In her Keynote Article, Paradis discusses the role of the interface between bilingual development and specific language impairment (SLI) on two different levels. On the level of theoretical explanations of SLI, Paradis asks how domain general versus domain-specific perspectives on SLI can account for bilingual SLI, as well as what bilingual SLI may contribute to the discussion of these theories. Paradis argues in favor of domain-specific deficits (in addition to well-documented processing deficits in SLI), and especially for the maturational model (Rice, Reference Rice, Rice and Warren2004). She argues against a mere processing deficit of input information and against deficits in working memory and processing speed as sole sources of SLI. On the practical level, Paradis focuses on the question of whether and how language tests that have been standardized for monolingual children are valid for the assessment of SLI in bilingual children. Both discussions, on theory and on practice, are based on empirical data from Canadian studies on bilingual children with and without SLI carried out by Johanne Paradis, Martha Crago, and Fred Genesee (e.g., Crago & Paradis, Reference Crago, Paradis, Levy and Schaeffer2003; Paradis, Reference Paradis2007; Paradis, Crago, & Genesee, & Rice, Reference Paradis, Crago, Genesee and Rice2003).

In the following, I comment on additional research topics that might be pertinent to the study of SLI in bilinguals. Paradis emphasizes the value of comparing bilingual SLI with monolingual SLI and unimpaired child second language (L2) to test theories on the nature of SLI by focusing on the interface of bilingualism and SLI, as well as evaluating commonalities and differences between language acquisition in different acquisition settings. I want to draw attention to another research option offered by the study of bilingual SLI. The fact that bilinguals acquire two different languages and that bilinguals with SLI experience an impaired language acquisition in these two languages enables us to use their data for cross-linguistic studies on SLI.

One major question for cross-linguistic comparisons of SLI is to what extent deviations in morphosyntax in SLI across languages can be explained by the typological characteristics of the languages. These insights may then reveal a basic (grammatical) impairment common to SLI in all languages. This is a central objective of cross-linguistic studies on SLI (Clahsen, Reference Clahsen, Ball, Perkins, Mueller and Howard2008; Crago & Paradis, Reference Crago, Paradis, Levy and Schaeffer2003; Leonard, Reference Leonard, Levy and Schaeffer2003). Cross-linguistic studies on SLI always run the risk of comparing apples to oranges: because of the heterogeneity of SLI, we never know for sure whether two SLI groups of different languages contain the same type of impaired children (Friedman & Novogrodsky, in press; van der Lely, Reference Van Der Lely, Levy and Schaeffer2003). This is especially relevant for comparisons of SLI in languages that are not as well studied as the main European languages, American English and Spanish, or as Hebrew, or the recently intensively studied Chinese languages (Fletcher, Leonard, Wong, & Stokes, Reference Fletcher, Leonard, Wong and Stokes2005; Law, Weekes, & Wong, Reference Law, Weekes and Wong2009). Studies on SLI in other languages are still few in number; see, for example, Crago and Allen (Reference Crago and Allen2001) for Inuktitut, Demuth and Suzman (Reference Demuth, Suzman, Hughes, Hughes and Greenhill1997) for Zulu, or Abdalla and Crago (Reference Abdalla and Crago2008) for Arabic.

The studies on simultaneous bilinguals with SLI, many of them carried out by Paradis and her colleagues, focus on both languages, and they can be summarized using the (somewhat simplifying) formula that balanced bilinguals with SLI are like monolinguals with SLI in both languages (for a discussion, see Paradis, Reference Paradis2007, and Keynote). These children are equally impaired in both languages and to the same extent as that known from monolinguals. This result constitutes the starting point for two further types of SLI studies in bilinguals. First, the study of SLI in bilinguals is the purest form of a comparative study of SLI in two languages, namely, the comparison of two languages within an individual. This individual level of comparison guarantees the two languages being affected by the same subtype of SLI. Furthermore, such cross-linguistic studies use the same criteria for the linguistic analysis of both languages. These points are relevant for studies aiming for a comprehensive description and explanation of the grammatical deficit in SLI beyond the specific grammatical deviations in the individual languages. Second, we may use the study of bilinguals with SLI as an effective setting for the investigation of SLI in languages for which we do not yet know how SLI is expressed. In this case, we would concentrate on bilinguals with one well-studied language and one language that has yet to be examined with respect to SLI. One of the benefits of this type of intraindividual cross-linguistic studies is that the identification of children with SLI in the language of focus will be supported by the identification of SLI in the other, well-studied language. SLI is identified via the well-studied language (according to the usual exclusionary and inclusionary criteria), and the other language becomes the subject of investigation. Especially the combination of languages from different language families should be at the center of this type of cross-linguistic study. One interesting language combination in Europe would be Basque–Spanish.

Many languages of the countries of the so-called Western world are well studied with respect to acquisition and SLI. Many of these countries are countries that attract large numbers of immigrants. The children of these immigrants, and often their descendents in the next generations as well, grow up bilingual, and some of these children are children with SLI. For many of the first languages of migrants and their children, we do not know how exactly SLI manifests itself. In this way, relevant language pairs offer the potential for the cross-linguistic study of SLI, like Arabic–French in France; Moroccan–Dutch and Turkish–Dutch in The Netherlands; Turkish–German in Germany; English–Hebrew and Russian–Hebrew in Israel; combinations of Urdu, Hindi, or Punjabi with English; and combinations of Asian and African languages with Western European languages. One relevant objection to this suggestion might be that bilingual children with such language combinations are mostly successive learners of the majority language. Using the language profile and proficiency in a well-studied language as decisive criteria for a child being labeled specifically language impaired conflicts with the child learning this language successively. In her Keynote, Paradis compares the results of studies on simultaneous bilinguals to studies on successive bilinguals with and without SLI. Comparing and weighing the results, she concludes that claims of the cumulative effects of child L2 difficulties and SLI deviations have hardly been substantiated. She proposes several factors that have to be taken into account for the comparison of the data of first language (L1)–unimpaired, L1–SLI, cL2–unimpaired, and L2–SLI children, because these factors influence the language proficiency of child L2 learners. These factors are age of onset, length of exposure, amount of exposure, and the majority or minority sociolinguistic status of the bilingual's two languages.

Rothweiler, Chilla, and Clahsen (Reference Rothweiler2009) compared the acquisition of subject–verb agreement (SVA) and the verb second rule in the spontaneous speech of Turkish–German successive bilinguals with SLI (ages 5–8, age of onset ~3) with monolingual SLI children of the same age. The results of the bilinguals with SLI match the results of the monolingual children with SLI after 18 to 24 months of exposure, as unimpaired successive bilinguals (age of onset 3) match the developmental course of monolingual unimpaired learners (Rothweiler, Reference Rothweiler and Lleó2006). One of the relevant factors is age of onset. If the acquisition of a L2 starts at age 3 or 3.5 years at the latest, the acquisition of morphosyntax parallels L1 acquisition in many respects (Meisel, Reference Meisel, Haznedar and Gavruseva2008, Reference Meisel2009; Rothweiler, Reference Rothweiler2009). The same holds for SLI in successive bilinguals compared to SLI in monolinguals (Rothweiler et al., Reference Rothweiler, Chilla and Clahsen2009). Thus, if we take the factors summarized by Paradis into consideration, especially age of onset and length of exposure, bilingual successive children with SLI may be considered just as relevant a group for intraindividual cross-linguistic studies on SLI as simultaneous bilinguals. Most studies on successive bilinguals with SLI focus on the acquisition of the L2 and compare it to SLI in monolinguals and to unimpaired child L2 (as Paradis summarized in her Keynote). In addition, we should turn our attention to the L1s of these children, whereas the L2s, the majority languages, should be used for identifying children with SLI and as contrastive languages.

There are ongoing research projects on bilingual SLI involving certain language combinations of the sort described here.Footnote ¹ In these projects, SLI is investigated in both of the children's languages; see, for example, the two research projects on bilingual SLI in Turkish–Dutch children (Orgassa & Weerman, Reference Orgassa and Weerman2008) in Amsterdam and in Turkish–German children in Hamburg (Chilla, Reference Chilla2008; Rothweiler, Reference Rothweiler and Lleó2006).

The Hamburg project studied case morphology and person and number markings in Turkish. Turkish is an agglutinative language with rich case marking and verb morphology. The morphological markings are regular and phonetically salient with a low rate of homophoneous affixes. Both systems are acquired early by unimpaired children, that is, before or around the age of 3. In this way, these systems should be in place before the successive bilinguals start to acquire German. This holds for the unimpaired children: the development of case and SVA morphology (person or number morphemes) parallels what is known from Turkish monolingual children in Turkey (Chilla & Babur, in press). For the SLI children, we found that the same children suffer from deficits in person or number verb morphology in Turkish (affix omissions) and from a SVA/finiteness deficit in German (Babur & Rothweiler, Reference Babur and Rothweiler2007; Chilla, Reference Chilla2008; Chilla & Babur, in press). As in monolingual SLI, a deficit in SVA (i.e., in person or number marking on the verb) is a clinical marker for SLI in German in bilingual children (Rothweiler et al., Reference Rothweiler, Chilla and Clahsen2009). These results deliver further evidence for the SVA deficit account proposed by Clahsen, Bartke, and Göllner (Reference Clahsen, Bartke and Göllner1997). Like the account of Rice (Reference Rice, Rice and Warren2004), this account argues for SLI as a domain-specific deficit. The SVA deficit account does not only apply for German. The results indicate that it also accounts for the person or number deficit in the nonrelated language Turkish. Furthermore, case morphology in Turkish is affected by SLI, and we observed omissions as well as comission errors in spontaneous speech data (Babur, Rothweiler, & Kroffke, Reference Babur, Rothweiler and Kroffke2007; Rothweiler, Chilla, & Babur, in press). To conclude, in Turkish, case morphology as well as verbal morphology (ps/n) are additionally delayed compared to mean length of utterance controls. Concentrating on verbal morphology, we can conclude that SLI affects finite verbs, especially SVA, in both languages, thus adding further evidence to support the hypothesis that the deficit in finiteness markings is a general clinical marker for SLI.

Paradis discusses the interface between bilingual development and specific language impairment. She presents relevant and convincing evidence and arguments from the comparative study of SLI and bilingualism and of bilingual SLI for the characterization of SLI as a domain-specific linguistic deficit (in addition to processing deficits). She thus prepares the field for cross-linguistic studies in bilinguals with SLI. Both lines of research will contribute to the search for the linguistic nature of SLI.