solved. The chapter would have had a more useful function early on in the book, where it could have motivated why a new account is needed. It is simply not clear why chapters 6 and 7 are presented as afterthoughts to the main argument.

Lastly, the book is very selective in its presentation of relevant earlier re-search. This is acknowledged in the preface, and many readers will of course appreciate that Goldberg’s discussion is brief and to the point. Nonetheless, some of the more general background information could have been sacri-ficed to make room for issues of more direct relevance, as for example the model of constructional knowledge that is presented by Boas (2003), who approaches the explain-me-this puzzle from the perspective of frame seman-tics. In the context of statistical preemption, also the corpus-based work by Stefanowitsch and Gries (2003) would have merited a more detailed review.

It is clear that none of these points take anything away from the out-standing scientific merit of the book. With Explain Me This, Goldberg once again leads the field of Construction Grammar into a new and exciting area of research, which is a remarkable achievement.

ReferencesAllen, Kachina, Francisco Pereira, Matthew Botvinick, Adele E. Goldberg

2012: Distinguishing gammatical constructions with fMRI pattern analysis. Brain and Language 123, 174-182.

Ambridge, Ben 2013: How do children restrict their linguistic generaliza-tions? An (un-) grammaticality judgment study. Cognitive Science 37(3), 508-543.

Ambridge, Ben and Adele E. Goldberg 2008: The island status of clausal com-plements: evidence in favor of an information structure explanation. Cognitive Linguistics 19(3): 349-381.

Ambridge, Ben and Ryan Blything 2015: A connectionist model of the retreat from verb argument structure overgeneralization. Journal of Child Lan-guage 43(6), 1245-1276.

Ambridge, Ben, Julian M. Pine, Caroline F. Rowland, Franklin Chang 2012: The roles of verb semantics, entrenchment, and morphophonology in the retreat from dative argument-structure overgeneralization errors. Language 88(1), 45-81.

184 L aMiCuS 2019 no. 3 (3)

Barak, Libby and Adele E. Goldberg 2017: Modeling the partial productivity of constructions. AAAI. Stanford, CA, 131-138.

Boas, Hans C. 2003: A Constructional Approach to Resultatives. Stanford: CSLI Publications.

Bybee, Joan L. 2010: Language, Usage, and Cognition. Cambridge: Cambridge University Press.

Citron, Francesca and Adele E. Goldberg 2014: Social context modulates the effect of hot temperature on perceived interpersonal warmth: a study of embodied metaphors. Language and Cognition 6(1), 1-11.

Croft, William 2001: Radical Construction Grammar. Syntactic Theory in Typo-logical Perspective. Oxford: Oxford University Press.

Goldberg, Adele E. 1995: Constructions. A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press.

Goldberg, Adele E. 2006: Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press.

Perek, Florent 2016: Using distributional semantics to study syntactic pro-ductivity in diachrony: A case study. Linguistics 54(1), 149-188.

Pinker, Steven 1989: Learnability and Cognition: The Acquisition of Argument Structure. Cambridge, MA: MIT Press.

Stefanowitsch, Anatol 2008: Negative entrenchment: A usage-based approach to negative evidence. Cognitive Linguistics 19(3), 513-531.

Stefanowitsch, Anatol and Stefan Th. Gries 2003: Collostructions: Investi-gating the interaction of words and constructions. International Journal of Corpus Linguistics 8, 209-243.

Suttle, Laura and Adele E. Goldberg 2011: Partial productivity of construc-tions as induction Linguistics 49(6): 1237-1269.

Wonnacott, Elizabeth, Jeremy Boyd, Jennifer Thomson, Adele E. Goldberg 2012: Input effects on the acquisition of a novel phrasal construction in five year olds. Journal of Memory and Language 66, 458-478.

van Trijp, Remi and Luc Steels 2012: Multilevel alignment maintains language systematicity. Advances in Complex Systems 15(3-4), 1-30.

Yang, Charles 2016: The Price of Linguistic Productivity. How Children Learn to Break the Rules of Language. Cambridge, MA: MIT Press.

185Reviews

As the reader compares the title of Peeters’s edited volume with that of the series in which it appeared (Routledge Focus), they may experience an ap-parent dissonance: the former suggests a broad coverage across constructs, “languages, cultures, and epochs”, the latter narrows down that scope into a focalised core. Yet, the mismatch is only superficial: if the series zooms in onto a manageable focus, the volume follows suit in that it presents the essence of a theoretical, methodological, and descriptive framework, along with a few applications of it, albeit without claims to the comprehensibility of the picture. Indeed, the very idea of a series of this kind is commendable: to present an issue in a relatively short volume but provide more in-depth coverage that one can find, for example, in Oxford Basics. A volume in this Routledge series should provide an overview, synopsis, or introduction that allows one to launch further inquiry or begin one’s own research – such is also the design and aim of this book.

The volume has a dual focus: on the theoretical side, it presents the latest developments in the Natural Semantic Metalanguage enterprise. In terms of analysis, it selects, from a plethora of semantic domains that NSM research-ers investigate, “heart- and soul-like constructs” in a few languacultures.

Bert Peeters (ed.) 2019: Heart- and Soul-Like Constructs across Lan-

guages, Cultures, and Epochs. (Routledge Focus: Routledge Studies in

Linguistics.) New York and London: Routledge. viii, 148 pp. ISBN: 978-

138-74530-8 (Hb), 978-1-315-18067-0 (E-book).

Reviewed by Adam Głaz (Maria Curie-Skłodowska

University [UMCS], Lublin, Poland)


186 L aMiCuS 2019 no. 3 (3)

The book consists of the Editor’s introductory chapter and four analytical inquiries from other authors: Yuko Asano-Cavanagh deals with the Japanese inochi (roughly speaking, ‘life’) and tamashii (approx., ‘soul’), Deborah Hill with anoa ‘spirit’, agalo ‘ancestor spirit’, and zabe ‘body and spirit’ in Longgu (an Oceanic language of Guadalcanal, Solomon Islands), Chavalin Svetanant with chai ‘heart’ in Thai, and Colin Mackenzie with Old Norse-Icelandic hugr ‘mind, heart, courage’. The book presents the current state-of-the-art of a consistent body of research and so is a valuable resource both for uninitiated beginners to NSM and for those, like myself, who are not thoroughly up-to-date with the framework’s latest developments.

The Editor’s control of the volume’s content is impressive: apart from the introductory chapter, he provides a postscript to each of the other con-tributions, noting their idiosyncratic features and the range of possibilities that the NSM framework allows for, such as Deborah Hill’s use of Minimal English (and its relationship to the NSM), Chavalin Svetanant’s explications in Thai, or Colin Mackenzie’s use of cultural scripts apart from standard NSM explications. In-between the lines, Peeters also provides the reader with the “geography” of the NSM research, when on p. 8 he refers to “linguists in Aus-tralia and elsewhere”. He could have, however, been a little more careful on p. 3, where it is suggested, perhaps inadvertently, that the main person behind the framework is Cliff Goddard. As far as current activity is concerned, this may well be correct but, with all due respect, is erroneous in the historical sense. This “error” is “corrected” on p. 11, where Goddard’s proposals are said to be “explicitly based on Wierzbicka’s earlier work”.

One of the book’s major goals is to show the diversity and richness of languacultures and in so doing to combat the hegemony of English.1 How-ever, in trying to achieve that goal, the Editor seems to inadvertently fall foul of it himself. From the discussion of NSM on p. 8, the reader can deduce that the metalanguage comes primarily in English but is fully translatable into other languages. This is not what the NSM was conceived to be: instead of “other NSMs” that Peeters evokes, Anna Wierzbicka talks about other, “lan-guage-specific versions of the same, universal Natural Semantic Metalan-guage” (Wierzbicka 1996: 23, emphasis added). Perhaps this is just a matter of wording, and yet, in a publication whose potential readership are not fully familiar with the framework it is of vital importance.

Indeed, Peeters’s volume may well inspire the reader to engage in di-vagations that go beyond mere wording. In what clearly is a continuation of Malinowski’s research programme to reconstruct the point of view of an indigenous speaker (i.e., what on p. 7 here is called the “insider perspectives”), one is led to conclude that the preference given to ethnopsychological “con-

1 This is in fact a common theme of NSM-based publications, cf. Levisen and Waters (2017).

187Reviews

structs” (EPCs), rather than words and concepts, takes us to the realm of lan-guacultures, rather than languages or a-cultural psychological experiences. But even if (and rightly so) the primary focus is on meaning, not on form, one must not underestimate the role of the latter. Its importance has been noted both from the cognitivist and the poetic-translational perspective. The former includes the voice of Dirk Geeraerts: “Language is not just con-tent: it is also form, and its formal side has an expressivity of its own, which does seem to create lexical configurations that can hardly be explained if we only take into account the conceptual expressivity of language” (Geeraerts 1988: 227). The latter is represented by, e.g., James Underhill: “Form makes the poem more meaningful; the poem hits harder than prose” (2016: 12). Of course, this is not to suggest that Peeters, a prime linguist and translator, as well as a speaker of several languages, relegates form to oblivion – it is more a matter of proportions. Indeed, unsurprisingly, the importance of form surfaces in Svetanant’s account of chai, which may figure as either the first or the second element of a compound (p. 103): we are entering here the realm of meaningful lexicogrammatical patterns that code the constructs behind them. In this light, it seems somewhat imprecise to say that word, concept, and construct are interchangeable (p. 3) – instead, they are sometimes (erroneously) interchanged! The very fact that EPCs are deemed a better choice than the other options speaks against their interchangeability.

The book’s greatest potential, however, is that is opens the door to inqui-ries of nature even more fundamental than linguacultural considerations. This is suggested in the short Preface, on p. viii, where the volume’s main in-terest is said to lie in “the topic of personhood” and the way it is built through heart- and soul-like constructs. And so, the most intriguing concept is that of dividualism (cf. the chapter by Hill), which means, in a rather straightfor-ward account, that “[a] person is not seen as a separate entity from the rest of the world, but as a complex of relationships interlinked with the world” (Sa-lisbury 2015, n.p.). In the Melanesian context discussed by Hill, humans are relational creatures whose nature links them with the living, the deceased, and land (p. 75). This is reminiscent of the interconnectedness of humans, animals, and land among Aboriginal Australians, which transpires in their use of expressions such as This land is me/us (cf. Malcolm 2017: 645; Sharifian 2017: 19). On a broader scale, it is an idea that forces one to also ask questions about European philosophy and theology, where over centuries the Christian understanding of person has undergone a rather major change.2 For Boethius

2 NSM researchers do not shy away from tackling religious and theological issues; cf. Wierzbicka 2001 on the message of the Gospel, or Habib 2015 (which Peeters actually references) on God and Allah.

188 L aMiCuS 2019 no. 3 (3)

(5th-6th c. A.D.), persona was “an individual substance of a rational nature”3 (in Migne 1878–1890, vol. 64: 1345). As expressed more precisely by Richard of St. Victor (12th c. A.D.), the person exists in themselves alone according to a singular mode of rational existence4 and is characterised by rational-ity, “absolute individuality, uniqueness and non-substitutability” (Erk 2011: 239). But in contemporary Christian personalism, the focus is very much relational: “The person is the subject who [...] exists in relation to another person, or to other people” (Gacka 2002: 72).

Is this conceptual affinity between this portion of Melanesian world-view and Western religious worldview a matter of accident or founded upon some more profound conceptual base? Or, are these considerations going too far off the point in the first place? I do not think so. Because we are concerned with languacultures here, it can only be expected that the cultural aspect of the inquiry would touch upon issues that are central to cultures, worldviews, systems of thought, and ideologies (in a broad sense). The NSM framework does have the potential to shed light on such issues, especially as it seeks to realise the Leibnizian ideal of discovering the “alphabet of human thought”. In the search of that alphabet, fundamental notions, such as humanity and personhood, are rightly interrogated.

Apart from those issues, three major questions kept drawing my at-tention as I was reading Peeter’s volume. I will relate to those by anchoring them in specific analytical contexts.

First, although the Editor claims that the assembly of semantic primes that has been being sought for decades is now almost ready, doubts are un-likely to disappear altogether. For example, when Deborah Hill explicates someone’s anoa ‘spirit’ in Longgu, she claims that for explicatory purposes ‘something’ is better than ‘part of someone’ (p. 67). I agree that she does have valid arguments for her choice; yet, ‘something’, even if abstract and non-corporeal, is nevertheless “some-thing”, an instance of reification, which – if I read Hill’s argumentation correctly – is better avoided in this instance. The conundrum may well be apparent but it still makes sense to ask whether there is a way out.

Because of the problematic nature of (some of the) primes, or more generally the impenetrability of the explications that may ensue, NSM prac-titioners have begun to rely to some extent on semantic molecules (pp. 14ff). The explications that rely solely on primes are incomprehensible and vague: the exact opposite of what they are designed to achieve. Such was the nature of the critique that came from Kalisz (1998) – and although, in all fairness, the NSM framework was not as developed then as it is today, the crux of that author’s argument remains valid. The lesson one could learn from it perhaps

3 “Rationalis naturae individua substantia.”4 “Existens per se solum juxta singularem quendam rationalis existentiae modum.”

189Reviews

has a more general import: as a scientific procedure, both in physics and semantics, reductionism is laudable and theoretically sound, but humans as cognitive beings do best when they operate at an optimal level of general-ity (recall the fundamental significance of the basic level of categorization), rather than at the level of indivisible semantic primes. Evidence comes not only from the attractiveness of semantic molecules, but also from recourse that is made to Minimal English (Hill).

Secondly, the effectiveness of semantic primes can also be thwarted by their greatest asset: basicness. While defining the Japanese inochi, Yuko Asano-Cavanagh proposes as part of the explication: “if something very bad happens to someone’s body, this something cannot be a part of this someone’s body anymore” (p. 38). However, without insight from Asano-Cavanagh’s general discussion in her chapter, one is left uncertain as to whether “very bad” is used here in the quantitative sense (bad enough for the body to die) or in the qualitative sense (something very bad of a certain sort) – and it hap-pens to be the former. Do we, again, need molecules in this kind of context or would there be another solution?

Thirdly, EPCs are ethnopsychological constructs – but what does this mean exactly? On p. 3 Peeters classifies the heart as a non-physical part of person, which appears to be somewhat one-sided, given that, after all, the heart is a body organ. The dual nature of the heart, both corporeal and non-corporeal, surfaces rather clearly in Svetanant’s analysis of the Thai chai, when it is said to be “linked to the body, but nowhere to be seen” (p. 109). At the same time, chai has unambiguously material properties, such as shape, size, colour, or temperature. Does this point to the inadequacy of the Western dualism of body vs. non-body, where the non-bodily aspects of personhood are located in the mind, soul, or heart? Certainly – and in fact Peeters’s volume, as well as indeed the whole of the NSM enterprise, is a very clear voice to beware of easy claims to universality one might be tempted to make based on Western philosophical tradition.

There is certainly a need to ask questions relating to universality, but one must ask them well. Peeters et al. may have perhaps aimed at a somewhat higher level of precision in this regard. For example, on p. 46 Asano-Cavanagh claims that no one has proved or disproved the reality of tamashii. Reality is a loaded notion – what does the Author mean here? A potentially open-ended ocean of philosophical considerations awaits those who dare venture into the ontological status of that which is real, actual, or virtual, notions that in some configurations need not be mutually exclusive. In a more straight-forward understanding, not only tamashii, but in fact all ethnopsychologi-cal constructs are culturally, symbolically, psychologically, and socially real by the very fact of being EPCs. Unsurprisingly, on p. 48 Asano-Cavanagh notes a lack of consensus on tamashii, whose very essence (as an EPC) must

190 L aMiCuS 2019 no. 3 (3)

be sought not in what it is, but in what people believe it is, including that very lack of consensus. In short, the fundamental assumption that we focus on ethnopsychological constructs means that we are not defining real-life ob-jects whose existence or ontological status can be established beyond human cultural groups.5 Rather, it is the cultural groups themselves that, in a way, call these entities into being, to various degrees of entrenchment, and so figure in explications such as that of “someone’s tamashii”: “many, not all, people think like this: people have this part” (p. 52). Consider also Peeters’s discussion of reification and Anglocentrism (p. 5), where the Editor evokes what I would classify as “bad” questions about the mind, such as whether anyone has ever seen or touched it. They are “bad” in the sense that they assume things must be seen or touched to count as real. But there are also “good” questions that go beyond the mind’s alleged physicality and relate not only to the dilemma whether it exists in “all human babies and all newborn chimpanzees” (p. 5) but whether it exists at all (even if the title of the famous journal Mind suggests that it does).

In the course of reading the book, one may also come across a few (very) minor problems or questions of clarity or editorial nature. For example, when at the very beginning the Editor reports on his personal experience, he leaves the reader confused by referring first to Flemish (p. 1) and then to Dutch (p. 2). Next, when he notes on pp. 4-5 that Levisen “prefers the noun personhood to the adjective ethnopsychological”, one wonders in what sense this is a real choice: the two terms can hardly be used interchangeably. Or, the typesetter may have been a little more careful on p. 68 and introduced clearer spaces in heading [D] of the explication of “someone’s anoa ‘spirit’”. But these are minutiae, without any serious negative impact on the volume’s quality.

And the quality of the volume is high, which I would like to express without hesitation, despite the critical comments that I have offered above. Or rather, I have had a chance to offer these remarks and engage in some philosophizing precisely because Peeters’s book has provided the inspira-tion for doing so. The book is intended for various audiences, especially for doctoral students or researchers in language and culture who are not well acquainted with the NSM framework, but also for readers like myself, who need to be made familiar with the recent developments in this area. I am sure both sections of readership will find the book appealing.

5 In an essentially parallel fashion, the linguistic worldview programme known as cognitive eth-nolinguistics focuses on how “the mental object, the concept or a cultural artefact” is entrenched in the language spoken by a given community (Bartmiński 2009: 216).

191Reviews

ReferencesBartmiński, Jerzy 2009: Aspects of Cognitive Ethnolinguistics. Ed. Jörg Zinken,

transl. Adam Głaz. London and Oakville, CT: Equinox.Erk, Christian 2011: Health, Rights and Dignity. Philosophical Ref lections on an

Alleged Human Right. Frankfurt, Paris, Lancaster, New Brunswick: On-tos Verlag.

Gacka, Bogumił Zygmunt 2002: Personalistic education. Personalism. Science, Philosophy, Theology 3/2002, 71-78.

Geeraerts, Dirk 1988: Where does prototypicality come from? In: Brygida Rudzka-Ostyn (ed.) Topics in Cognitive Linguistics. Amsterdam/Phila-delphia: John Benjamins, 207-229.

Habib, Sandy 2015: Can God and Allah promote intercultural communication? RASK (International Journal of Language and Communication) 42, 77-103.

Kalisz, Roman 1998: Is it possible to operate with primitives in every expli-cation? In: Barbara Lewandowska-Tomaszczyk (ed.) Lexical Semantics, Cognition and Philosophy. Łódź: Łódź University Press, 55-63.

Levisen, Carsten and Sophia Waters (eds.) 2017: Cultural Keywords in Discourse. Amsterdam/Philadelphia: John Benjamins.

Malcolm, Ian G. 2017: Terms of adoption: Cultural conceptual factors under-lying the adoption of English for Aboriginal communication. In: Farzad Sharifian (ed.) Advances in Cultural Linguistics. Singapore: Springer, 625-630.

Migne, Jacques-Paul (ed.) 1878-1890: Patrologia Latina. Paris.Salisbury, R. 2015: On dividualism. Post-Scarcity Anarchism: A journal of Social

Ecology 1. http://postscarcitymagazine.com/Volume/View/1#Dividual-ism, accessed 6 June 2019.

Sharifian, Farzad 2017: Cultural Linguistics. Amsterdam/Philadelphia: John Benjamins.

Underhill, James W. 2016: Voice and Versification in Translating Poems. Ottawa: University of Ottawa Press.

Wierzbicka, Anna 1996: Semantics. Primes and Universals. Oxford: Oxford Uni-versity Press.

Wierzbicka, Anna 2001: What Did Jesus Mean? Explaining the Sermon on the Mount and the Parables in Simple and Universal Human Concepts. Oxford: Oxford University Press.

Documents

Language Mind Culture Societyfiles.clickweb.home.pl/34/60/346047fd-5467-4c0e-a260-17ef4bfa99ea.pdfsical layer of the aria Vissi d’Artez from Puccini’s Tosca. Her study serves as