Invited Colloquium II

"Comparing child L2 and SLI: Crosslinguistic perspectives"
Organizer: Theres Grüter (Stanford University)

Recent work in a variety of linguistic frameworks has shown remarkable similarities between children acquiring a nonnative language (L2) and children diagnosed with Specific Language Impairment (SLI): Similar grammatical phenomena appear to be vulnerable in both cases. The aim of this panel is to bring together researchers working on child L2 and SLI in different languages within a linguistic framework, in order to address and discuss questions such as the following:

- To what extent are similarities/differences between child L2 and SLI observed crosslinguistically?
- What domains of the grammar seem to be particularly (in)vulnerable in child L2 and SLI crosslinguistically?
- Are there aspects of grammatical development that clearly distinguish child L2 learners from children with SLI?
- To what extent are the vulnerabilities grammatical phenomena and/or processing phenomena?
- What are the implications of these similarities/differences for developmental theories of child L2 and SLI, and for linguistic theorizing more generally?

Presentation schedule:
8:30-8:40 Introduction (TG)
8:40-9:05 Rothweiler and Chilla
9:05-9:30 Marinis et al.
9:30-9:55 Scheidnes et al.
9:55-10:20 Simon-Cereijido and Gutiérrez-Clellen
10:20-10:40 COFFEE BREAK
10:40-11:05 Orgassa et al.
11:05-11:30 Armon-Lotem and Walters
11:30-11:45 General discussion

Paper 1:
Comparing child L2 and SLI: The acquisition of German sentence structure
Monika Rothweiler and Solveig Chilla (Hamburg University)

Paper 2: Comprehension of pronouns/reflexives in L2 children compared to children with SLI
Theodoros Marinis, Vasiliki Chondrogianni, Halit Firat (University of Reading)
Paper 3:
Acquisition of Wh-Questions in French: L2 Children and L1 Children with SLI
Maureen Scheidnes, Sandrine Ferré, Martin Haiden, Philippe Prévost, and Laurie Tuller (François Rabelais University, Tours)
Paper 4: Argument Structure in typical and atypical English L1 and L2
Gabriela Simon-Cereijido (San Diego State University/University of California, San Diego) and Vera Gutiérrez-Clellen (San Diego State University)

Paper 5: A double delay in L2-SLI acquisition: Evidence from Dutch agreement inflection
Antje Orgassa, Jan de Jong, Anne Baker and Fred Weerman (University of Amsterdam)
Paper 6: Indicators of SLI in bilingual children: Inflections and preposition
Sharon Armon-Lorem and Joel Walters (Bar Ilan University, Israel)

Paper 1: Comparing child L2 and SLI: The acquisition of German sentence structure
Monika Rothweiler and Solveig Chilla (Hamburg University)

The first authors discussing striking similarities between SLI and (child) L2 acquisition were Håkansson & Nettelbladt (1993, 1996), followed by Paradis & Crago (2000). Taking into account that simultaneous bilinguals pass through the same developmental sequences as monolinguals respectively, it is not reasonable that similarities between (child) L2 and SLI are an effect of the acquisition of an additional language in bilinguals. One hypothesis may be that (child) L2 learners as well as SLI children have only restricted access to first language learning mechanisms. Even though the observed differences to L1 appear in the same (vulnerable) domains (Platzack 2001), the causes are different in L2 and SLI. We assume that the observed differences between L1 and cL2 result from a changing access to first language learning mechanism, meaning that the differences between (2)L1 and cL2 should extend with growing age. The aim of this paper is to show that age of onset of L2 acquisition matters for the question of which subgroup of cL2 shows similarities to SLI.

Up to now, only few studies deal with the early second language acquisition of German in childhood. Following these studies, early successive acquisition (age of onset, AO 3-4) does not differ from L1 with respect to the acquisition of German sentence structure (CP) and related grammatical aspects (finiteness distinction and verb placement see Kroffke 2006, Prévost 2003, Thoma & Tracy 2006, wh-questions see Bonnesen & Chilla 2008, negation see Kroffke & Rothweiler 2006, subordinate clauses see Rothweiler 2006). Monolingual German children with SLI, however, persistently show problems in the acquisition of V2 and associated morphosyntactic aspects (Clahsen et al. 1997).

In this paper, we are going to present data from 4 monolingual SLI children and 4 cL2 (AO3) children. The cL2 participants are within their first and second year of consistent exposure to German. First, we are going to show that cL2 (AO 3-4) is a variant of 2L1 with respect to the acquisition of German sentence structure, i.e. for the acquisition of CP (V2 in main clauses, subordinate clauses, negation, subject realisation). The two groups of children, cL2 and SLI, differ with respect to clinical markers of SLI, especially non-finite elements in V2. These results will be contrasted with results from Kroffke (2006) who shows that children acquiring German successively (AO6) exhibit more similarities to adult L2 and thus to SLI than the younger cL2 children (AO 3).

[Back to top]


Paper 2: Comprehension of pronouns/reflexives in L2 children compared to children with SLI
Theodoros Marinis, Vasiliki Chondrogianni, Halit Firat (University of Reading)
Recent child L2 acquisition studies have focused primarily on language production and have shown 1) L1 to L2 transfer effects at early stages, 2) difficulties with morpho-syntax, and 3) similar pattern of performance to children with Specific Language Impairment (SLI) (Chondrogianni, 2007; Hakansson & Nettelbladt, 1996; Haznedar, 1997; Paradis, 2005; Whong-Barr & Schwartz, 2002). To date, only one study has investigated L2 children’s off-line comprehension showing that despite L2 children’s problems in production, their off-line comprehension of clitics is target-like (Grueter, 2005). Finally, two studies investigating L2 children’s on-line comprehension have showed no qualitative differences between L2 and L1 children in the processing of actives/passives and pronouns/reflexives, but differences in off-line accuracy (Marinis, 2007; Marinis, in press). The present paper investigates off-line comprehension of pronouns and reflexives in L2 children using the revised version of the Advanced Syntactic Test of Pronominal Reference (A-STOP-R) (van der Lely, 1997) that has been previously used on children with SLI (van der Lely & Stollwerck, 1997). 38 6-to-8 year-old Turkish-English children and 34 controls matched on age and socio-economic background participated in this study. All Turkish-English children were successive bilinguals, i.e., they grew up in a Turkish-speaking family and started having systematic exposure to English after the age of 3 when they went to the nursery/school. All children underwent a battery of tests to assess their verbal and non-verbal abilities. A-STOP-R is a picture-sentence pair judgment test involving pronouns and reflexives using subordinate clauses, such as in (1) below.

(1) The dog says that the cat is touching him/himself.

For the sentences with pronouns, mismatch pictures involved a reflexive action, whereas for the sentences with reflexives, mismatch pictures involved a transitive action. The results showed that L2 children were overall slightly less accurate than L1 children. Both groups were more accurate when the sentence matched the picture than when there was a mismatch between sentence and picture, and both groups were equally accurate in pronouns and reflexives in the matching conditions. However, there was a significant difference between the groups in the mismatch conditions. L1 children performed significantly better in reflexives compared to pronouns (Delay of Principle B effect), whereas L2 children performed significantly better in pronouns compared to reflexives. The pattern of the L2 children differs from the pattern of the children with SLI from the van der Lely & Stollwerck study. The children with SLI showed a similar pattern to the L1 children, but lower accuracy over the board. L2 children showed a disproportionately low accuracy in the reflexive mismatch condition compared to both the L1 children and the children with SLI. This relates to the status of the pronominal elements in the children’s L1s; the so called reflexive pronoun kendisi in Turkish unlike the English himself can be bound by an NP in the embedded or in the main clause. Our findings demonstrate differences between L2 children and children with SLI in the comprehension of pronominal elements and reveal transfer/misanalysis effects at later stages of child L2 acquisition.
[Back to top]


Paper 3: Acquisition of Wh-Questions in French: L2 Children and L1 Children with SLI
Maureen Scheidnes, Sandrine Ferré, Martin Haiden, Philippe Prévost, and Laurie Tuller (François Rabelais University, Tours)

Comparing children with SLI and children acquiring the same language as an L2 has recently been shown to provide insight for understanding acquisition and the syntactic operations involved in key areas of grammar (Grüter, 2005; Håkansson, 2001; Paradis, 2004; Paradis & Crago 2000, 2001). These children are similar and different in crucial ways, which allows for disentangling of often related variables such as pathology, timing of acquisition, L1 transfer, etc. In this paper, we propose to enrich the SLI/L2 comparison paradigm by focusing on the acquisition of wh-questions in French.

These constructions provide particularly fertile ground for investigating the hypothesis that derivational computational complexity predicts which areas of grammar are harder to acquire (Jakubowicz, 2004, 2005). Unlike English, a wide variety of syntactic strategies are used in French to derive wh-questions, not all of which involve syntactic movement:
(1) a. Quii asj-tu tj vu ti?
b. Quii tu as vu ti?
c. Quii est-ce que tu as vu ti?
d. C’est qui que tu as vu?
e. Tu as vu qui?
(2) a. Whoi havej you tj seen ti?
b. *Whoi you have seen ti?
c. Whoi is it that you have seen ti?
d. *It’s who that you have seen?
e. *You have seen who?
These different strategies correspond to varying degrees of computational complexity, where the most complex strategy in French (1a) corresponds to the most common strategy in English (2a), and the least complex strategy in French, wh-in-situ (1e), is ungrammatical in English as a non-echo question (2e) (Jakubowicz 2004, 2005, 2007, to appear; Hamann, 2006; Strik, 2002, 2003; Zuckerman, 2001). These studies show that children learning L1 French tend to avoid constructions involving movements, with high rates of wh-in-situ questions in typically developing children and even higher rates in children with SLI. In English-speaking learners of French, the in-situ strategy is preferred initially by adults (Scheidnes 2007), but develops late in children (Grondin & White 1996). However, only two children were investigated, and the data came from spontaneous production only.

We conducted a longitudinal study comparing the development of wh-questions in French-speaking children with SLI (n=30) and British immigrant children acquiring French in France (n=15), using an elicited production task and a comprehension task (Jakubowicz 2006). In the former, six wh-constituents were tested (subject, animate and inanimate objects, adjunct-where,-how,-why) (54 items in total). Children had to ask questions about missing images of corresponding arguments/adjuncts in pictures depicting transitive verbs. In the comprehension task, only argument questions were used (n=72), but with different strategies (e.g. wh-anteposition and wh-in-situ). Questions were asked about pictures representing three characters involved in an action, and children had to point to the characters targeted by the questions. The British children were further assessed for English via the CELF-4 Core Language Subtests, and all children were assessed for French via standardized measures of vocabulary, phonology, and morphosyntax (BILO, Khomsi et al., 2007). The results will be presented and analyzed from the vantage point of derivational computational complexity.

[Back to top]


Paper 4: Argument Structure in typical and atypical English L1 and L2
Gabriela Simon-Cereijido (San Diego State University/University of California, San Diego) and Vera Gutiérrez-Clellen (San Diego State University)
Studies with English- and Spanish-speaking children have shown that children with language impairment (LI) tend to use a limited number of three-argument verbs and present with occasional argument omissions (Grela & Leonard, 1997; Simon-Cereijido & Gutierrez-Clellen, 2007; Thordardottir & Weismer, 2002). In addition, measures developed for monolingual English speakers (such as finite verb morphology tasks) do not assist in the differential diagnosis of children with LI who are bilingual, speak a nonmainstream English dialect, and/or are English as a Second Language (EL2) speakers (Gutierrez-Clellen& Simon-Cereijido, 2007; Gutierrez-Clellen, Simon-Cereijido, & Wagner, 2008; Paradis, 2005). Intact knowledge of phrase structure in their first language may help EL2 children learn to use complex phrases, such as ditransitive predicates, in English. The purpose of this study was to explore whether spontaneous narrative measures of verb and argument use reveal differences between children with and without LI who speak English as their first language (EL1), and between typical EL1 and EL2 children.

Fifty-six Mexican-American children were divided in 3 age-matched groups (18 English speakers with typical language development (EL1TLD), 18 English speakers with LI (EL1LI), and 18 EL2 children with TLD (EL2TLD); mean age = 71.3 (9.5) months). Children with LI met criteria based on parent and/or teacher concern, and performance on the English morphosyntax subtest of the Bilingual English Spanish Assessment (Peña, Gutierrez-Clellen, Iglesias, Goldstein, & Bedore, n.d.). Children had no evidence of other special needs based on parent report, teacher report, and school records.

A method for calculating a total argument structure (TAS) score was devised based on narratives elicited with wordless picture books (Mayer, 1973, 1975). The storybook plates that elicited three-argument predicates were identified with an independent sample of children randomly selected from a larger study. A rubric based on the complexity of the verb was then created. The maximum point per target was 4 (verb and 3 arguments) and the minimum point per target was 0 (target or related verb was not used). Morphological errors were not penalized. Reliability was calculated by two graduate students who independently coded and scored the samples with 92% agreement. In addition, grammaticality (percentage of grammatical utterances) was separately calculated following the Systematic Analysis of Language Transcripts conventions (Miller & Chapman, 2000).

As expected, the EL1TLD grammaticality scores were significantly higher than both the EL1SLI group (t(34)=5.287, p<.001) and the EL2TLD group (t(34)=7.007, p<.001). In contrast, the TAS score of the EL1TLD group was significantly higher than the EL1SLI children (t(34)=2.264, p =.03), but not significantly different than the EL2TLD group (t(34)=1.985, p=.06). See Table 1 for means and SDs.

The language ability comparisons supported previous research on children with LI and EL2 speakers. A morphological measure (e.g., grammaticality) does not help differentiate between children with LI and nonnative English speakers. In contrast, verb and argument use appears to have the potential to tell apart these two groups. Although this study’s measure is limited, the results suggest that elicited tasks of
ditransitive predicate contexts may be promising unbiased indicators of LI in English.

[Back to top]


Paper 5: A double delay in L2-SLI acquisition: Evidence from Dutch agreement inflection
Antje Orgassa, Jan de Jong, Anne Baker and Fred Weerman (University of Amsterdam)

With the aim of specifying the relationship between SLI and L2 acquisition, two perspectives on the nature of Specific Language Impairment (SLI) are tested: the agreement-deficit account (Clahsen 1989; 1993) and the generalized processing account (e.g. Miller, Kail, Leonard and Tomblin 2001; Ellis Weismer and Evans 2002). This will be done by comparing the acquisition of Dutch agreement inflection as marked on gender and verbs in four child learner groups. These linguistic domains represent well-known vulnerable areas in SLI and L2, respectively. Verb agreement is particularly vulnerable in Dutch SLI. In the nominal domain, attributive adjectival inflection is a locus of difficulties in (child) L2 acquisition. Data analysis focused on error types in terms of qualitative and quantitative similarities and differences.

If SLI is assumed to entail a deficit in linguistic representation such that one or more principles of Universal Grammar are not available and if acquisition of grammar is age-dependent, then children with SLI and older L2 learners might have to rely on other learning mechanisms and should pattern qualitatively different from unimpaired L1 and L2 children. If, on the other hand, SLI is assumed to be the result of limitations in natural processing capacities, then SLI children and their unimpaired L1 and L2 peers should rely on the same learning mechanisms, the only difference being that the SLI children are delayed in development. Consequently, we predict that all child groups produce the same error types. This perspective allows for further specification of the impact of bilingualism. As in SLI, L2 children have a poorer intake (albeit for different reasons) than unimpaired L1 peers: The reduced intake is related to the fact that these children have had uneven exposure to two languages resulting in a reduced input in their L2. Accordingly, we also predict that L2 children with SLI (L2-SLI) should show a delay compared to all other child groups marking a cumulative effect of SLI and bilingualism.

To test the predictions, we conducted experiments on several groups: child L1 (n=55; 4-8y); child L1-SLI (n=25; 6-8y), child L2 (n=20; 5-8y) and child L2-SLI (n=20; 6-8y). The L2 participants all have Turkish as L1 to control for possible transfer effects. Group comparisons revealed that the representational deficit account is not supported by our data since all child groups pattern alike in terms of error types but are different from what is known about acquisition in older L2 learners (e.g. Blom, Polišenská and Weerman 2006/2007). The observed delay (i.e. higher error frequencies) in the SLI and child L2 groups as compared to the unimpaired L1 controls is in line with the processing account. On the basis of that evidence, it appears that it is problems with the intake (in SLI and L2) rather than access to grammatical principles that explain differences between the groups. The fact that we also find a cumulative effect of bilingualism and SLI supports this: the L2-SLI group not only differs from the child L2 controls but also from the Dutch L1-SLI group.
[Back to top]


Paper 6: Indicators of SLI in bilingual children: Inflections and prepositions
Sharon Armon-Lorem and Joel Walters (Bar Ilan University, Israel)
Substitutions and omissions of inflections are often used as indicators of Specific Language Impairment (SLI), while prepositions are a locus for code interference in bilingual populations, but are not necessarily considered an indicator of SLI. Using an elicited imitation task, we show that while inflection errors demonstrate quantitative differences between SLI and typically developing bilingual children, obligatory prepositions provide a qualitative measure showing unique SLI errors. Subjects were preschool sequential bilinguals from English-speaking homes who had been exposed to L2 Hebrew for more than two years. Eight language impaired children showing atypical language development (ATD) in both languages, are compared with eleven typically developing (TD) children from the same neighborhood, who scored within the norm in both languages. Inflections in Hebrew were tested targeting past tense morphology in 1st and 2nd person, and in English targeting third person morphology in the present tense and regular past tense. Prepositions were tested in both languages targeting free (non-subcategorized) prepositions, e.g., at school, and obligatory prepositions, e.g., laugh at, which serve a grammatical function.

Significant quantitative difference was found between the two groups in the proportion of inflection errors in both languages. For both groups, past 2nd person morphology was the hardest in Hebrew and tense and person inflections were most difficult in English. Some errors were unique to the ATD group, however. In Hebrew only ATD children omitted the inflections, while in English only ATD children substituted –ing for 3rd person -s. For preposition use, no significant difference was found between TD and ATD children in the proportion of errors. Both groups made errors in the use of prepositions due to code interference, but only the ATD children showed errors which could not be explained by crosslinguistic influence. This was true for both languages, and a significant difference was found in both languages for obligatory prepositions. That is, both groups made errors which reflect crosslinguistic interference, such as omission of inflection in the third person in English, where Hebrew has no overt inflection, as well as substitution of person in Hebrew, due to the absence of person features in English, but the proportion of errors for ATD children was higher as reported for monolingual SLI. Moreover, while code interference in the use of prepositions is typical of bilingual children, the unsystematic substitution and omission errors in the use of obligatory prepositions are unique to ATD.

Restriction of the unsystematic substitutions and omissions to obligatory prepositions can be explained by their central grammatical role in case assignment, with a limited contribution to the semantics of the sentence. Similarly, the substitutions of –ing in English are in places where the inflection does not add to the meaning of the sentence, whereas the inflections omitted and substituted in Hebrew carry the semantics of person and tense. In the bilingual context, when L1 has no person features, the ATD children seem to associate these suffixes with tense only, which is also encoded in Hebrew by the interdigited vowel pattern, making the suffix semantically redundant.


[Back to top]


 

 

Home | Organizers | Sponsors | Call for proposals | Registration | Conference program | Social events | Transportation |
Accommodations | Hawai‘i links | Call for 2009 SLRF host | Past SLRF | SLRF '08 FAQs | Presentation guidelines |

If you need additional information, please contact us.
E-mail: slrf2008@gmail.com
Mail address: SLRF 2008, Department of Second Language Studies, University of Hawai'i
1890 East-West Rd. Honolulu, HI, 96822, USA. Fax: 1-(808) 956-2802
Webdesign by Marta Gonzalez-Lloret, Deborah Masterson, and Yukiko Watanabe (Webmaster)