Constructional Properties versus Lexical Specific Transfer: Overgeneralized Causatives in L2 English and L2 Spanish
Mónica Cabrera and María Luisa Zubizarreta
University of Southern California
This study provides evidence that overgeneralized causatives in L2 English
and L2 Spanish is driven by selective L1 transfer. Not all aspects of the grammar
are transferred at once. Properties of the causative construction are transferred
first; verb-specific properties, later. Pre-advanced learners overgeneralize
L1 properties of the causative construction: they prefer overgeneralized causatives
with verbs of change of state/location. Advanced learners transfer properties
of specific verb classes: they restrict the overgeneralization of causatives
to verbs that are allowed in lexical causatives in their L1. Our analysis provides
a unified account of the overgeneralization of causatives.
Syntax First: Mismatches between Morphology and Syntax in first
language acquistion elucidate linguistic theory
Cristina Dye, Claire Foley, Maria Blume, and Barbara Lust This paper examines several of the critical conceptual and analytic components
of a theory of "Distributed Morphology" which appear to differentiate it from
other theories and relates these to current studies of the acquisition of the
morpho-syntax of verbal inflection in first language acquisition. It is argued
that a wide set of results involving both experimental and natural speech analyses
based on English, German, and Dutch initial child grammars between two and four
years of age converge with the 'syntax first' properties of the "Distributed
Morphology" paradigm. Implications for current linguistic theory and current
acquisition theory are discussed. Task pragmatics and the lexicon: A re-examination of the role
of language in cognition
Akiko Fuse and Laraine McDonough We examined spatial relational terms in Japanese and English in order to investigate
how differences in collecting and coding data can influence our views on the
role of language in cognition. Monolingual Japanese-speaking two-year-olds and
adults were shown relations such as placing a ring on a pole and were asked
to describe the activity. Adults produced 28 different spatial verbs and children
produced 17. Tests using the same stimuli by Choi & Bowerman showed that English-speaking
adults and children produce 6 and 3 different spatial prepositions respectively.
However, verbs contain dynamic information (manner/path of action) whereas prepositions
provide static information. We tested English speakers and coded for verb plus
preposition combinations to capture categorization of dynamic and static aspects
of spatial categories. The results showed no differences in the number of spatial
categories produced by English and Japanese speakers; however, responses were
significantly less variable among the Japanese speakers. Optionality as 'demarking' in L2 advanced state
Masahiro Hara This poster reports L2 advanced-state optionality in the grammaticality judgment
of Japanese passives. This optionality is discussed as induced by 'demarking'
(see Sorace, 2000) and originating in the lexicon. Data consist of grammaticality
ratings of 84 sentences on a five-point scale collected from 81 English and
85 Chinese learners. It was found that highly advanced learners of both L1s
consistently accepted grammatical ni indirect passive sentences, but optionally
rated one of its ungrammatical versions as grammatical. It is argued that in
rating the ungrammatical ni indirect sentences, these subjects analyzed them
as the unmarked ni direct passive, smaller and less complex in structure than
the marked ni indirect passive. The difference in structural complexity between
these two passives derives from the properties of the two different passive
verbs Watanabe (1996). Hence, the etiology of optionality is traceable to selection
of different lexical items in the Numeration. Why do children say did you went?: the role of do-support
Ryoko Hatttori This paper examines "doubling errors" in L1 acquisition, where "tense and/or
agreement is incorrectly expressed twice-- once on the 'fronted' auxiliary and
once on the main verb" (O'Grady 1997; 166), as in:
*Did you went home?
Hurford (1975) claimed that doubling errors arise from an incorrectly internalized
Subject Auxiliary Inversion rule, while Mayer et al. (1978) claimed that they
are the result of incorrectly formulated movements, i.e., copying without deletion.
Through the examination of doubling errors in yes/no questions and negative
declaratives, which I classified into three groups that differentiate between
these two hypotheses (those involving do-support, those involving be, and those
involving modals) in three English-speaking children--Adam, Eve, and Sarah (Brown
1973) from CHILDES database (MacWhinney 2000), I show that neither hypothesis
accurately accounts for the distribution of doubling errors. Instead, I propose
that the main factor underlying doubling errors is the presence of the typologically
unusual do-support. (150 words) The acquisition of the German focus particle auch/too:
Comprehension does not always precede production
Tanja Hüttner, Hein Drenhaus, Ruben van de Vijver, The German focus particle auch ('too') occurs as a stressed and an unstressed
variant, yielding two possible interpretations of sentences with auch, depending
on the respective focus domain. As shown by Nederstigt (2001) children produce
both variants of auch at age of 1;5. A picture selection task with children
(age 2;11 - 7;8) and adults has shown that this productive knowledge of auch
significantly precedes the interpretive knowledge. In addition, the data reflect
that the comprehension of stressed auch is acquired before that of unstressed
auch.
These results are explained as an integration problem with the different types
of focus information. While in the stressed variant of auch the lexical and
the prosodic information coindices, in the unstressed variant the prosodic information
is not on auch itself, but on a following element. In this case the lexical
information and the prosodic focus information must be integrated for a correct
interpretation. The role of argument structure and object familiarity in Japanese
children verb learning
Mutsumi Imai, Etsuko Haryu, and Hiroyuki Okada Previous research showed that young children had difficulty learning a new
action verb, failing to generalize a novel verb to the same action when the
object in the action was changed (Imai et al., 2002). The present research examined
factors that might help children segregate the object from the representation
of verb meaning. We tested whether explicit specification of the arguments facilitates
verb learning even for Japanese children, who often hear verbs without arguments.
Also manipulated was familiarity of the theme object. Because acquisition of
verb meaning requires children to grasp the relation between nouns, children
may learn verbs more easily when they have solid knowledge about the object.
The effect for object familiarity was found, but specification of the argument
structure did not improve their performance. These results are compared to the
results from English-speaking children (Mayer et al., 2003), and implications
about universal and language-specific properties of verb meaning acquisition
is discussed. The Interaction of Lexical Aspect and Phonological Salience
on Regular Past Tense Affixation in L2 English
Elaine C. Klein, Iglika Stoyneshka, Kent Adams, Yana Pugach, Stephanie Solt,
and Tamara Rose This study investigates the influence of lexical aspect and phonology on variability
in the perception and production of the past tense -ed morpheme among beginner
and intermediate learners of L2 English. Under controlled conditions, we tested
the hypothesis that telic (e.g. achievement) verbs are more readily associated
with accuracy in regular past tense than are atelic (e.g. activity) verbs. To
test the effects of phonology, we measured the degree to which accuracy depends
on the allomorphic realization of a given verb's past tense ending (i.e. -[t]
as in stopped; v. [d] as in closed v. [Id] as in started). Results show little
impact of lexical aspect, but very strong phonological effects, regardless of
proficiency level. This investigation of learner variability suggests a weakening
of the Primacy of Aspect hypothesis, and a more prominent role played by phonology
in past tense affixation of L2 English. Addressing the Syntax/ Semantics/ Pragmatics interface:
the Acquisition of the Japanese Additive Particle _mo_
Kazumi Matsuoka The semantic interpretation of adverbs such as _only_, _also_, _even_ has been
a central issue of alternative semantics for focus (Rooth 1996). Experimental
research was conducted to investigate how children associate the syntactic position
of the Japanese additive particle _mo_ with the range of the possible alternative
set. Unlike its English equivalents such as _also_, the range of the alternative
set in the interpretation of _mo_ is syntactically determined. Children from
a daycare center in Osaka (mean 6;4) were tested with a Truth-value Judgment
task (Crain and Mckee 1985).
Nearly half failed to demonstrate an adult-like comprehension of the sentences
containing _mo_. Their response pattern indicated the possibility that their
grammar provided a non-adult focus interpretation of _mo_, by choosing an alternative
set regardless of the syntactic position of the focus item. Based on the finding,
we will consider the process by which different focus items are interpreted.
On the acquisition of causatives in Japanese
Keiko Murasugi, Tomoko Hashimoto, and Sachiko Kato We investigate the acquisition process of the Japanese causative construction,
which employs the morpheme -(s)ase. Based on a four-year longitudinal observational
study with a Japanese-speaking child, we show that the construction is acquired
in three steps. First, causative sentences are produced without the morpheme
-(s)ase, and then sentences like (1a) with a non-agentive causee are observed.
And finally, children acquire more "regular" causatives with an agentive causee
as in (1b).
1. a. Taroo-ga ningyoo-ni kutu -o hak -ase -ta b. Taroo-ga Hanako-ni hon-o yom-ase-ta This study provides for Matsumoto's (2000) hypothesis that causatives with
the morpheme -(s)ase are structurally ambiguous: they may have a complex structure
with -(s)ase as an independent verb as generally assumed in the literature,
but they may also have a simple structure with a non-agentive causee and V-(s)ase
as the predicate. Against the Aspect First Hypothesis
Olesya Olbishevska The study reports on two experiments testing the predictions of the Aspect
First Hypothesis on Ukrainian. Experiment 1 tests young children's production
by observing what verbal aspectual morphology (perfective or imperfective) children
predominantly use to express past and present. Experiment 2 is a comprehension
study. It tests whether children use perfective morphology to encode telicity,
and imperfective atelicity, i.e. whether they mark lexical aspect with the help
of overt grammatical aspect marking.
The findings of the experiments provide evidence that 2.5 - 4.5 year olds do
not restrict their use of perfective grammatical aspect to past tense and imperfective
to present. The results demonstrate that even the youngest children could comprehend
past tense equally when it is applied to events that are completed and to events
that are incomplete. Therefore, the results do not seem to support either version
of the Aspect First Hypothesis. Learning definite determiners: Genericity and definiteness
in English and Spanish
Ana T Perez-Leroux, Alan Munn, Cristina Schmitt, and Michelle Deirish This paper reports 2 studies comparing the interpretation of definite plurals
in English and Spanish. In Spanish definite plurals can be interpreted as generics
but English definite plurals cannot be, even in favoring contexts (present tense).
Study 1 tested children's interpretation of definites with 8 stories about two
atypical members of their kind (e.g. zebras with spots.) Yes/no questions evaluated
generic or specific interpretations of the subject with definites, bare plurals
and demonstratives. Study 2 examined children's ability to use tense to restrict
definite generic interpretations. Our results show that while English children
over-generalize the use of definite plurals to generics, both Spanish and English
children are sensitive to Tense as a restrictor of generic interpretations.
Reappraising Poverty of Stimulus Argument: A Corpus Analysis
Approach.
Florencia Reali and Morten Christiansen The poverty of stimulus argument for innateness of grammar is based on the
assumption that the information in the environment is not rich enough to allow
a human learner to attain adult competence. Auxiliary fronting in polar interrogatives
has been taken as strong support for the poverty of stimulus argument. Here
we reassess the assumption of absence of evidence for aux-fronting through a
corpus analysis of child-directed speech. We used bigram/trigram models to compare
the probability of correct (Is the lion that is roaring hungry?) and incorrect
(Is the lion that roaring is hungry?) hypotheses for auxiliary fronting. We
found that the probability of correct aux-question sentences was about twice
as high as the incorrect ones. These results show that the statistical information
present in the corpus allows to select the correct fronting hypothesis in the
95% (trigram) and 92% (bigram) of the cases. Alternative Grammars in Acquisition: markedness- vs.
faithfulness-oriented learning
Anthi Revithiadou and Marina Tzakosta Cross-linguistically a fixed set of markedness constraints derives children's
templatic truncations. It has been argued that children's early productions
realize the least sonorous segments of the preserved syllables of adult words.
The proposal advanced in this paper, however, brings forward another crucial
factor in the selection of the segmental make-up of such outputs, namely faithfulness
to edgemost segments. More specifically, it is claimed that in the course of
acquisition co-grammars, which exploit fine-grained markedness or faithfulness
distinctions, are developed in parallel. In markedness driven co-grammars, segmental
selection is performed on the basis of sonority scales: xri.'stu.lis ['tulich]
'Christ-DIM' whereas in faithfulness driven co-grammars, the edgemost consonant
is produced: fri.ãa.'nu.la ['fula]/*'nula 'toast-DIM'. Consequently, Greek raises
significant questions about the nature of constraints and the dynamics of their
rankings that construct early grammars and shape intermediate ones (i.e. co-grammars)
towards the final stages of phonological development. The production of SE- and SELF-anaphors in Dutch child
language
Esther Ruigendijk, Sergio Baauw, Sergey Avrutin, and Nada Vasic Some Dutch verbs can take either SE- (zich) or SELF-anaphors (zichzelf), depending
on the context. Zich is preferred in non-contrastive, and zichzelf in contrastive
situations. According to Reuland (2001), zich encodes the formation of a referential
dependency in narrow syntax. Concretely, zich enters a feature checking relation
with its local antecedent, resulting in the establishment of an A-Chain between
zich and its antecedent. Zichzelf, on the other hand, is identified with its
local antecedent through processes that lay outside narrow syntax.
19 Dutch children participated in a story elicitation task eliciting zich and
zichzelf. They correctly produced SELF-anaphors in 72.7% of the zichzelf situations.
Whereas they realized zich in only 30.6% of zich contexts, producing non-anaphoric,
or empty elements. Children's poor performance on zich shows that they have
problems with the use of syntactic mechanisms (i.e., feature checking and A-Chain
formation) to establish referential dependencies (Avrutin 1999). The results
also show that children correctly differentiate between the discourse situations,
which is important since this can hardly be learned from input.
Acquisition of copulars ser and estar in Spanish: Learning lexico-semantics,
syntax and discourse
Christina Schmitt, Carolina Holtheuer, and Karen Miller Two experiments (PMT and Acceptability Tast) examine Chilean Spanish-speaking
children's knowledge of the be vers ser/estar. It is well-known in the linguistics
literature, however, that the distribution of the be verbs ser/estar in Spanish
depends on lexical and syntactic properties of the predicates they appear with
and on discourse conditions. We hypothesize that, if discourse conditions are
harder to master, then children should master first the other constraints and
if children default to the unspecified form, they should default to ser. PMT
examines children's ability to use the context to make a choice. The AT examines
children's abililty to choose the copula based on lexical, syntactic and/or
contextual information. Our results show that children use the unmarked ser
as a default when the context is the sole determinant of the copula but not
when other constraints can be used to guide the choice. Influence of Parental Input on Learning Argument Structure
Constructions
Nitya Sethuraman How do cues change in mothers' speech early in development? Does changing input
help children learn argument structure patterns? Mothers' speech to 20- and
28-month-olds was examined longitudinally (Bates et al. 1988; MacWhinney 1995).
Specific patterns are used with one single high-frequency verb in mothers' speech
to 28-month-olds (Goldberg, et al. In Press). Children learn to associate the
meaning of the pattern with the meaning of that verb, thereby progressing from
specific knowledge of individual verbs to more general knowledge of argument
structure (e.g., Bowerman 1982; Tomasello 2003). Mothers addressing younger
children use one single verb in higher frequency in a particular pattern. Using
a high-frequency token more often in a certain pattern may help younger children,
barely starting to use the pattern, lock onto the meaning of that pattern more
efficiently; older children already using the pattern are then provided with
a larger variety of verbs in the pattern.
Cornell University and MIT
City University of New York
Truman State University
University of Hawaii, Manoa
University of Potsdam, Jürgen Weissenborn, Humboldt University, Berlin
Keio University, University of Tokyo, and Tokai University
The CUNY Graduate Center, New York
Keio University
Nanzan University, Nanzan University, and Massachusetts Institute of Technology
Taroo-Nom doll -DAT shoes-Acc put.on-cause-Past
'Taroo put the shoes on the doll.'
Taroo-Nom Hanako-Dat book-Acc read-cause-Past
'Taroo made/let Hanako read a book.'
University of Ottawa
University of Toronto, Michigan State University, University of Toronto
Cornell University
University of the Aegean, Leiden University
Ultrecht University
Michigan State University, University of Canberra, Michigan State University
Indiana University, Bloomington