Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Why is aspect so hard to acquire in a second
language?Roumyana Slabakova
University of Southampton and NTNU, the Norwegian University of
Science and Technology
Affiliations and Acknowledgements
Overarching Goaln Are there "easier" and "harder" linguistic properties, constructions,
and even parts of the grammar? n Principled distinctions, well understood in linguistic theory, and
solid bodies of data should be used.n Applying the insights achieved by generative SLA research and
theory in the last thirty years also informs language teaching. n It makes practical sense that if teachers know what is hard to
acquire and thus target instruction by practicing it more in the classroom, learners will achieve better fluency and higher accuracy in the second language.
Functional morphologyMy daughter often take-s the bus.3 sg NP present-3sg
Agreementbut also
TenseHabitual aspectOvert subjectNominative subjectVerb stays in VP
Functional morphology
My daughter often take-s the bus.
AgreementTenseHabitual aspectOvert subjectNominative subjectVerb stays in VP
EXPONENT
FEATURES
Features live in the lexiconFeatures are properties of morphemes and words. They can be of three types: n phonological features such as [± sonorant], [± voice],
building up the sound side of the linguistic signal; n semantic features such as [± specific], [± definite], [±
countable] for nouns and nominal phrases, [± stative] [±telic] for verbs and verbal phrases, etc.;
n morphosyntactic features such as [Case: {Nom, Acc, Gen…}], [Person: {1, 2, 3}], [Number: {Sg, Pl}], etc.
What has to be acquired?n The semantic as well as morphosyntactic
features can be assembled in different feature bundles, expressed by the words in the target and native languages.
n Not only the features themselves but the linguistic environments in which they appear are important to acquire.
Aspectual featuresTwo types of Aspectn Lexical (situation), calculated on
predicates not on verbsn Grammatical (viewpoint), calculated
based on other functional morphemes: -ing, -ed, se.
Vendler lexical classes and aspectual features
Dynamic Durative Telic Examples
State – + – be asleep, trust, believe, love
Activity + + – sleep, run, sing
Accomplishment + + + run a mile, write a book
Achievement + – + wake up, reach the top, die
Grammatical Aspect
Perfective aspect
Imperfective aspect
Grammatical aspect examples1. John was making a chair, and he may still be making a chair.
2. #John made a chair, and he may still be making a chair. (that same chair)
Cross-linguistic Variationn The linguistic elements that determine lexical
aspect do not vary much; their exponents (expression) varies
n As for grammatical aspect, all languages have ways to express notions such as stop or start an action, or be in the middle of it, even if only with aspectual adverbials.
n However, there is a lot of cross-linguistic variability with grammatical encoding—whether or not it is encoded at all, and, if so, which aspects are expressed how.
Cross-linguistic variationPolish, aspectual pairs1. Klaun budował most
clown BUILD.IMPERF.PAST bridge‘The clown was building a bridge./The clown used
to build a bridge’
2. Klaun z-budował mostclown BUILD.PERF.PAST bridge‘The clown built a bridge.’
Cross-linguistic variationSpanish, aspectual tenses1. El payaso construía un puente.
the clown BUILD.IMPERF.PAST a bridge‘The clown was building a bridge./The clown used
to build a bridge’
2. El payaso construyó un puente.the clown BUILD.PERF.PAST a bridge‘The clown built a bridge.’
Cross-linguistic variationMandarin Chinese, aspectual particles1. Zhe-ge xiao-chuo zai xiu qiao.
clown bridge ASP build bridge‘The clown was building a bridge/The clown used
to build a bridge’
2. Zhe-ge xiao-chuo xiu-le yi-zuo qiao.clown build-ASP a-CL bridge‘The clown built a bridge.’
Feature Reassembly Hypothesis (Lardiere 2009)n L2A is a two-step process of mapping and
(re)assembly of formal feature bundles.n The first step is based on perceived similarities
between the functional meanings of L2 and L1 lexical items (reminiscent of Contrastive Analysis)
n Similarities lead to an initial feature mapping of L1 items onto target items. E.g., Spanish-speaking learners of English typically map the Spanish imperfect onto the English past progressive because they partially overlap in meaning.
Feature Reassembly Hypothesis (Lardiere 2009)
n The next step involves feature reassembly: features can be added or deleted, gradually adjusted based on input-based evidence for meaning and usage.
n For the English progressive tense, noticing its unavailability with stative verbs and the absence of habitual interpretation should result in alterations to the feature set.
n Feature reassembly may occur slowly or not at all if the relevant evidence in the input is rare or ambiguous.
n Hence, no instantaneous resetting of parameters is expected.
Form–meaning mismatchn Different meanings are encoded in seemingly similar
morphemes
Guillermo robaba en la calle. (habitual)Guillermo rob-IMP in the street‘Guillermo habitually robbed (people) in the street.’
Guillermo robó en la calle. (one-time event)Guillermo rob-PRET in the street‘Guillermo robbed (someone) in the street.’
Felix robbed (people) in the street. (habitual)Felix robbed a person in the street. (one-time event)
Form–meaning mappings
Acquisition tasksn The pieces of inflectional morphology
(-ing, -ed) come from the functional lexicon. n The functional projections (e.g., AspP) where
semantic features are checked are part of sentence syntax.
n The aspectual meanings (ongoing event, habitual event, one-time finished event) reside in conceptual structure and are universal.
n Different languages have different form-to-meaning mappings.
The Semantic Complexity Theoryn Certain combinations of lexical and grammatical aspect
are semantically simpler than others, e.g.n Telic plus perfective is simpler than telic plus
imperfective. WHY?n Telic predicates implicate a culmination moment, hence,
a telic eventuality description creates an expectation of culmination. Perfective aspect is compatible with this expectation, but imperfective and progressive aspect cancel the culmination expectation. (e.g. John was making a chair, and he may still be making a chair.)
n Similarly, atelic and imperfective/progressive are an better match.
Predictionsn This theory of Semantic Complexity predicts an
asymmetry in the development of perfective and imperfective aspect as related to lexical aspect classes.
n For telic predicates, perfective aspect is acquired earlier than imperfective aspect; telic plus imperfective is semantically more complex.
n A mirror-image prediction for atelic predicates: imperfective aspect is acquired earlier than perfective aspect. An atelic plus perfective combination is more complex.
Child language studiesn The research designs use complete situations
in which an event reaches its natural culmination,
n versus incomplete or ongoing versions of the same event, capitalizing on the completion entailment of telic plus perfective, and the lack of this entailment for telic plus imperfective
n Completion entailment: a telic predicate and a perfective marker = complete event
Study 1Kazanina and Phillips (2007)
child language
Experimental setting
Test sentences
Overall accuracy (Experiment 1)
Aspect Percent correct
Perfective 95%
Imperfective 37%
Interim conclusionThere are many (moving) parts to aspect:n Two types of aspect (lexical, grammatical)n Expressed variously in different
languagesn Some combinations are harder, some are
easiern Children acquire the easier combinations
earlier and more accurately.
Study 2Montrul and Slabakova (2002)
Form–meaning remapping
Form–meaning mappings
Sentence Conjunction Task1. La clase era a las 10 pero empezó a las 10:30.
The class was at 10 but started at 10:30.–2 –1 0 1 2
2. La clase fue a las 10 pero empezó a las 10:30.The class was at 10 but started at 10:30.
–2 –1 0 1 2
Yes or No morphologyn 71 adult learners of Spanish, English
native speakern Advanced and intermediate learnersn Based on a test of inflectional morphology
of aspectual tenses, the intermediate learners were further divided into a Yes-morphology group (80% achieved accuracy) and a No-morphology group.
Acceptance–rejection contrast with statives
NativeSpeakers
Advancedlearners
YesMorphology
NoMorphology
Imperfect 1.56 0.92 0.53 0.12Preterite -1.5 -0.9 -0.32 -0.25
-2-1.5
-1-0.5
00.5
11.5
2
ImperfectPreterite
* n.s.
Study 3Dudley and Slabakova (in press)
The Reportive PresentIn certain types of oral narratives, namely narrative (sports), novels and stories, commentaries, etc., the semantic distinction between the simple and progressive forms is cancelled out. Two meanings:
1. “So I go to pay for my sandwich, and the guy asks me for two pounds – but I don’t have any money on me!” complete past events
2. “The ball hits the net, and she loses the point.”ongoing eventsThe simple form can adopt some of the functional meanings of the progressive form, in special context.35
Form–meaning mismatchThe reportive present simple is PERFECTVE in meaning. It is triggered by a special context.
Present ProgressivePresent Simple
HABITUAL ONGOING
36
PERFECTIVE
One form withthree meanings but also one meaning with two forms!
Experimental participantsn 29 Chinese native speakers, average
proficiency 27.9 out of 40, range 20–37, on a standardized proficiency test
n 31 French native speakers, average proficiency 30.3 out of 40, range 23–39
n 29 English native speakersn Participants were university students and
were tested in Southampton and in Rennes
37
Example stimuli, Aspectual Interpretation Task
REPORTIVE: Tom is in London today watching the marathon. His friend is a professional runner who is taking part in the London Marathon. Tom is listening to the commentator speaking through the overhead speaker: “Paul starts a sprint past the front runner.” Only 100 metres remain between Paul and the finish line…
A. … Paul is running towards the finish line.”B. … Paul runs towards the finish line.”C. Both sentences in A and B are true descriptions. D. Don’t know.
38
AIT: aspectual tense choice in the REPORTIVE context
39
Study 4Slabakova (2003)
Additional entailments
English simple present tense The English simple tense cannot denote ongoing events.
*She eats an apple right now. (#ongoing event)She is eating an apple right now. (ongoing event)She eats an apple (every day). (habitual event)
With stative predicates, however, the ongoing reading of the English present is possible.
Mike is lazy. (characteristic state)Mike is being lazy today. (temporary state)
English bare infinitive The English bare infinitive denotes not only the process part of an event but includes the completion of that event
I saw Mary cross the street. (completion entailed)I saw Mary crossing the street.
(no completion entailed)Explanation: bare verbs are marked with a feature [Perfective] in the lexicon, because English inflectional morphology is impoverished.
Bulgarian present tense No present progressive tense and the present simple tense is ambiguous between a habitual and an ongoing event or state.
Maria sega jade torta. (simultaneous event)Maria now eat-PRES cake‘Mary is eating a cake right now.’
Maria jade torta vseki den. (habitual activity)Maria eat-PRES cake every day‘Mary eats cake every day.’
Bulgarian present tense This is true of stative predicates as well, ambiguous between a characteristic and a temporary state.
Maria ləže. (characteristic state)Maria lies-present‘Mary is a liar.’
Maria v momenta ləže. (temporary state)Maria at this moment lies-present‘Mary is lying (at the moment).’
Bulgarian infinitivesBulgarian verbs do not need to be marked [Perfective] in the lexicon. They are amply marked with person, number, and tense endings. Consequently, Bulgarian equivalents to bare infinitives do not entail completion of the event.
Ivan vidja Maria da presiča ulicata.Ivan saw Maria to cross street-DET‘John saw Mary crossing/*cross the street.’
(no completion entailed)
Learning Task (n .
English eventive verbs have the feature
[Perfective]
Bare verbs denote a complete event.
Present tense has only a habitual interpretation
Progressive is needed forongoing interpretation
States in the progressivedenote temporary states
Experimentn 112 Bulgarian-speaking learners of English and
24 native speakersn 3 proficiency levels: low intermediate, high
intermediate, and advancedn A production task for ascertaining knowledge of
inflectional morphologyn A truth value judgment task for checking
knowledge of interpretation
ExperimentA quadruple testing completed interpretation of English bare forms: Matt had an enormous appetite. He was one of those people who could eat a whole cake at one sitting. But these days he is much more careful what he eats. For example, yesterday he bought a chocolate and vanilla ice cream cake, but ate only half of it after dinner. I know, because I was there with him.
I observed Matt eat a cake. True FalseI observed Matt eating a cake. True False
Results (Slabakova 2003)
Mean Accuracy on Bare Verb vs. -ing Form on Perceptual Reports ( percentage )
5865
73
64
7568
76 75
8779 82
68
8375
95
86
0
10
20
30
40
50
60
70
80
90
100
Bare verb (F) -ing (T) Bare verb (T) -ing (T)
Incomplete event Complete event
Low IntHi IntAdvancedControls
Conclusions (Slabakova 2003)
n It is possible to acquire semantic properties in the second language that do not come from the native language
n All semantic effects of the triggering inflectional property appear to be engaged at the same time.
n Any impact of instruction? NOAll groups perform equally well on all conditions.
Bottom linenAspect is difficult for all learners, including
children learning their native languagen In the second language, the same aspectual
meanings (semantic features) are expressed, but the exponents may differ, or be mismatched on different functional morphemes
n HARD
Bottom linen However, there may be certain entailments
that come for free as part of knowing the functional morphology
n EASY
Recommendationsn Teachers should know which properties, combination of morphemes or meanings are easier and more difficult for learners, hence
n Language teachers should study
Generative classroom-oriented research is growingn Applications of generative SLA to the
language classroom (Whong, Gil and Marsden volume, Rothman, VanPatten, Slabakova)
n The Meaning in Language Learning network http://millnetwork.org/
n Recent issue of Language Teaching Research
n New journal Educational Linguistics
Thank you!