Age effects in the acquisition of English article semantics
Tania Ionin, University of Illinois, Urbana-Champaign
Maria Luisa Zubizarreta, University of Southern California
Vadim Philippov, Orel State University

Investigations into age effects on L2-acquisition have traditionally looked at end-state learners, with focus on phonological or syntactic domains. More recently, focus has shifted to the processes behind child vs. adult L2 acquisition (Schwartz 2004), and age effects at the syntax-semantics interface have received more attention (Unsworth 2005, Song and Schwartz forthcoming). Our paper continues this investigation by examining the role of age in the domain of article semantics.

Our starting point is Ionin, Ko and Wexler's (2004) (IKW) proposal that adult L2-English learners whose L1 lacks articles are sensitive to UG-provided semantic universals. While English groups articles on the basis of definiteness, other languages, notably Samoan, group articles on the basis of specificity (1). IKW found that adult L2-English learners from article-less L1s fluctuate between these two possibilities, resulting in a non-random error pattern: errors occur only when definiteness and specificity are in conflict, in (4) and (5), but not (3) and (6).

However, recent work on Samoan (Fuli 2007) has shed doubt on IKW's proposal. Samoan makes a morphological specificity distinction for indefinites but not for definites: all definites are marked with the same article as specific indefinites (2). This is consistent with Lyons's (1999) claim that no language distinguishes specific and non-specific definites. In this new light, IKW's findings no longer provide evidence for UG-access in L2-acquisition: while the error in (5) is consistent with cross-linguistic data, the error in (4) is not. We argue that adult L2-learners use explicit strategies in their article choice, rather than relying exclusively on UGprovided implicit knowledge (cf. Trenkic 2008). We hypothesize that child L2-learners rely more on implicit knowledge, showing a pattern of article use consistent with natural language data.

We tested 18 L1-Russian children (ages 10-12) and 21 L1-Russian adults, both groups learning English in Russia, and age-matched controls. We used written elicitation with dialogues, like IKW. Examples of target sentences are in (3) through (6). L1-English adults and children used the in nearly all definite and a in nearly all indefinite contexts, and were not influenced by specificity, while L2- English learners were influenced by specificity, as expected. Both adult and child L2-learners overused the significantly more with specific than with non-specific indefinites (7). In contrast, only adults made a specificity distinction with definites (8). Thus, children but not adults exhibit an error pattern consistent with natural language data (2).

We conclude that (a) all L2-learners are sensitive to the semantic universal of specificity; (b) children make use of this implicit knowledge directly; and (c) adults, who have greater metalinguistic knowledge, overrely on explicit strategies which overextend the specificity distinction to definites, at least in an explicit task (cf. Ellis 2005). In a more implicit task (narrative production in IKW), adults made the specificity distinction with indefinites only, consistent with natural language data. In contrast, child L2-learners are guided by implicit knowledge regardless of task type, as shown by similar results obtained in controlled elicitation (our study) and naturalistic oral production (Zdorenko and Paradis 2008).

Please click here for diagrams and references