Cormier, K., Schembri, A., Vinson, D., & Orfanidou, E. (in press). First language acquisition differs from second language acquisition in prelingually deaf signers: Evidence from sensitivity to grammaticality judgement in British Sign Language. Cognition. Accepted 13 April 2011. First language acquisition differs from second language acquisition in prelingually deaf signers: Evidence from sensitivity to grammaticality judgement in British Sign Language Kearsy Cormier, Adam Schembri1, David Vinson, Eleni Orfanidou2 Deafness, Cognition & Language Research Centre University College London 49 Gordon Square London WC1H 0PD England CORRESPONDING AUTHOR Kearsy Cormier [email protected] Phone: +44 (0)20 7679 8674 Fax: +44 (0)20 7679 8691 Email addresses for co-authors: [email protected], [email protected], [email protected] 1 Present address: National Institute for Deaf Studies and Sign Language, La Trobe University, Melbourne Victoria 3086, Australia 2 Present address: Department of Psychology, University of Crete, Rethymno, 74100, Greece GRAMMATICALITY JUDGEMENT IN BSL 2 Abstract Age of acquisition (AoA) effects have been used to support the notion of a critical period for first language acquisition. In this study, we examine AoA effects in deaf British Sign Language (BSL) users via a grammaticality judgment task. When English reading performance and nonverbal IQ are factored out, results show that accuracy of grammaticality judgement decreases as AoA increases, until around age 8, thus showing the unique effect of AoA on grammatical judgement in early learners. No such effects were found in those who acquired BSL after age 8. These late learners appear to have first language proficiency in English instead, which may have been used to scaffold learning of BSL as a second language later in life. Keywords: deaf; British Sign Language; critical period; acquisition; grammaticality judgement; syntax GRAMMATICALITY JUDGEMENT IN BSL 3 First language acquisition differs from second language acquisition in prelingually deaf signers: Evidence from sensitivity to grammaticality judgement in British Sign Language 1. Introduction According to a hypothesis originally proposed by Lenneberg (1967), there is a critical (or sensitive) period for acquisition of a first language (L1) – i.e., the first accessible language to which an individual is exposed, typically from birth – linked to neural plasticity which decreases as an individual grows older. This hypothesis has also been extended to apply to subsequent, second language (L2) acquisition. Based on a variety of studies which have examined L2 acquisition (e.g., Birdsong, 1999b), it does indeed seem clear that earlier exposure to a second language leads to more native-like competence of that language, although whether an actual critical period exists for L2 acquisition due to changes in the brain is unclear. Some have argued for a critical period for L2 acquisition, based on studies showing that native-like proficiency declines when the age of acquisition is after puberty (e.g., Coppieters, 1987; Eubank & Gregg, 1999; Johnson & Newport, 1989; Weber-Fox & Neville, 1999). However, others have argued against a critical period based on studies showing that individuals acquiring a second language may achieve near-native levels of proficiency at any age (e.g., Birdsong, 1992; Bongaerts, 1999; Hakuta, Bialystok, & Wiley, 2003; White & Genesee, 1996).3 Unlike the L2 critical period hypothesis, the L1 critical period hypothesis is untestable in the general population, because all typically-developing, hearing children have access to a first language from the surrounding language community. Evidence to support the L1 critical period hypothesis comes primarily from two groups with atypical language development. 3 For useful overviews covering key literature on the critical period hypothesis in second language acquisition, see Birdsong (1999a), Hyltenstam and Abrahamsson (2005) and Harley and Wang (1997). GRAMMATICALITY JUDGEMENT IN BSL 4 One group is children who have been deprived of language via isolation from humans early in life. One well-known case is Genie, a child who was isolated for the first 13 years of her life. After she was discovered and exposed to English at age 13, Genie exhibited better lexical than syntactic development and better comprehension than production. Her rate of language acquisition was overall very slow as well, and she had long-lasting problems with particular grammatical structures. However, it is difficult to draw strong conclusions about language development from cases such as Genie, not only due to their extreme rarity, but also because their isolation has broad consequences on all aspects of cognitive and social development, not only their linguistic development; thus any effects of age of language acquisition are confounded with these other variables (Curtiss, 1977; Skuse, 1988). The second group which provides evidence for a critical period for L1 acquisition are profoundly deaf children born into hearing families. In North America, for example, language acquisition and transmission for deaf individuals differs substantially from that of hearing individuals, with a maximum of only 5-10% of deaf children acquiring a sign language natively from deaf or hearing signing parents (Mitchell & Karchmer, 2004). This minority of deaf children who are exposed to a sign language from birth acquire that sign language in much the same way that hearing children acquire their native spoken language and with similar maturational milestones (Newport & Meier, 1985; Petitto & Marantette, 1991). The remaining 90-95% of deaf children are born to hearing families who typically do not use a sign language, instead focusing on teaching the child spoken language. Such an approach has had typically highly variable outcomes. Spoken language acquisition, if it does occur, is significantly delayed in many deaf children compared to hearing children, depending not only on the degree and kind of hearing loss but also the child’s home and school environments, their intelligence and time spent reading (Blamey, 2003; Sarant, Holt, Dowell, Rickards, & Blamey, 2009). In terms of the school environment, there have traditionally been three main GRAMMATICALITY JUDGEMENT IN BSL 5 types of communication that schools use in the education of deaf students: auditory-oral methods (where the focus is on speech, listening and/or speechreading), bimodal methods (where artificial sign systems are used to represent spoken language, e.g., Signed English or Sign Supported English in the UK), and natural sign languages such as British Sign Language (BSL). During the twentieth century, oral and bimodal methods were most prevalent worldwide until the 1980s which saw the introduction of bilingual-bicultural approaches to deaf education in some countries, due to changes in awareness of and attitudes towards the language and culture of Deaf communities and the recognition that deaf children are equally as capable of learning as their hearing peers (Swanwick & Gregory, 2007). The basic premise of bilingual-bicultural education was that full access to education via a natural sign language would facilitate age-appropriate language and cognition; this would then provide the basis for a transition to text-based literacy in the majority spoken language (Cummins, 1989; Johnson, Liddell, & Erting, 1989; Lane, Hoffmeister, & Bahan, 1996). Bilingual-cultural models of deaf education are still popular today. However, a lack of empirical evidence to support the effectiveness of bilingual approaches, along with technological advances such as newborn hearing screening and cochlear implants, both pose possible challenges to the continuation of bilingual programmes for deaf children in the UK and worldwide (Mayer & Leigh, 2010; Powers, 2002). Whether or not spoken language proficiency is successfully attained, some degree of proficiency in the surrounding spoken language via reading and writing can be achieved by some deaf individuals. However, such successes with literacy are again highly variable and not common. The average reading age for the adult deaf population in the UK and the USA is generally believed to be around 8 or 9 years, based on data from Conrad (1979) and Traxler (2000). GRAMMATICALITY JUDGEMENT IN BSL 6 Regardless of their success or failure to acquire spoken or written language, many deaf individuals from non-signing families may begin to use a sign language. This may occur only later in childhood when they encounter signing in school, or even much later after having left school. Some of these individuals can be considered to have delayed acquisition of a sign language as L1 (i.e., those for whom acquisition of the surrounding spoken language as an L1 has clearly failed). However, many deaf individuals have some degree of proficiency in the surrounding spoken language, so determining true cases of delayed L1 acquisition of a sign language, as opposed to second language (L2) acquisition of a sign language after successful acquisition of the surrounding spoken/written language as L1, can be a challenge. Distinguishing these cases of late L1 versus L2 sign language acquisition is crucial for making claims about the L1 critical period hypothesis, since it is widely believed that first language acquisition differs from second language(s) in that the former typically leads to native proficiency. Although “the earlier, the better” applies in both cases, early acquisition is considered relatively less important for ultimate attainment of L2 skills compared to L1. Also, some research suggests that second language acquisition is not a good test for the notion of a critical period for language in any case, because the outcome of second language acquisition may be influenced by circumstances of first language acquisition (Werker & Lalonde, 1988). Therefore, first language acquisition is a better test of the notion of critical periods overall. Acquisition of a first language from birth will ultimately lead to native proficiency, whereas delayed first language acquisition is unlikely to lead to complete acquisition at all, much less native/near-native proficiency. However, as noted above, cases of delayed first language acquisition of a spoken language are rare and confounded with other variables. Deaf individuals acquiring a signed language at varying ages provide a better testing ground for hypotheses about L1 critical periods. GRAMMATICALITY JUDGEMENT IN BSL 7 The current study examines the effects of age of acquisition on sensitivity to grammatical judgement in BSL. BSL is a natural sign language used by the deaf community in the United Kingdom (Sutton-Spence & Woll, 1999). Many aspects of its linguistic structure at the levels of phonology, morphology and syntax are different from English. BSL is also unrelated to, and for the most part mutually unintelligible with, most other sign languages, including American Sign Language (ASL) and Irish Sign Language (ISL). Many previous studies have shown age of acquisition effects within sign languages such as ASL and BSL, at phonological, morphological, and syntactic levels. Mayberry and colleagues (Mayberry & Eichen, 1991; Mayberry & Fischer, 1989) found that native signers outperformed non-native signers of ASL in sentence recall and sentence shadowing tasks, while Emmorey and colleagues (Emmorey, 1991; Emmorey, Bellugi, Friederici, & Horn, 1995; Emmorey, Corina, & Bellugi, 1995) found age of acquisition effects in ASL morphological repetition priming, lexical decision, sign monitoring, and probe recognition tasks. Differences have also been found between native and non-native signers in brain activation. MacSweeney, Waters, Brammer, Woll and Goswami (2008) reported greater activation in the left inferior frontal gyrus in deaf non-native signers than native signers during a BSL phonological judgement task, while Newman, Bavalier, Corina, Jezzard and Neville (2002) found more right hemisphere angular gyrus activation in hearing, native signers of ASL (i.e., ASL-English bilinguals) than in hearing late learners of ASL. These data have been used in support of the L1 critical period hypothesis. However, some of these studies did not distinguish between first language and second language acquisition, because they did not control for the extent to which the non-native signers were proficient in a spoken language before acquisition of a sign language, an important issue when considering whether their sign language was an L1 or an L2. This is a crucial issue when considering whether the results are relevant to critical periods for L1 acquisition. GRAMMATICALITY JUDGEMENT IN BSL 8 Some other studies have recognised the need to distinguish between L1 and L2 acquisition by attempting to ensure that their non-native signer participants had a sign language, rather than a spoken/written language, as L1. For example, Newport (1990) reported that native signers outscored non-native signers in a variety of production and comprehension tests of ASL morphology and syntax.4 Morford, Grieve-Smith, MacFarlane, Staley and Waters (2008) found age of acquisition effects in perception of handshape and place of articulation in lexical signs of ASL. Specifically, deaf native signers (with exposure to ASL from birth) showed higher discrimination with handshapes that were on the periphery of the category prototype or which straddled the category boundary, compared to deaf late learners of ASL (age of acquisition ranging from 10 to 18 years old). Mayberry (1993) directly compared L1 acquisition with L2 sign language acquisition effects, by comparing sentence recall in deaf ASL signers who reported acquiring ASL as a first language at various ages – from birth, early (between ages 5 and 8), and late (between ages 9 and 13) – with postlingually deaf individuals who had acquired English as L1 and acquired ASL as L2 after becoming deaf later in childhood. In this study, Mayberry found age of learning L1 effects: native signers performed better than early learners who performed better than late learners. Those who had acquired ASL as an L2 performed more similarly to the early L1 learners than the late L1 signers. From this, Mayberry concluded that L1 acquisition of a sign language (in prelingually deaf individuals) differs from L2 acquisition of a sign language (in postlingually deaf individuals). Mayberry, Lock and colleagues (2003; 2002) extended these 4 With regard to syntax, Newport (1990) found no differences between native and non-native signers in the comprehension of basic word order. This was measured by a task where signers watched simple ASL sentences with three different word orders (SVO; OSV; VOS) combined with different non-manual markers to signal changes in information structure. Participants were asked to choose one of two pictures in which the subject and object were reversed. Newport (1990) used the word order task from the draft Test Battery for ASL Morphology and Syntax (Supalla et al., n.d.) which included practice items that demonstrated correct responses (Schembri et al., 2002). It is possible that the non-native signers could have learned the rule from these practice examples. GRAMMATICALITY JUDGEMENT IN BSL 9 findings by examining L2 English proficiency in deaf L1 signers of ASL (with varying ages of acquisition) compared with L2 English proficiency in native speakers of spoken French, German, Italian and Greek. Their findings indicated that deaf native/near-native ASL signers (who had acquired English between ages 3 and 7 as L2) and hearing non-signers (who had acquired English between ages 6 and 13 as L2) performed better on English tests than deaf ASL signers who had acquired English between ages 6 and 13 as essentially a delayed first language.5 Finally, Boudreault and Mayberry (2006) examined grammatical judgement in signers who had acquired ASL as a first language from birth, early childhood, or later in life, finding that as delay in age of exposure increased, performance in grammatical judgement decreased, similar to Mayberry (1993). Taken together, these studies on deaf populations strongly suggest that age of first language acquisition of a sign language affects ultimate proficiency in that language. However, in all of these studies, the status of ASL or BSL as the first language of the non- native signers was based on their inability to use spoken English (Boudreault & Mayberry, 2006; Mayberry, 1993; Mayberry & Lock, 2003) or their “limited knowledge of English” (Morford et al., 2008: 43) as determined by self-report, or their “limited skills in English” with no indication of how this was determined (Newport, 1990: 14). The problem with relying solely on self-report for determining L1 in deaf signers is that deaf individuals are typically bilingual to some degree (even if only a small degree) in the surrounding spoken and/or written language (Ann, 2001; Grosjean, 1992), and there is great heterogeneity in age of both L1 and L2 acquisition, as discussed above. Determining the degree of competence in spoken/written language via only self-report can be difficult, particularly in late learners who 5 Mayberry & Lock (2003) claimed that the group of deaf people with a delayed first language in their study were not exposed to any accessible language at all until age 6, at which time they were enrolled in schools which used Total Communication, a bimodal strategy where signs (some from ASL, some invented for representing English grammatical morphemes) are produced in English word order. At these schools they were also taught English via lipreading, reading and writing. GRAMMATICALITY JUDGEMENT IN BSL 10 would have had more time to potentially develop proficiency in the spoken/written language. In most of these studies (MacSweeney et al., 2008, being one exception), measures which could have more convincingly eliminated the possibility of English as L1, such as reading ability (either via formal assessment or self-report), were not reported. In the current study, we examine age of L1 acquisition effects in a sign language by employing a grammaticality judgment task based on the ASL task designed by Boudreault and Mayberry (2006). Crucially, we take into account English reading performance and nonverbal IQ, in addition to self-reported demographic information from participants, in order to more directly assess the critical period hypothesis. Because our study is based closely on materials developed in ASL by Boudreault and Mayberry (2006), we first outline their methods and results. 2. ASL Grammaticality Judgement Task (Boudreault & Mayberry, 2006) Boudreault and Mayberry (2006) designed an ASL grammaticality judgement task to test whether age of acquisition had differential effects with regard to off-line grammaticality judgement of different syntactic structures. They selected various morphosyntactic structures based on their reported developmental timecourse in ASL acquisition. This included ASL structures which have been claimed to be acquired relatively early (by age 2;6-3;0), such as basic word order with plain verbs (Pichler, 2002) and negation and ‘agreement’ verbs6 (Anderson & Reilly, 1997; Meier, 1987). Boudreault and Mayberry also included structures reportedly acquired later such as wh-questions, relative clauses, and ‘classifier’ constructions.7 Although elements of wh-questions and classifier constructions first occur at ages similar to simple structures (Anderson & Reilly, 2002), adult-like use of these structures 6 Note that we use the term ‘agreement verb’ here for ease of comparison with the literature, but we accept the analysis proposed by Liddell (2000) that these forms do not actually mark for person agreement. 7 For a review of ‘classifier’ constructions in signed languages, and issues surrounding the terminology to refer to them, see Schembri (2003).