1

ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT

OF COPULATIVE PERCEPTION VERBS IN ENGLISH:

A COGNITIVE PERSPECTIVE

KAZUMI TANIGUCHI

Osaka University

This paper considers copulative perception verbs in English (e.g., Johnlooks happy; the flower smells sweet) in the framework of cognitivegrammar. Their syntactic and semantic properties, i.e. the subcategori-zation of the adjectival complement, the alignment of the subject to the“logical object,” and the implication of the actual Experiencer to be

either the speaker or generic individuals, might be seen as idiosyncraticin synchronic terms, but they can be precisely characterized with regardto their diachronic development consisting of local analogical exten-sions from their precursor, each of which is motivated by cognitive-grammatical principles. This study will also demonstrate the process inwhich grammatical knowledge is organized with interaction to languageusage, as proposed by the usage-based model of grammatical theory(Langacker (1988), Kemmer and Israel (1994), Israel (1996)).*

1. Introduction

This paper deals with synchronic and diachronic aspects of certain

perception verbs in English which occur with adjectival complements,as exemplified below:

(1) a. John looks happy.

* This is developed from my paper read at the fourteenth Annual Meetings of

Japan Linguistics Society, held at Kwansei Gakuin University, November 16, 1996,

part of which has already appeared in JELS 14. I would like to thank SeisakuKawakami, Yukio Oba and Michael T. Wescoat for their continuous support, and I

wish to express my special thanks to Michael T. Wescoat for insightful comments

and discussions on earlier versions of this paper, as well as extensive stylistic

suggestions. I am also grateful to Yo Matsumoto and anonymous EL reviewers,

who gave me valuable comments on this work. Of course, I am solely responsible

for remaining inadequacies.

English Linguistics 14 (1997) 270-299(C) 1997 by the English Linguistic Society of Japan

-270-

Page 2: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 271

b. This cake tastes good.c. This cloth feels soft.d. That sounds reasonable.e. This flower smells sweet.

These perception verbs are copulative in nature, if one assumes a roughdefinition of copulas as verbs followed by a complement denoting someproperty of the subject participant (Quirk et al. (1985: 54)). Hence,I shall refer to the predicates in (1) as copulative perception verbs(CPVs), and to the configurations in which they occur as CPV con-structions.

One may observe various intriguing properties peculiar to CPVs: (i)the obligatory occurrence of the adjectival complement, (ii) thealignment of the subject, which semantically corresponds to theperceptual object, and (iii) the implication of the existence of anExperiencer, who may be the speaker or else a generic individual.

This work attempts to account for these characteristics of CPVs froma cognitive-grammatical perspective. In fact, the data in (1) are ofconsiderable theoretical interest, since they provide strong motivationfor the usage-based theory of language advanced by Langacker (1988).1According to this notion, a particular basic linguistic pattern with aschema S may undergo extension through speakers' usage and experi-ence, yielding a more abstract schema S' subsuming the newly addedmembers of that pattern. The addition of S' does not, however,cancel the original schema S; they are both stored and available inlinguistic knowledge. Thus, even if the non-basic, extended usagemight appear idiosyncratic, it can be accommodated by positing ahigher-level schema which allows for the existence of the extendedusage as derived and motivated from the basic one. Moreover, alinguistic expression which does not seem to conform to generalcognitive-grammatical principles synchronically can be characterized

properly by articulating the basic phenomenon from which it has beenextended, together with the relevant process of development. The

present study will show that modern CPVs have acquired their currentcharacteristics through the same process, i.e. a series of analogical

1 See Kemmer and Israel (1994) for a detailed discussion of the usage-basedmodel, and also Israel (1996) as an application of this model to the analysis of way-constructions from a diachronic viewpoint.

Page 3: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

272 ENGLISH LINGUISTICS, VOLUME 14 (1997)

extensions from basic to non-basic types. Thus an examination of the

historical process in which CPVs have developed will provide a compre-

hensive explanation for the idiosyncrasies observed above.

2. The Synchronic Properties of CPVs

Here, I shall review in greater detail the three properties of CPVslisted above, the first of which concerns obligatory adjectival comple-ments. Unlike optional secondary predicates occurring with someaction verbs such as John arrived safe and sound or Mary walked lame,the complements of CPVs cannot be dropped without a change inmeaning:

(2) a. *John looks. (John cannot be the perceptual object)b. *That sounds.

The usual rationale for the subcategorization of adjectival complementsdoes not work here. Other copulative predicates (e.g., John keptsilent; Sally remained restless) basically denote spatial positions of theirsubject participants (John kept out of the room; Sally remained in theclassroom), and we can reasonably assume that the class of spatial

prepositional phrases is extended to subsume adjectival complementsdenoting current states of the subject participant, via a generalmetaphor of spatialization: "STATE IS A LOCATION" (Lakoff (1990:63) and Goldberg (1995: 83)). Such an account, however, would notapply to CPVs, whose basic semantic values pertain not to spatiallocations but to perceptual experience. Thus, some other source forthem must be sought.

Secondly, CPVs characteristically align their grammatical subjectswith the "logical objects" of the perceptual experience denoted by theverb; in John looks happy, for example, John is not the Experiencerbut rather the object of visual experience. Thus, this argument maybe regarded as a "passive" subject, just like that of the middle con-struction (e.g., This car sells well). We will take a closer look at theissue of subject selection in the next section.

Thirdly, CPVs imply the existence of an actual Experiencer, whichmay either go unexpressed or be represented in a to-phrase. Asillustrated below, the default value of the implied Experiencer, whoalso serves as a "point of view" for evaluation, is either the speakerherself (to me) or a generic individual (to everyone).

(3) a. This cake tastes good.→This cake tastes good to me.

Page 4: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 273

b. John looks happy. →John looks happy to everyone.

3. Subject Alignment in CPVs

In this section, let us consider subject alignment in CPVs from acognitive-grammatical perspective. As was already noted, the subjectsof CPVs are semantically "logical objects." This raises a problem formany theories of subject alignment. However, the difficulty posed bythis phenomenon may be overcome if one adopts what I shall call a bi-directional cognitive model of event structure. With such a represen-tation, the theory of subject alignment advanced within cognitive

grammar provides demonstrably superior results. This approach none-theless leaves a residue of unexplained cases, which are best handledin a diachronic analysis to be presented in later sections.

3.1. Subject Alignment in Cognitive GrammarLet us begin by reviewing how semantic arguments can be aligned

with grammatical relations, especially the subject. The notion that onecan predict the alignment of arguments with grammatical relations haswon growing support among practitioners of a variety of theories. Itwas once a commonplace of generative grammar to assert that such

grammatical relations as subject and object were purely syntacticnotions and impossible to define semantically, since they are linked tomore than one thematic role (for example, the subject can be an agent,instrument or patient). This criticism was overcome to some degreeby techniques based on thematic hierarchies, which, for instance,equated the subject with the "highest" thematic role in a clause

(Gruber (1965), Jackendoff (1972)). However, while successive modi-fications of this approach have achieved increasing levels of sophistica-tion (Wilkins (1988), Jackendoff (1990)), the fact remains that theirrange of empirical coverage is limited to a class of predicates thatfeature a rather severely restricted set of argument types. Further-more, such approaches are oriented toward characterizing only"logical" or, in transformational terms, "underlying" grammatical

relations (see the notions of "actor" and "undergoer" in Foley and VanValin (1984)). Thus they cannot directly address the problem ofcharacterizing passive subjects. A more general approach to subjectalignment is afforded by the cognitive-grammar view which equateslinguistic meanings with conceptualization.

Page 5: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

274 ENGLISH LINGUISTICS, VOLUME 14 (1997)

Langacker (1990a: 224) claims that such grammatical notions assubject and object, like any other linguistic categories, can be describedin terms of prototypes and schemas. A prototype represents the mosttypical and central semantic property of the category in question. Incontrast, a schema characterizes the commonalities observable in thefull range of instances of a given category. Thus, schemas are highlyabstract in nature. The subject prototype identifies the most typicalcandidates for subjecthood; it turns out that prototypical subjects areagents and topics. As for the subject schema, which should embracethe common characteristics shared by all subjects, Langacker's (1990a)

proposals on this matter are summarized below:2(4) Subject schema:

The subject encodes a relational figure, i.e. the most saliententity.

The schema in (4), which is based on cognitive salience, applies tosubjects of both transitive and intransitive clauses; note that in thelatter case, (4) necessarily selects the sole relevant participant as therelational figure. This schematic definition has a variety of advan-tages. It can correctly identify the subjects for a broad range of

predicates, even in cases that are problematic for a thematic-hierarchyapproach. For instance, it trivially handles static relations, as in Johnresembles Mary, where the subject John is highlighted and evaluated interms of the object Mary, which functions as landmark. Furthermore,the schematic definition subsumes not only active subjects but also

passive ones (Langacker (1990a: 229)), since the passive constructiondescribes a marked construal of a situation in which the relational

2 Notice that the schema in (4) is intended only to identify the argument whichbears the grammatical relation of subject. The association of a particular case withthe subject is a separate issue handled by distinct mechanisms that vary amonglanguage groups. In nominative-accusative languages, there are cases specificallyassociated with certain grammatical relations. In ergative-absolutive languages, thesystem of case-marking is determined by the conceptual autonomy of events; theabsolutive case marks a participant that constitutes the conceptual "core" of aprocessual event, i.e. a single participant's change of location or state (hencethe absolutive marks the intransitive subject and the transitive object), whilegrammatical relations still correlate with relative cognitive salience. Still otherlanguages associate cases with thematic roles (see Langacker (1990a: 234-254) formore details). The diversity of case-marking patterns strongly suggests that caseand grammatical relations should be handled separately, so there is no need for (4)to mention such issues.

Page 6: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 275

figure is exceptionally imposed on some participant (e.g. the patient)other than the one which would usually be construed as most promi-nent (e.g. the agent).

With regard to the schema in (4), the default-case construal of anasymmetrical relation such as energy transfer or perception and idea-tion may be summarized as follows (see Langacker (1990a: 221-222)):

(5) The relational figure is imposed upon the head of an asym-metrical relation involved in the described event.

In the generalization above, the "head" indicates a participant whoinitiates an asymmetrical relation. Thus, the agent who initiatesenergy transfer and the experiencer who initiates perceptual activity arelikely to be the relational figure and therefore encoded as subject.

3.2. The Bi-directional Model of PerceptionIn light of this definition, let us reconsider the subject alignment seen

in perception verbs. Kemmer (1993: 136-137) comments that percep-tion verbs fall into two types according to whether the semantic role ofthe subject participant is the Experiencer as in (6a), or the Stimulus asin (6b). Kemmer refers to these as Experiencer-based (EB) andStimulus-based (SB) perception verbs respectively:

(6) a. Joe smelled the garlic. [EB]b. Garlic smells good. [SB]

Obviously, the CPVs of interest here are a subset of the SB perceptionverbs in the categorization above. Indeed, CPVs constitute a propersubset of SB predicates, since the latter include such non-copulativeforms as that smells.

Kemmer's observation regarding the availability of two alternativesfor the role of subject would follow directly from the subject schema ifone adopted what I shall call a bi-directional cognitive model of

perceptual events, as depicted in Figure 1, which contrasts with anormal, or uni-directional model. Such a bi-directional model featurestwo participants, the Experiencer and its perceptual object, which standin two asymmetrical relations indicated by dashed arrows in themodel.3 One of the two relevant relations is the Experiencer's cogni-

3 The notion of bi-directional models may also be found in various analyses ofexchange of possession predicates like buy and sell (Jackendoff (1990), Croft (1991)).

Page 7: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

276 ENGLISH LINGUISTICS, VOLUME 14 (1997)

zance of the object, technically known as "mental/perceptual contact."The other is the emission of a stimulus by the perceptual object toexcite the senses of the Experiencer.

With regard to the EB perception verb in (6a), the subject participantis the Experiencer (Joe) who initiates mental/perceptual contact withthe perceptual object (garlic) and therefore is the head of the relevantrelation. As for the SB perception verb in (6b), the object of per-ception (garlic) itself lies not in the head but rather in the "tail" orendpoint of the mental/perceptual contact. Nonetheless, it is the headof the stimulus emission, exciting the Experiencer's sense of smell.Thus, the subjects in both EB and SB perception verbs are compatiblewith the characterization of subject in (4) and (5). There are twocandidates for subject, simply because there are two complementaryrelations to which the schema can be applied.

3.3. Some Unexplained CasesAt this point, one should note that there are certain cases for which

the foregoing analysis does not seem to be entirely suitable. Considerthe following pair comparable to (6):

(7) a. I felt the blanket.b. The blanket feels nice.

Whereas the subject of the EB perception verb feel in (7a) is in accordwith the subject schema, the same probably cannot be said of (7b). Itis quite doubtful that one can recognize any discrete event of emittingan independent stimulus on the part of the subject of the copulative

predicate feel in (7b) (namely the blanket). Rather, the object ofperception and the stimulus are indistinguishable, unlike the garlic and

<Figure 1>

E: ExperiencerO: Object of perception

Note that the function of the mental-perceptual contact and the stimulus emission

are different in nature, the latter involving a causing effect upon the Experiencer.

However, both are indicated by dashed arrows in the diagram for the sake of

graphical simplicity.

Page 8: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 277

its gaseous emission in (6b). Crucially, the emission of gasses from

garlic constitutes a recognizable event separate from the mental/perceptual contact initiated by the Experiencer. However, in the caseof feel, no such distinct event of stimulus emission can be recognized,and it would be quite ad hoc to impose the foregoing bi-directionalcognitive model that was applied to the perception verb smell, sincethere is no motivation for positing an independent relation of stimulusemission for feel. As a result, the relation of stimulus emission onwhich the analysis of subject alignment for smell depended is notavailable to explain the choice of subject in (7b). Thus, it appears that there are two types of perception verbs, namely

bi- and uni-directional ones. Empirical motivation for this distinctionmay be found in contrasts like the following:4

(8) a. The smell reached me.b. The sound reached me.

(9) a. ??The feeling reached me.b. ??The taste reached me.c. ??The sight reached me.

In (8), the stimulus is describable as having a trajectory, whence itfollows that the stimulus is perceived as separate from the object fromwhich it was emitted. This constitutes recognition of an event ofstimulus emission. Consequently, the verbs in (8) are bi-directional,while those in (9) are not.

The categorization of bi-and uni-directional perception describedabove may prove controversial, especially the status of visual experi-ence. While (10a) is acceptable, this kind of expression cannot beapplied to all the visual objects:

(10) a. The light reached me.b. ??His appearance reached me.

It seems that the status of visual experience depends on the kind ofobject involved; when it comes to light emitted by the object, as in

(10a), the visual experience is regarded as bi-directional. However, Icharacterize visual experience as essentially uni-directional, since thesituations where visual objects emit stimuli are considered to be limit-ing cases.

In summary, the classification of perceptual events and CPVs corre-

4 I owe the discussion here to Michael T. Wescoat.

Page 9: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

278 ENGLISH LINGUISTICS, VOLUME 14 (1997)

sponding to each class is represented below. Note that, although I

categorize copulative look basically as uni-directional, it is situated

closer to bi-directional perception since it can be bi-directional on some

occasions.

Given the underlying cognitive models for the various perceptionverbs illustrated in Figure 2, the cognitive-grammar theory is hard

pressed to provide an adequate explanation of subject selection in theuni-directional CPVs, feel, look and taste. While the subjects of theCPVs smell and sound head relations of stimulus emission in theirrelevant cognitive models, and therefore conform to the subjectschema, the same cannot be said of the other problematic subjects.5Such subject selection indicates that uni-directional CPVs encode amarked construal of a perceptual event where the perceptual objectlying in the tail of the mental/perceptual contact relation is renderedmore salient than the Experiencer, against the general cognitive incli-nation that the head of an asymmetrical relation is ceteris paribusmore prominent than the tail. One must then wonder why such aspecialized pattern of construal and encoding is allowed and conven-tionalized in uni-directional CPVs. It seems unlikely that an explana-

<Figure 2>

(a) bi-directional (b) uni-directional

5 CPVs differ from ordinary copulative expressions like you are tired in that thereis a relevant participant, i.e. the Experiencer, which can be expressed in the former(e.g., you look tired to me) while it is often impossible to express in the latter (*youare tired to me). Thus, although CPVs lack objects and therefore appear to beintransitive, the perceptual events they describe nonetheless inevitably include twoparticipants, i.e. the Experiencer and the perceptual object. Consequently, subjectalignment in CPVs should be treated in the same way as in transitive predicates.

Page 10: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 279

tion can be formulated in purely synchronic terms. I therefore claimthat a key to the solution to this problem will be found if one looks atthe diachronic processes by which CPVs have evolved into what theyare at present. Historical observations presented in the next sectionwill show that the difference in status with respect to the subjectschema between bi- and uni-directional CPVs is clearly reflected intheir historical development.

4. Historical Observations

Here I present a chronological outline of the development of CPVsin Table 1, which provides a summary of the first attested appearanceof each CPV as recorded in the Oxford English Dictionary (OED)together with the earliest dates for the corresponding EB and non-copulative SB usages. The last category is exemplified by the bellsounded or that smells.6

<Table 1>

11 12 13 14 15 16 17 (C)

[bi-directional]smell (EB) 1175------

(SB) 1175------(CPV) 1220------

sound (SB) 1352------

(CPV) 1374------[uni-directional]look (EB) 1000------

(CPV) 1400------taste (EB) 1340------

(CPV) 1552------feel (EB) 897------

(CPV) 1581------

6 In Table 1, I have limited my attention to the information furnished by theOED. The relevant evidence that I have been able to locate elsewhere in thelinguistics literature has been sparse, but it nonetheless supports the generalizationillustrated in Table 1. For example, Nakao (1972: 282) reports that look began toaccompany an adjectival complement in the fifteenth century, as is consonant withthe date reported for the inception of this usage in Table 1. The present analysis isalso partially echoed in the conclusions of Frantz (1939), who speculates that thecopulative usage of feel (e.g., it feels stiff) might be obtained by analogy withcopulative look and taste.

Page 11: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

280 ENGLISH LINGUISTICS, VOLUME 14 (1997)

First, let us look at the bi-directional perception verbs smell and sound.The verb smell can be employed in three ways, i. e. in EB, SB or CPVusages, while sound has two usages, SB and CPV.7 With both smelland sound, the CPV usage appeared almost simultaneously with thecorresponding EB and SB usages. This seems to indicate that theCPV usage of bi-directional perception verbs was quite basic from thebeginning, just as were their EB and SB usages.

The situation of uni-directional CPVs, on the other hand, displays aremarkable contrast with that of bi-directional ones. While EB usagesof look, taste and feel had already appeared in earlier stages, their CPVusages did not emerge until late in the Middle English period.8 Thisdelay in the emergence of uni-directional CPVs leads us to assume thatuni-directional CPVs arose as extensions from already established bi-directional CPVs; if uni-directional CPVs were an unmarked phenom-enon, they would have appeared as early as their EB usages, just as inthe case of the bi-directional CPVs observed above. Furthermore,comparing bi- and uni-directional CPVs, we find that the latter tendedto appear later than the former.

As a consequence of the foregoing observation, we may assume thatCPVs underwent the following process of development:9

(11) CPVs were first established in bi-directional perceptionverbs in which the perceptual object is encoded as a clausalsubject, and afterwards the CPV usage was extended intouni-directional perception verbs as well.

7 The EB coding of auditory experience is independent from the SB one, and isrepresented by distinct single verbs, e.g. hear and listen. The formal independenceof the SB (and copulative) sound also seems to reflect remarkable conceptualautonomy of the stimulus emission in auditory experience.

8 The reason why copulative look appeared earlier than the other uni-directionalperception verbs might be related to the ambiguity of visual experience pointed outin 3.3, i.e. the fact that visual perception can be bi-directional as well.

9 A reviewer points out the possibility that CPVs evolved from such putativeimpersonal constructions as Me smells the flower by some analogical mechanism.However, such attested impersonal constructions as me thinks were supplanted bycorresponding personal constructions like I think through a general process ofreplacing dative subjects with nominatives; this observation would lead us to expectthat the speculative archaic form Me smells the flower would yield to the personalconstruction I smell the flower, rather than undergoing the conjectured grammaticalchange described above. See Visser (1963: 29-30) and Nakao (1972: 298-300) onthis issue.

Page 12: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 281

Note that we are dealing here with a relationship between a basic or"core" phenomenon and a marked or "peripheral" one arising as an

extension from the core case. The subject alignment of bi-directionalCPVs adheres to the generally well-motivated characterization ofsubject in (4) and (5), and may therefore be viewed as a basic

phenomenon while the corresponding alignment in uni-directionalCPVs contradicts that schema and is therefore peripheral. This typeof core-peripheral distinction overwhelmingly correlates with historicaldevelopment; the most basic and unmarked types of linguistic expres-sions establish themselves early and motivate subsequent extensions.The latter may be akin to their precursors in significant respects, whiledeviating from them in some other ways and thus evolving into noveltypes of expressions. In the case of CPVs, the bi-directional CPVswhose subjects are straightforwardly analyzable given existing cognitive-

grammar definitions appeared as a basic or unmarked phenomenon,and triggered a structural extension which allows for the application ofthe same argument alignment in uni-directional CPVs as well, throughvarious factors including analogy of sensory modality.

Although the deviant subjects of uni-directional CPVs can be char-acterized in terms of their historical development, there still remainunexplained facts concerning CPV constructions, especially the obliga-tory co-occurrence of adjectival complements. These issues will behandled in the next section, in a more detailed analysis of the devel-opment of CPVs.

5. The Development of CPVs

In the present section, we shall consider the process of extension inwhich the CPV usage, originally established in bi-directional perceptionverbs, has been extended into uni-directional ones. This developmentmay be characterized as a sequence of modest steps, each motivated bycognitive-grammatical principles.

5.1. The Origin of the Adjectival ComplementFirst, let us consider the original form in which bi-directional CPVs

initially emerged. In particular, we claim that the adjectival comple-ment seen in modern CPVs finds its origin in ordinary adverbialadjuncts applied to perception verbs. It will be shown that theconversion from adverbs to adjectives took place due to coincidental

Page 13: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

282 ENGLISH LINGUISTICS, VOLUME 14 (1997)

overlaps in morphological form and cognitive function.When we think about the grammatical category of adjectives speci-

fically in historical terms, we also need to consider their adverbialcounterparts, whose relevancy to adjectives is considerable both mor-phologically and functionally. The morphological relation betweenadjectives and adverbs is quite trivial in present-day examples such assmooth/smoothly, with the suffix -ly being the marker of the adverb.Sometimes adjectives and adverbs even coincide, as in fast. As for thesemantic and functional similarity they bear, the characterization inLangacker (1987) offers a helpful insight:

(12) a. adjective: an atemporal relation with a thing astrajector.

b. adverb: an atemporal relation with a relation astrajector. (Langacker (1987))

Both adjectives and adverbs evoke the notion of a scale; in the case ofsmooth/smoothly, for example, both involve a scale of smoothness.The only difference between them is seen in their actual targets ofmodification. What is given a positive value with respect to smoothnessis a thing (coded in a nominal) in the case of the adjective smooth, anda relation (coded in a relational predicate) in the case of the adverbsmoothly.

Such considerable similarity between adjectives and adverbs has longbeen present in English. According to Nakajima (1979: 152-154),adverbs in Old English were generally formed by adding -e toadjectives (e.g. softe "softly," deope "deeply"), and the combination ofthe adjectival suffix -lic and -e, namely -lice, was also used to formadverbs. With regard to the earlier examples of bi-directional CPVs, wecan find a considerable number of data which seem to include adverbsrather than adjectives, as illustrated below:

(13) a. jenne pe mon wule tilden his musestoch he bindeouppon pa swike chese and bret hine for Jon pet hesclode swote smelle.

[1225 Lamb. Hom. 53; MED s.v. smellen]b. Garleek...smellej strongliche.

[1398 Trev. Barth. 214b/b; MED s.v. smellen]These examples lead us to assume that bi-directional CPVs originallyoccurred with adverbs, which later evolved into adjectives, as inpresent-day English. This assumption is supported by the followingcomment in Visser (1963):

Page 14: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 283

Since Old English adverbs in -e generally appear in Early MiddleEnglish without this -e, it is not easy to decide whether in theearlier instances we have to do with adverbs or adjectives.

(Visser (1963: 210))The observation above indicates that there was a morphological coinci-dence of adjectives and adverbs in Middle English.10 Furthermore,Visser actually provides the following remark which seems to be con-sonant with our assumption, together with some examples listed in (14)that support his idea:

[W] e may have to do with an adverb instead of an adjective(e.g. in 'to smell swete'), especially because formally identicalcombinations with adverbs in -ly actually occur.

(Visser (1963: 215))(14) a. To call them a slip, would indeed sound strangely.

[1789 T. Twining, Aristotle's Treatise on Poetry, 216: OED]b. The rags smelt unpleasantly.

[Anstey, Vive Versa XVI, 305: Visser 1963]Now, recall that it was also in the Middle English period when bi-directional CPVs first appeared, as illustrated in Table 1; this allows usto infer that the adjectival complements of bi-directional CPVs shouldhave originated as adverbs. Moreover, the morphological coincidenceof adverbs and adjectives in Middle English presumably set the stagefor a grammatical reanalysis in which the former category wasreinterpreted as the latter. Thus, I conclude that bi-directional CPVsinitially occurred with adverbs, and I would like to refer to this form asan "original CPV."

Let us consider the original meaning of bi-directional CPVs in moredetail, taking this flower smells sweet(ly) as an example. Presumably,this would have meant roughly this flower emits smell sweetly. Asnoted in (12b), the semantic function of the adverb is to modify arelation between things with respect to some scale. In the currentcase, the target of the modification of the adverb (sweetly) will be thestimulus emission whose head is the perceptual object (this flower);hence, the expression this flower smells sweetly describes the effect ofthe smell emitted by the flower. The semantic structure of this bi-

10 As for the drop of the adverbial suffix -e in Middle English, see Jespersen

(1942: 408) for example.

Page 15: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

284 ENGLISH LINGUISTICS, VOLUME 14 (1997)

directional CPV with an adverb is illustrated in Figure 3:

In the diagram above, the upward arrow indicates the scale ofsweetness provided by the adverb sweetly, and the shadow representsthe target of modification, i.e. what is given a positive value on thescale. Note that what is shadowed in the diagram is the whole

perceptual event, since the adverb itself could potentially modify eitherthe mental/perceptual contact or stimulus emission. In the case ofCPVs, however, the modified relation is limited to the latter, as isindicated by bold lines.

Now, it is clear that there is nothing idiosyncratic in the semanticsand grammatical relation displayed by original CPVs. Selecting the

perceptual object as subject is quite compatible with the schematiccharacterization of subject in (4) and (5), since that object is the headof an asymmetrical relation. Furthermore, modifying the relation ofstimulus emission with an adverb is also reasonable. Hence, the pre-cursor of present-day CPVs appears to have been quite an unex-ceptional linguistic phenomenon. In what follows, I would like toexamine step by step the process through which the original CPVs havebeen extended to uni-directional perception verbs, ultimately acquiringtheir present idiosyncrasies.

5.2. Extension to Uni-directional Perception VerbsIntuitively, it would be no surprise if the CPV expressions established

in bi-directional perception verbs came to apply also in uni-directionalones by analogy of sensory modality. Actually, we can find someearlier examples of the generally uni-directional perception verb lookoccurring with apparent adverbs ending in-ly:

(15) a. His grace looks cheerfully and smooth.[1594 Shakespeare, Rich. III, III, iv, 50]

<Figure 3>

Page 16: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 285

b. Good gentleman look fresh and merrily.

[1601 Idem, Jul. Caes. II, i, 224]c. You look most shockingly.

[1773 Goldsmith, She Stoops I, 168]

(Visser (1963: 210))In the examples above, the coordination of cheerfully and smooth in

(15a) and of fresh and merrily in (15b) provides evidence of the mor-phological and functional homogeneity of adverbs and adjectives, giventhe background assumption that elements combined by a coordinateconjunction are supposed to be of the same grammatical category. Bethat as it may, these data suggest that the form of the original CPVs,which take adverbial adjuncts, gradually applied to such uni-directional

perception verb as look. Notice, however, that adverbs are semanti-cally incompatible with uni-directional look, since there is no relationof stimulus emission for the adverb to modify. How then could theCPV usage be extended to uni-directional perception verbs despite suchfunctional incompatibility?

A solution to this may be found in the assumption that adverbs inthe original CPVs came to be reinterpreted as adjectives in the midstof the morphological and functional confusion surrounding the twocategories. This is not so hard to imagine, since expressions like this

flower smells sweet(ly) can entail that sweetness is one of the propertiesof the flower when it is smelled. As a result, bi-directional CPVssemantically came to express a property of the subject participant asadverbs were reinterpreted into adjectives. Diagrammed in Figure 4 isthe semantic structure of bi-directional CPVs with adjectives:

<Figure 4>

Comparing Figures 3 and 4, we find the shift from an adverb to anadjective quite straightforward, due to their functional similarity; bothevoke the notion of a scale, and their actual difference is found only in

Page 17: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

286 ENGLISH LINGUISTICS, VOLUME 14 (1997)

the target of modification. While the adverb in Figure 3 modifies therelation of stimulus emission whose head is the perceptual object, theadjective in Figure 4 applies just to the perceptual object itself, whosesalience is inherited from the original CPV.

With the CPV construction established in bi-directional perceptionverbs, the stage was set for application to their uni-directional counter-

parts. Since the target of modification of adjectival complements isthe perceptual object, as Figure 4 clearly shows, these complementscan also be accommodated by uni-directional perception verbs, whichhappen to feature such objects as well. This state of affairs is in starkcontrast to the situation with adverbs, whose target of modification wasstimulus emission, which is present in bi-directional perception verbsbut absent in the uni-directional variety. Thus, the advent of adjec-tival complements removed the impediment to extending the CPV con-struction to uni-directional perception verbs. Diagrammed below is thesemantics of uni-directional CPVs as an extension from bi-directionalones:

Once the adjectival complements of CPVs were conventionalized,they also came to allow for predicative nouns, as shown below:

(16) a. He smells April and May.[1598 Shakespeare, Merry W. III, ii, 70: Visser 1963]

b. Whose entreaties in this case sounded commands in theears of such as were piously disposed.

[1639 Fuller, Holy War ix, (1840) 14: OED, s.v. sound]c. All pale he lies, and looks a lovely Flow'r.

[1697 Dryden, AEneid xi, 99: OED, s.v. look]The data above show that the CPVs with predicative nouns appeared inlater times, further suggesting that the predicative function of the CPVswas acquired gradually, as adverbs in the original CPVs were re-

<Figure 5>

Page 18: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 287

interpreted into adjectives.Finally, the extension of CPVs to uni-directional perception verbs

brought additional semantic change to this construction. Whereas bi-directional CPVs essentially profiled the relation of stimulus emission

(see Figures 3 and 4), with uni-directional CPVs it is the Experiencer'smental/perceptual contact that is profiled (see Figure 5). This shift ofthe profiled relation yields a certain implication in CPVs to the effectthat the perceptual object is evaluated through described perception,i.e. the mental/perceptual contact of the Experiencer; this affects bi-directional CPVs by feedback. As a result, the verb smell in this

flower smells sweet becomes ambiguous, the profiled relation beingeither the stimulus emission (as in the original sense) or the mental/

perceptual contact as diagrammed in Figure 6. This ambiguity wouldnot be seen in sound, due to the existence of other verbs like hear orlisten, which encode mental/perceptual contact exclusively."

5.3. The Vestiges of Original CPVs in Modern English

As we have seen so far, the precursor of CPVs is the bi-directional

perception verb occurring with an adverb which modifies the relevantstimulus emission. In the course through which the initial CPV has

been extended to uni-directional perception verbs, obtaining the cur-

<Figure 6>

11 Another relevant question arises with regard to the range of verbs that havedeveloped into CPVs. The reasons why the CPV usage is established in look butnot in see might pertain to the semantics observed here, since at this stage the CPVconstruction with uni-directional perception verbs conveys that the Experiencerevaluates the perceptual object by agentively making mental/perceptual contact withit. This implication of agentivity or volitionality on the part of the Experiencermay favor look, which denotes agentive visual experience, over a perception verbdenoting rather passive visual experience (i.e. see).

Page 19: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

288 ENGLISH LINGUISTICS, VOLUME 14 (1997)

rent syntactic and semantic properties as a result, the original sense hasnot vanished, and is somehow preserved even in Modern English.

First, let us consider an example cited by Jespersen (1927: 368):

(17) There was very little furniture, but it all seemed so differentfrom the room downstairs. It smelt differently, looked dif-

ferent, and felt different.[1916 S. Aumonier, Olga Bardel, 36]

In the expression above, we find three CPVs, smell, look and feel.The second and third are accompanied by the adjective different, whilesmell occurs with the adverb differently. Concerning this datum,Jespersen himself notes that "[t]he reason for the two forms [in (17)] isnot very obvious (ibid.)." However, from the fact that the confusionof the adverb and adjective seen in Middle English should be alleviatedin early Modern English due to the conventionalized use of the suffix-ly in adverbs, the alternating usage of the adjective different on the

one hand and the adverb differently on the other could not be consideredto be arbitrary. Actually, the reason for the two forms in (17) isaccounted for straightforwardly in the analysis proposed thus far.Recall that the CPV appeared in the first place as a combination of abi-directional perception verb and an adverb modifying the relationof stimulus emission. This original sense having survived, the bi-directional perception verb smell reasonably occurs with an adverb,since it involves stimulus emission which is susceptible to modificationby an adverb. As for the remaining uni-directional CPVs (look and

feel), the presence of adverbs must be blocked since these verbsessentially do not include any stimulus emission for the adverb tomodify.

The same explanation will apply to the observation in Quirk et al.

(1985: 407-408) that, while present-day CPVs in principle do not allowfor the co-occurrence of adverbs instead of adjectives, some CPVsseem to be more tolerant of the use of adverbs than others, as shownbelow:

(18) a. The flower smells {sweet/?sweetly}.b. The food tastes {marvelous/? *marvelously}.

The reason why smell sweetly is not so odd as taste marvelously isobvious under our analysis; bi-directional perception verbs feature therelation of stimulus emission for the adverb to modify, while their uni-directional counterparts have no such relation, depriving the adverb ofa target to which it can apply. As a result, (18b) is less acceptable,

Page 20: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 289

just as it would have been prior to the development of CPV con-structions in English.

5.4. Backgrounding of the ExperiencerEven if the CPV construction can be applied to uni-directional per-

ception verbs, there still remain some problems to be solved regardingthis usage. The perceptual experience they denote consists of a singlerelation whose head is the Experiencer, and selecting the perceptualobject lying in the tail of the relation as a subject crucially contradictsthe schematic characterization of subject in (4) and (5). In order tomake the perceptual object salient enough to be coded as a subject, weneed to downgrade the inherent prominence of the Experiencer andbring it into the background. For this purpose, the Experiencer isrendered generic or else associated with the speaker herself; this givesrise to an implicit "to everyone" or "to me" interpretation for CPVs.

5.4.1. Generic ExperiencerOne of the factors in backgrounding the Experiencer pertains to

genericity, as is pointed out by Kemmer in the following passage:In the Stimulus-based type, the experiencer is generic and hencehas a low degree of saliency; the most salient participant, giventhe semantics of the construction, is the Stimulus itself.

(Kemmer (1993: 136))This observation stems from a basic cognitive principle that may besummarized as follows: other things being equal, specific and concreteentities stand out more readily than general and abstract ones. In thecase of uni-directional CPVs illustrated in Figure 4, the Experiencerlying in the head of the relation will inevitably be more salient than the

perceptual object, if both participants are equally specific. However,if the Experiencer is rendered generic, the relatively more specific

perceptual object comes to gain cognitive salience according to thecognitive principle above, despite the disadvantage of being situated inthe tail of an asymmetrical relation.

5.4.2. Sub j ectification in CPVsIn the alternative interpretation for CPVs, the Experiencer can be

the speaker herself as in John looks happy (to me). The contributionof this strategy to the backgrounding of the Experiencer will beexplained in terms of the notion of "subjectification" proposed in

Page 21: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

290 ENGLISH LINGUISTICS, VOLUME 14 (1997)

Langacker (1990b).Let us begin by surveying the nature of subjectification. Briefly,

this refers to a process in which the "subject" of conceptualization (i.e.the conceptualizer, or the speaker of a linguistic expression) is inte-

grated implicitly into the meaning of an utterance. As a demonstra-tion, consider the following expressions each of which involves the

preposition across (Langacker (1990b: 17-21)):(19) a. Vanessa jumped across the table.

b. Vanessa is sitting across the table from Veronica.c. Vanessa is sitting across the table (from me).

Figure 7 (a-c) depict the situations described in (19a-c) respectively. 12The basic semantic value of the preposition across is taken to be aspatial path that goes beyond the entity profiled by its prepositionalobject, as in (19a), where the path Vanessa traverses is observed andconstrued objectively from the speaker's vantage point. In (19b), onthe other hand, what moves along the path designated by across is notVanessa as in (19a), but rather the speaker's mental contact, whichmoves towards Vanessa starting from Veronica, who serves as thereference point (R). Furthermore, (19c) does not even mention anyreference point in the path, but we can easily interpret the reference

point as coinciding with the speaker herself.In the foregoing examples, we can see two ways in which aspects of

<Figure 7>

(a) (b)

12 In the diagram, G stands for the ground, which refers to an integrated con-ception of the speech event, its participants (speakers and addressee) and itsimmediate circumstances. See Langacker (1990b: 9) in this regard.

Page 22: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 291

an entity may receive subjective construal by integrating the speaker's

presence implicitly into the meaning: replacement of an objectivemotion with a subjective one (i.e. mental contact) as in (19b), andidentification of a reference point with the speaker as in (19c).Although the subjectively-construed. reference point is usually inte-

grated into the meaning, it is possible and even more appropriate torepresent it overtly by providing some special context as in (20):

(20) Look! My picture's in the paper! And Vanessa is sittingacross the table from me! (Langacker (1990b: 20))

The effect of subjectification is also seen in CPVs. If we assume theExperiencer in CPVs to be a kind of reference point, which representsthe starting point of mental/perceptual contact and a norm for evalu-ating the perceptual object, what happens in CPVs is essentially thesame as in the across example in (19c): a reference point comes tobe identified with the speaker. The Experiencer in CPVs is thussubjectively construed, and implicitly integrated into the overallsemantics of CPVs as sketched in Figure 8. Consequently, theExperiencer is backgrounded despite its position of privilege in thehead of the relation, and the perceptual object gains relative

prominence instead.

Here, we have to consider Langacker's (1990b) proposal that sub-

jectification, or subjectivity in the construal of an entity, is also asignificant factor in "grammaticization," the diachronic change of asemantically-rich "content" word into a "grammatical" one with anabstract meaning. A question naturally arises as to whether thesubjectification identified in the foregoing discussion has led to the

grammaticization of CPVs in Modern English. To answer this ques-tion, one might in turn ask if CPVs feature the type of "semanticbleaching" that typifies grammaticization. Indeed, we can notice a

<Figure 8>

Page 23: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

292 ENGLISH LINGUISTICS, VOLUME 14 (1997)

certain attenuation in the semantics of CPVs; however, this process hasclearly not been carried to its ultimate completion since CPVs have yetto acquire the properties that Horton (1996: 320) associates with the

prototypical copula, which he characterizes as "a colorless linkingdevice", "semantically empty" or "reduced." More detailed discussionof the matter of semantic attenuation of CPVs will be provided in thenext section.

6. Further Syntactic and Semantic Extension of CPVs

In the foregoing discussion, we have presented a sketch of the entire

process in which original CPVs have been extended to what they are atpresent. Current CPVs are not the endpoint of the process, however;we can still observe further extensions of CPVs in their meaning andsyntactic behavior, partly due to analogy with the personal use of seem

(e.g., John seems happy), one of the basic copulas in English, whichhistorically became established much earlier than CPVs. CPVs areakin to seem in that they all occur with adjectival complements whichdenote subjective evaluation of an entity; therefore, it is quite reason-able to assume that CPVs might come to obtain some formal propertiesof seem as an iconic reflection of their semantic similarity.

6.1. Seem and CPVsFirst, let us look at the influence of seem upon CPVs with regard to

syntax. Although CPVs originally occurred with an adverb modifyingthe stimulating relation in bi-directional perception verbs, as we haveseen previously, the use of the adverb in the CPV construction is gen-erally prohibited in current English usage. The following descriptionin the OED indicates that the constraint on the use of adverbs incopulative look results from its semantic and grammatical similarity toseem:

This use [with advs. of manner] is often indiscriminately con-demned, but is justly censurable only where look is virtuallyequivalent to seem, so that it requires a predicative complementand not a qualification of manner. Owing, however, to the

prejudice excited by the inaccurate use, look now rarely occurswith advs. of manner other than well, ill, badly. [OED, s.v. look]

Furthermore, look exhibits a grammatical behavior comparable tothat of seem, which can take an infinitive complement (e.g., John seems

Page 24: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 293

to be happy) .13

(21) a. John seems to be a fool.b. John looks to be a fool.c. *That sounds to be a good idea.d. *It {tastes/feels/smells} to be a fruit.

Among the various CPVs above, only look allows for the co-occurrenceof an infinitival complement. Given that the following datum is theearliest attested usage of an infinitive with look, it would appear thatthis structure is a fairly recent development in the language:

(22) It looks to me to be narrow and pedantic, to apply theordinary ideas...

[1775 Burke, Sp. Conc. Amer. Sel. Wks. I, 192: OED]Obviously, had look not first become a CPV, thereby acquiring the

property of taking complements in general, it would not have beenpossible for this verb to subcategorize for the specific subclass ofcomplements formed from infinitives. We assume that this syntacticchange was motivated by semantic analogy with seem, which takesinfinitives, in addition to adjectival complements comparable to thosefound normally in CPV constructions. As for the reason why onlylook can accompany an infinitival complement, we can point out atleast the psychological fact that visual perception is especially central tohuman inference and thought (as a matter of fact, seem itself typicallyrepresents inference based on visual experience). Thus, look is mostlikely to approximate seem which denotes subjective inference, theirsemantic proximity also triggering convergence in grammatical behavioriconically.14

Finally, let us examine semantic change in CPVs, especially look,whose evolution has been driven by analogy with seem. As mentioned

previously, we can find considerable commonality between seem and

13 The use of an infinitival complement in copulative look is preferred especiallywhen it includes a predicative nominal. See Quirk et al. (1985: 1173) on this issue.Note that the infinitival complement of look is generally limited to to be, unlike thatof seem, though our analysis does not exclude the possibility that expressions like (i)might become acceptable as a result of further extension:

(i) *John looks to have a cold.(ii) John seems to have a cold.

14 Note that the verb see exhibits the same kind of semantic expression, as in Ican see what you mean. See Sweetser (1990: 25-48) for a discussion of the closerelationship between vision and intellect.

Page 25: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

294 ENGLISH LINGUISTICS, VOLUME 14 (1997)

look which stems from the fact that both denote subjective inferencetypically based on visual experience. Actually, look is getting closer toseem and is inclined to represent the speaker's inference in general,regardless of the relevant kind of perception. Consider for examplethe expression below:

(23) Judging from her letter, she looks to be the best person forthe job. [Longman Dictionary of Contemporary English, s.v. look]

What the speaker of (23) in fact sees is her letter, not the subject

participant of copulative look. Hence, the speaker's evaluation on thesubject participant is made by inference, which is not necessarily basedupon direct visual experience. The departure of look from its funda-mental semantic value also shows the centrality and dominance of visualexperience in the cognitive processing of inference, which I assume hastriggered the semantic convergence of look and seem in particular.

6.2. Extension of the CPV Construction to Action VerbsAs we saw in the previous section, original CPVs with bi-directional

perception verbs plus adverbs extended to uni-directional ones partlydue to the fact that they all pertain to sensory modality. As a result,the CPV construction conveys an evaluation on some entity coded inthe subject position through the perceptual experience of the impliedExperiencer. This meaning can undergo further extension, to code asituation where the implied Experiencer evaluates some entity throughher action, not limited to perceptual experience. In fact, we can findcopula-like semantics. and behavior in the action verbs read and touchin Middle English. Cited below are parts of their respective definitionsand examples from the OED that seem to display the characteristics ofcopulative predicates:

(24) read: To turn out (well or ill), or have a specified char-acter, when to read; to produce a certain impressionon the reader. Also, to convey a statement whenread.

(25) a. Whose productions...read better than they act.[1789 Aristotle's Treatise on Poetry. I, 254]

b. Nothing can read more free and easy than his presenttranslation. [1828 Examiner, 84/2]

(26) touch: (with descriptive extension) To 'feel' to the touch;to cause a specified sensation when touched.

(27) a. We say this beast touches nicely upon its rib.

Page 26: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 295

[1770-4 A. Hunter, Georg. Ess. IV, 575]b. They touch rough dusty rough, as books touch that

have been lying unused. [1885 Jefferies, Open Air, 104]In the examples above, we find a significant parallel with CPVs in thateach subject participant corresponds not to the agent but to the object

(or theme) of the denoted activity, through which the agent evaluatesits theme as expressed in the complement. One might suspect thatsome of the complements in the data above are adverbs rather thanadjectives; better in (25a) might be the comparative of the adverbwell, and nicely in (27a) is formally an adverb. But, if we take intoconsideration the morphological and functional coincidence of adjec-tives and adverbs as mentioned earlier, the data which seem to occurwith adverbs cannot be excluded from this discussion. In fact, judgingfrom the example in (15), the confusion of the adjective and the adverbhad survived at least until the eighteenth century, when the datum in

(27a) happens to have been penned.However, the existence of CPV-like usages of read and touch was

temporary, and they are almost unacceptable in current English, asreflected by the elicited data in (28):

(28) a. *?This table touches hard.b. *?This book reads interesting.

As for touch, its copular sense might be blocked by copulative feel.With respect to read, on the other hand, its copula-like nature can beseen preserved in another kind of expression illustrated in (29), wherethe object of the reading activity is encoded as the clausal subject.

(29) a. This article reads as follows:...b. The thermometer reads 22℃.

Furthermore, in present-day English, some other verbs exhibit thefollowing important characteristics of CPVs: their subject participantsare semantically the objects of the denoted events, and their adjectivalcomplements represent properties of the subject participants, evaluatedby the implied Experiencer (or agent) through the perceptual experi-ence or activity denoted by the verb. Consider the following expres-sions: 15

15 The paraphrases in (30b) and (30c) are also due to Horton (1996: 329). Asfor the acceptability of the data, some native speakers report that (30b) and (30c)are distinctly odd. Such partial acceptability shows that these expressions have notbeen fully established yet.

Page 27: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

296 ENGLISH LINGUISTICS, VOLUME 14 (1997)

(30) a. He tested positive for HIV.b. The meat cuts tough.

[This meat is tough when someone cuts it](Horton (1996: 329))

c. The cake eats short and crisp.

[The cake is short and crisp when someone eats it](ibid.)

Since the subject, participant in each example in (30) is the logicalobject of the verb involved, and since the property evaluated throughthe denoted activity is expressed by the adjectival complement, we cansafely take these expressions to have come into being as furtherextensions of the CPV construction.

7. Concluding Remarks

In this paper, we have investigated the historical course of thedevelopment of CPVs, whose evolution is summarized below:

<bi-directional> <uni-directional>stimulus emission (S)

mental/perceptual contact (MC) mental/perceptual contact (MC)

This flower smells sweet(ly).

(S) (adverb)

This flower smells sweet. This flower looks sweet.

(S) (adjective)(MC) (adjective)

This flower smells sweet.

(MC) (adjective)This flower looks a kind of rose.

(MC) (predicative noun)

[personal seem construction]

This flower looks to be a rose.

(infinitival complement)

Page 28: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 297

It has been shown that the properties of CPVs, which might beidiosyncratic at present, can be characterized precisely with respect tothe process of the extensions. First, the encoding of the perceptualobject as a clausal subject was facilitated in original CPVs based on bi-directional perception verbs by the fact that the perceptual object canbe seen as the head of the relation of stimulus emission, and thus canbe accommodated in the schematic characterization of subject in (4)and (5). Secondly, the co-occurrence of a complement is also anythingbut mystifying if it originated not as an adjective but as an adverbmodifying stimulus emission in bi-directional perception. Finally, theimplication of the existence of an Experiencer resulted from the needto background it and make the perceptual object relatively salient evenin uni-directional perception verbs.

This study has also revealed that the overall process of thedevelopment of CPVs consists of local analogical extensions, as is thecase with the growth of get-passives (Givon and Yang (1994)) or way-constructions (Israel (1996)). The extension from the original bi-directional CPVs toward uni-directional ones was driven by semanticanalogy based on the fact that all these forms describe sensorymodalities. The shift from adverbs to adjectival complements can beattributed to morphological and functional similarity between thesecategories. Furthermore, the additional grammatical behavior foundin CPVs, especially look, is taken to be brought out by semanticanalogy to copular seem. These observations seem to follow Israel's

(1996) Production Principle of analogical usage which maintains thatutterances should sound like things the speaker has heard before; it isdue to this principle that the range of verbs available as CPVs hasexpanded, from bi-directional perception verbs to uni-directional onesand even to non-perceptual action verbs.

Cognitively-based works on historical linguistic change have also

proposed that "any synchronic model of grammatical organization mustbe reconcilable with the observed facts of linguistic change (Israel

(1996: 227))." This thesis naturally stems from the usage-based viewof grammatical knowledge as "a structured inventory of convention-alized linguistic units (Langacker (1987: 57))" obtained through one'sexperience concerning language use. Although speakers are usually

quite unaware of the history of the expressions they employ, the factremains that linguistic units have been conventionalized through timeand therefore naturally bear distinct signs of their diachronic evolution.

Page 29: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

298 ENGLISH LINGUISTICS, VOLUME 14 (1997)

In the case of CPVs, one finds in Modern English a seemingly eclecticcollection of expressions which appear to lack a unified synchroniccharacterization, but which reflect their historical development in thesense that one can enumerate a series of empirically verifiable andcognitively motivated steps that have widened the class of CPVconstructions at each successive stage, ultimately yielding. the currentlyobservable state of affairs. And, as is suggested in the example ofextension to action verbs, the present CPVs exhibit just a passing phaseof their growth. Of course it is hard to imagine how one couldaccommodate each stage in the development of CPVs in synchronic

grammar without resorting to the usage-based model of language; thus,this study hopefully contributes to the mounting evidence supportingthis cognitive view of the organization of linguistic knowledge.

REFERENCES

Croft, William (1991) Syntactic Categories and Grammatical Relations: TheCognitive Organization of Information, Univerisity of Chicago Press,Chicago.

Franz, Wilhelm (1939) Die Sprache Shakespeares in Vers und Prosa. [JapaneseTranslation by Shizuka Saito, Hideo Yamaguchi and Akira Ota, ShinozakiShorin, Tokyo.]

Foley, William A. and Robert D. Van Valin Jr. (1984) Functional Syntax andUniversal Grammar, Cambridge University Press, Cambridge.

Givon, Talmy and Lynne Yang (1994)“The Rise of the English GET-

Passives,” Voice: Forms and Functionin, ed. by Paul Hopper and Barbara

Fox, 89-117, John Benjamins, Amsterdam.Goldberg, Adele E. (1995) Constructions: A Construction Grammar Approach

to Argument Structure, University of Chicago Press, Chicago.Gruber, Jeffrey S. (1965) Studies in Lexical Relations, Doctoral dissertation,

MIT.Horton, Bruce (1996)“What Are Copula Verbs?” Cognitive Linguistics in the

Redwoods, ed. by Eugene H. Casad, 319-346, Mouton de Gruyter, Berlin

and New York.

Israel, Michael (1996)“The Way Constructions Grow,”Conceptual Structure,

Discourse and Language, ed. by Adele E. Goldberg, 217-230, CSLIPublications, Stanford.

Jackendoff, Ray S. (1972) Semantic Interpretation in Generative Grammar, MITPress, Cambridge, MA.

Jackendoff, Ray S. (1990) Semantic Structure, MIT Press, Cambridge, MA.Jespersen, Otto (1927) A Modern English Grammar part III (Syntax: second

Page 30: ON THE SEMANTICS AND DEVELOPMENT A COGNITIVE …

ON THE SEMANTICS AND DEVELOPMENT OF COPULATIVE PERCEPTION VERBS 299

volume), Ejnar Munksgaard, Copenhagen.Jespersen, Otto (1942) A Modern English Grammar part VI (Morphology),

Ejnar Munksgaard, Copenhagen.Kemmer, Suzanne (1993) The Middle Voice, John Benjamins, Amsterdam.Kemmer, Suzanne and Michael Israel (1994)“Variation and the Usage-Based

Model,” CLS 30,165-179.

Lakoff, George (1990)“The Invariance Hypothesis: Is Abstract Reason Based

on Image-Schemas?” Cognitive Linguistics 1,39-74.

Langacker, Ronald W. (1987) Foundations of Cognitive Grammar vol.1:Theoretical Prerequisites, Stanford University Press, Stanford.

Langacker, Ronald W. (1988)“A Usage-Based Model,” Topics in Cognitive

Linguistics, ed. by Brygida Rudzka-Ostyn, 127-161, John Benjamins,Amsterdam.

Langacker, Ronald W. (1990a) Concept, Image and Symbol: A Cognitive Basisof Grammar, Mouton de Gruyter, Berlin and New York.

Langacker, Ronald W. (1990b)“Subjectification,” Cognitive Linguistics 1, 5-

38.Nakajima, Fumio (1979) Eigo Hattatsu Si (The Historical Development of

English), Iwanami Shoten, Tokyo.Nakao, Toshio (1972) Eigoshi II (History of English), Taishukan, Tokyo.Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1985)

A Comprehensive Grammar of the English Language, Longman, London.Sweetser, Eve (1990) From Etymology to Pragmatics: Metaphorical and Cul-

tural Aspects of Semantic Structure, Cambridge University Press, Cambridge.Viberg, Ake (1983)“The Verbs of Perception: A Typological Study,” Linguis-

tics 21, 125-162.Visser, Frederik Theodoor (1963) An Historical Syntax of the English Language

part I, E. J. Brill, Leiden.Wilkins, Wendy (1988)“Thematic Structure and Reflexivization,” Syntax and

Semantics 21, ed. by Wendy Wilkins, 191-213, Academic Press, New

York.

Dictionaries

Longman Dictionary of Contemporary English, second edition. (1987)Middle English Dictionary. (a.k.a. MED)The Oxford English Dictionary, second edition. (a.k.a. OED)

3-10-3-207 Miyayama

Toyonaka-shi

Osaka 560

e-mail:[email protected]