BUCLD 32 Online Proceedings Supplement
Edited by Harvey Chan, Enkeleida Kapia, Heather Jacob
Acquisition of Japanese wh-questions: The effects of processing strategies on L2 sentence judgment
abstract | paper
Object clitic pronouns, definite articles and genitive possessive clitics in Greek preschool children with Specific Language Impairment (SLI): implications for domain general and domain-specific accounts of SLI
Nafsika Smith, Susan Edwards, Vesna Stojanovik, and Spyridoula Varlokosta
abstract | paper
When do infants develop an adult-like lexicon, with an interconnected structure? In Primed Preferential Looking, a picture is labelled shortly before dual image presentation. The target word is preceded by a ëprime,í and prime-to-target relationships are manipulated. In the current study, both prime word and target word were manipulated (Prime: related, unrelated; Target: label, no-label), allowing the independent contributions of ‘naming’ and ‘priming’ to be observed. 72 infants at 18-, 24-, and 30- months-of-age participated. In 12 trials, each infant saw both priming and both target conditions, with no stimulus repetition. Eye movements were recorded, and coded offline, frame-by-frame. Across all age groups, effects of both ‘namingí and ‘priming’ were evident. Detailed analysis of the time- course provides support for a primed lexical processing account.
Narrative competence includes the discourse-pragmatic ability to represent events, actors, and motivations from an evaluative-interpretive perspective (Bamberg and Damrad-Frye 1991). Comparing evaluative content in bilingual and monolingual children’s narratives, we examine whether bilingualism affects the use and functions of evaluative devices in narrative performance. Data were elicited from ten subjects (six Hungarian-English bilinguals, two monolinguals in each language), ages six to nine, using the picture book Frog, Where Are You? (Mayer 1969). The analysis indicates a bilingual advantage with respect to the number and range of evaluative devices, an advantage that increases with age. The variety of evaluative devices also correlates with the use of advanced vocabulary. Only bilinguals use voicing strategies, and mostly so when narrating in Hungarian. Appraisal, however, is more frequent in English. We argue that awareness of an evaluative-interpretive perspective in constructing narratives constitutes an aspect of discourse-pragmatic competence where bilinguals have an advantage over monolinguals.
Caregivers of 62 toddlers (Mean age = 22.55 months, SD = 3.90) acquiring English in bilingual homes provided information on home language use and on the toddlers’ English vocabulary development using the MacArthur-Bates Communicative Development Inventory. The overall portion of the children’s input that was in English was significantly related to the children’s English vocabulary percentile scores r (60) = .34, p = .007. For those toddlers with older siblings, the percent of sibling input that was in English was a significant correlate r (51) = .411, p < .05. CDI percentile scores were higher for children who spoke to their sibling(s) in English (M = 45.45, SD = 27.88, n = 11) than in Spanish (M = 5.00, SD = 0.00, n = 2) or a mix of English and Spanish (M = 20.00, SD = 26.01, n = 16), F (2, 26) = 5.21, p < .01.
This study investigates the sensitivity to two Cantonese elliptical structures among Cantonese-speaking children aged between 4 to 6. Previous studies (Thornton and Wexler 1999, Matsuo and Duffield 2001, Foley et al. 2003) found that English-speaking children (4-6-year-olds) are sensitive to VP Ellipsis Construction (VPEC). Chinese has VPEC and Null Object Construction (NOC) (Li 2002, Xu 2003). Despite superficial resemblance, the two constructions differ crucially in the recovery of the adverb in the antecedent clause. The adverbial is recovered in VPEC, but not in NOC. In the experiments, the subject matched the stimulus sentences against act-out scenarios. The results show that they are sensitive to the different possibility of adverbial recovery in these constructions. The judgment is particularly robust in their interpretation of VPEc. The study suggests early acquisition of elliptical structures in Cantonese.
Mintz (2003) proposes that very local distributional contexts of words in the input-so-called ‘frequent frames’-function as reliable cues for categories corresponding to the adult verb and noun. He shows that categories resulting from frequent frames align with English grammatical categories for over 90% and that American 12-month-olds use these frequent frames to form a verbal category. Based on Dutch input and child data, I will show that frequent frames are not generally valid as a cue to categories.
In a replication of Mintz (2003) for the input to Dutch children, I found that the frame-based categories aligned with Dutch grammatical categories for only 40%-71%. Furthermore, Dutch 12-month-olds did not use these cues in an experiment designed parallel to Mintz (2006). Even Dutch 16-month- olds did not use the cues, although there was some development towards the English pattern between the two age groups.
Discussing Japanese-speaking children’s erroneous verbs and -sase causatives, Murasugi and Hashimoto (2004) proposes that the children have difficulty in assigning appropriate phonetic contents to the functional head small v’s associated with the features [Å}cause], and the “Verb- Functional Head” (See also Murasugi, Hashimoto and Fuji (2007)). Just as in causatives, they make some errors in potentials and passives. For example, some children omit the potential suffix -rare, intending to express potential meaning: “tabe-ru” instead of “tabe-rare-ru” (I can eat). Some children overgenerate the potential suffixes -rare as well as -e: “mot-e- rare-ta” instead of “mot-e-ta” (I could hold it). This paper discusses that given Murasugi and Hashimoto’s (2004) v-VP shell analysis of causative – sase complex predicates, the omission and the overgeneration phenomena observed in the intermediate acquisition stage of potential -rare complex predicates and passives can also be empirically and theoretically explained in a uniform way.
We report results of an experiment exploring the mass/count distinction in Hebrew-speaking school-age children and adults. Adopting Barner & Snedekerís (2005) experimental methodology, we tested how speakers use singular/plural morphology to distinguish nouns that quantify over individuals from nouns that do not. Our results show that Hebrew-speaking adults, categorically base their judgments on number of individuals in the count condition as well as in the flexible-count and the object-mass conditions, choosing the character with the most individual items around 100%. Conversely, in the substance-mass and the flexible-mass conditions, adult speakers clearly judge the character with the overall more volume to have “more”. While these results are essentially identical to B&Sís English-speaking adults, surprisingly, Hebrew-speaking school-age children behave very differently from both Hebrew-speaking adults and English- speaking preschoolers. We propose that this discrepancy is due to the relative scarcity of Hebrew structures encoding the mass/count distinction, which makes acquisition more laborious.
We study CHILDES data (MacWhinney 2000) for seven children and their parents, confirming Stephany’s (1986) findings that gonna precedes will in child language acquisition. We test two hypotheses about the emergence of will and gonna, based on Copley’s (2002) semantic analysis, which suggests will is the simple future, and that gonna consists of will scoped over by a progressive operator. The Frequency Hypothesis: There is more gonna in the input, resulting in earlier acquisition. This is rejected; parental data suggests that will is more prevalent. The Progressive Hypothesis: Children initially treat gonna as a simple progressive, before modality emerges. This is also rejected; gonna is used with states freely, and gonna is used with future meaning from earliest instances. We propose a refined semantic account; that will, and not gonna, involves contextual integration, accounting for the later acquisition of will.
The present study investigates how L1-Korean/L2-Japanese, and L1-Chinese/L2- Japanese learners perform with regard to Japanese wh-questions. L2 learners of Japanese have been said to resolve scopally ambiguous question fragments in a manner similar to native speakers; they use the processing strategy that requires shorter dependency between a wh-phrase and a question morpheme -ka (Lieberman et al., 2006). In this study, two experiments, testing both unambiguous and ambiguous sentences, were conducted to investigate (i) whether the learners can distinguish the difference between a yes/no-question and a wh- question, and (ii) what processing strategy affects L2 sentence judgment. We have demonstrated that (a) the learners distinguish the differences between two types of Japanese questions in the unambiguous cases, obeying the wh-island constraint, and that (b) not only the strategy to require shorter dependency but also the strategy related to scrambling affect the acceptability judgment of Japanese wh-questions.
We present a computational technique that automatically identifies a set of lexically-specific templates (linguistic constructions consisting of specific words combined with variable slots, e.g. “where’s your X ?”, “thatís a X one”) present in a corpus of child-directed English speech. Distributional information about the occurrence of single words in the slots of these templates is used to form simultaneous clusters of words and templates, corresponding closely to traditional lexical categories such as nouns, verbs and adjectives. Clustering information from both word and template is then used to assign lexical categories to specific instances of words occurring in template contexts. The resulting lexical category assignment is highly accurate when compared against a manual tagging of the corpus, and is able to deal correctly with words that can belong to more than one lexical category.
The acquisition of tense in morphologically rich languages has been shown to be relatively error-free and early. However, these data come primarily from spontaneous speech. The use of controlled experiments in Hebrew reveals a clear delay in the acquisition of tense.
We tested 57 TD monolingual Hebrew-speaking children aged 4;2-12;9 and 9 adult controls on their knowledge of tense inflection with a completion task. While all children were (near) adultlike in their retention of agreement morphology and verb pattern, the younger children produce appropriate (past and future) tense morphology only 63% of the time.
We argue that these errors result from the fact that the present tense in Hebrew is participial, and thus non-finite. When young children do not succeed in mapping past or future to the correct tense morphology, they rely on the non-finite properties of the present tense form, allowing for present, past or future interpretation.
This study analyzes child-directed ASL inside and outside episodes of joint attention at three to four ages (9-24 months) for each of three mother- child dyads. Signs were coded for modifications from their citation form: spatial displacement, lengthening, and repetition, among others. Our findings were consistent with the idea that joint attention episodes are a privileged state for vocabulary learning. At all sessions, the number of lexical sign tokens produced per minute was higher inside joint attention than outside joint attention. At most sessions, a higher percentage of the mothers’ sign productions were modified inside joint attention episodes. Modifications such as lengthening and repetition did not appear to function to attract an inattentive child’s gaze, as the child was already looking at the parent’s face or hands at the beginning of a majority of modified sign tokens. These modifications may serve other discourse functions, e.g., eliciting a child’s imitation.
Only recently has there been an increased interest in generative L3 acquisition studies (see Leung 2007 inter alia). Studying L3 acquisition is interesting in its own right, since there are different variables to consider; nevertheless, studying L3 acquisition can also provide insight into theoretical debates within formal approaches to adult L2A (cf. Leung 2005, 2007). In this paper, we test for nominal ellipsis (N-drop) at the initial state of two L3 groups: English-Spanish additive adult bilinguals (n=22) and English-Spanish successive childhood bilingual learners of L3 Portuguese (n=18). We compare these groups to an independent group of English learners of L2 Portuguese at the initial state (n=20). Both L3 groups (unlike the L2 group) demonstrate knowledge of N-drop at the initial state of Portuguese, suggesting that the additive Spanish bilinguals acquired the interpretable and uninterpretable Spanish gender and number features crucial to acquiring N-drop (White et al. 2004), which provides evidence against theories of partial access for L2 acquisition.
Object clitic pronouns, definite articles and genitive possessive clitics in Greek preschool children with Specific Language Impairment (SLI): implications for domain-general and domain-specific accounts of SLI
Authors: Nafsika Smith, Susan Edwards, Vesna Stojanovik (University of Reading, UK) Spyridoula Varlokosta (University of the Aegean, Greece)
Nine Greek preschool children with Specific Language Impairment (SLI) were compared to an age-matched (CA) and a language-matched (LA) typically-developing group on their production of pre-verbal object clitic pronouns, definite articles and genitive possessive clitics through novel picture-based elicitation tasks. The aims were to examine whether these structures are significantly impaired in Greek SLI, and to evaluate domain- general and domain-specific accounts of SLI (Surface Hypothesis, SH, Leonard 1989; Interpretability Hypothesis, IH, Tsimpli & Stavrakaki 1999; Representational Deficit for Dependent Relations, RDDR, van der Lely 1998).
The SLI group scored significantly below both control groups on object clitics but only below the CA on definite articles (errors consisted of both substitutions and omissions), while they did not differ from the groups on genitive possessive clitics.
The findings disconfirm accounts predicting similar impairment on object clitics and definite articles (SH, IH) and are more consistent with accounts predicting a difference between object clitics and definite articles (RDDR). A novel combinatory model combining the notions of interpretability and movement from the IH and the RDDR is proposed to explain the exact pattern of findings and further explanations are discussed.
Two theoretical approaches to gender and declension in adult grammar are evaluated in this paper: Declension-to-Gender (Corbett 1982) and Gender-to-Declension (Crockett 1976). These approaches differ in their acknowledgment of what comes first: can a noun’s gender be predicted from its declension, or can nominal declensional class paradigm be derived from the gender of a noun. This paper reports the results of an elicited production experimental study, which investigated what kind of context children use more readily for the successful acquisition of the novel nouns’ gender. The results support the hypothesis that children’s performance on gender assignment depends on the kind of exposure: the condition where the novel noun was introduced in the context of its declensional paradigm is more facilitating for correct gender assignment than the condition of adjectival agreement context. This difference is taken as supporting the Declension-to-Gender model.
Three components of determiner meaning has been identified: Truth conditions, implicatures, and presuppositions. In this paper, I investigate children’s understanding of presuppositions associated with jeder `every’ and beide `both’ in German, using Presupposition Judgment Task (30 children each for age 6, 7, 8, and 9). Heim (1991) proposes that there are two types of presuppositions, lexical and implicated. Implicated presuppositions are derived like implicatures. The result of the experiment shows that the lexical presuppositions of jeder (existence presupposition) is acquired much earlier than the implicated presupposition of jeder (anti-uniqueness presupposition). This is expected: Heim (1991) proposes that implicated presuppositions are derived, using the same mechanism as implicatures, and previous research on implicatures show that children have difficulties with implicatures (Noveck 2001), predicting that children have difficulties with implicated presuppositions as well. Children had difficulties with duality presupposition of beide `both’, however, although it is a lexical presupposition.
This paper addresses the issue of attrition of grammatical aspect and other grammatical categories in bilingual (Russian-English) adults and children. Prior research has shown aspectual, lexical, agreement and case marking errors in this population. Of specific interest was a possible hierarchy of attrition in various areas of L1, based on the length of uninterrupted exposure to L1. An elicited narrative (One Frog Too Many by M. & M. Mayer) and a Grammaticality Judgment task were used to address the participantsí competence in Russian grammar. Our results indicate that both groups show signs of attrition in all above-mentioned grammatical categories. Significant between groups difference were found in the magnitude of attrition, based on the length of exposure to L1. However, the aspectual errors were seen only in children immersed in L2 (English) from birth. Thus, our results indicate that grammatical aspect may be the least sensitive category of attrition.