L1 attrition in a wider perspective

In proposing a formal model for grammatical attrition, Hicks and Domínguez (2020, henceforth H&D) present some valid arguments but also expose some limitations of their approach. First, they state that ‘the possibility of grammatical attrition is crucial to our understanding of bilingual acquisition (p. 144).’ I am in complete agreement with this view. For a long time, research on first language (L1) attrition was not systematically connected to research on second language (L2) acquisition. The emergent view now is rather to look at L1 attrition and L2 acquisition as two inseparable sides of bilingualism, both in the same individual speaker and in bilingual communities. Second, the new picture requires considering L1 attrition not only in terms of loss or erosion but, more generally, as change (although not necessarily grammatical change): this is a view that has been gaining strength in recent years, and is also explicitly recognized by H&D. Third, H&D acknowledge that incorporating L1 attrition in a generative model requires ‘a model of the language faculty that retains a role for input in maintaining an L1 grammar beyond the primary years of language acquisition’ (p. 145): they are absolutely right, although they ignore the equally strong need to understand how language interacts with general cognition and how these interactions modulate both L1 attrition and L2 acquisition. Ultimately, H&D’s main concern seems to be to provide a formal generative account of L1 attrition, to the exclusion of additional (not necessarily alternative) explanations. I will briefly expand on each of these points.

I would start by distinguishing between L1 attrition in first-generation L1 speakers who learn another language as adults and generational attrition in heritage speakers who learn their minority language as children in a bilingual context. H&D are only concerned with individual attrition, not with generational attrition, but the distinction is important because the effects on language in first vs. second generation attrition are different, both in terms of scope and in terms of stability. As H&D recognize, individual attrition involves no ‘erosion’ or ‘permanent loss’ but rather fluctuations and increasing optionality: this is because attrition in this sense crucially does not affect the grammar itself but rather how the grammar is accessed (Sorace, 2011, 2016). Parental input affected by attrition is then transmitted to the next generation of heritage speakers, who regularize variable input as part of their grammar. The degree of parental attrition has generally not been studied as a variable in heritage language development, but it is crucial for an understanding of the diachronic dimension of language change.

Therefore, the answer to H&D’s question of ‘where grammatical attrition arises, what kind of mental representations are affected?’ (p.145) is that in fact mental representations of grammatical knowledge in mature L1 speakers are not affected. Supporting evidence comes from studies of re-immersion in the L1 community (Chamorro, Sturt and Sorace, 2016), which shows that attrition effects are not permanent but at least partly reversible. Sensitivity to re-exposure/re-immersion is incompatible with permanent changes: one more reason to think that the mental grammar itself is not affected by another language that becomes dominant in the environment.

H&D also ask: ‘what accounts for the strikingly low levels of attrition in L1 grammar?’ (p. 146) A potential answer to this question comes from evidence that attrition is both selective and partly independent of the specific languages involved. Research on pronominal reference, which H&D quote extensively, shows that grammatical properties interfacing with non-linguistic external conditions are particularly sensitive to attrition, both when the L2 is typologically similar to the L1 and when it is typologically different (see, for example, Bini, 1993; Bonfieni, 2018; Lozano, 2006). One plausible hypothesis is that what is affected by attrition, at least in adult late bilinguals, are the non-linguistic processes involved in connecting pronominal choices with ever-changing discourse, pragmatic, and contextual conditions. The frequency of implementing the connections between choices of grammatical forms and contextual conditions is a factor modulating attrition, which explains the sensitivity of these effects to re-immersion in the L1 community. H&D prefer to locate ‘interpretive dependencies’ within the grammar itself and treat attrition as involving (re)assigning L2 feature bundles to L1 lexical items; however, it is unclear not only how this treatment can accommodate the existing evidence but also what it can gain in terms of parsimony: why should we try to account for everything in formal grammatical terms when there are plausible non-grammatical explanations?

It is interesting that H&D revive the input–intake distinction that used to be common in early work on L2 acquisition (e.g. Corder, 1982). ‘Intake’ is the proportion of the input that can be processed by the learner for acquisition. One of the factors modulating how much ‘input’ becomes ‘intake’ is the extent to which L2 learners rely on their L1. There is individual variation among L2 learners in the degree to which they are able to inhibit their L1 when learning and then using the L2. Successful inhibition of the L1 may play a role both in making it more susceptible to attrition and in developing L2 knowledge; conversely, less successful inhibition of the L1 leads to L2 development that continues to be filtered through the L1. As H&D put it, ‘changes in the L1 grammar require successful [L2] intake’ (p. 159): in other words, successful L2 acquisition builds on the possibility that the L1 also (selectively) changes, which may be influenced by individual differences in cognitive profiles among learners. This is an alternative way of accounting for the ‘attrition via acquisition’ relationship that H&D propose: the relationship certainly exists but it is based on complex interactions of language and general cognition that go beyond a purely grammatical account. Extralinguistic factors are mentioned as part of the ‘inference component’ in H&D’s model, whose job in attrition is to update the L1 grammar: essentially, to modify it in the light of dominant L2 input, ‘supplanting’ L1 feature assemblies with the corresponding L2 ones. However, it is unclear how exactly these factors work. Furthermore, this kind of permanent change due to specific L1–L2 correspondences is not what we see in L1 attrition of mature grammars.

Finally, it would be hard to dispute the need for ‘a model of the language faculty that retains a role for input in maintaining an L1 grammar beyond the primary years of language acquisition’ (p. 145). Generative linguistic models are traditionally based on monolingual acquisition and implicitly assume that acquiring a native language leads to an unmodifiable steady state. But now that bilingualism is becoming more common even in the so-called Anglosphere, we need a formal model of language that recognizes that bilinguals are not the sum of two monolinguals, as Grosjean (1989) famously put it. Is it true that ‘there is currently no model of grammatical attrition compatible with formal generative models’ (p. 144)? It depends on whether the ideal model has to encompass different aspects of acquisition and attrition, such as knowledge representations and real-time processing. The Interface Hypothesis is cited as a generative model that fails to account for attrition in grammatical representations. However, H&D seem to consider an old version of the interface model, based on a binary split between ‘narrow syntax’ and ‘interfaces’. In the light of current research (see Chamorro and Sorace, 2019; Sorace, 2011, 2016), it seems more appropriate to assume a continuum of conditions on syntactic realization, ranging from more ‘internal’ to more ‘external’ and involving different types of cognitive processes: there is no syntax that is unaffected by conditions of any kind. At this stage, there is still much that we do not know about the relative sensitivity of different conditions to changes in the environment. What we do know is that grammatical structures subject to more ‘external’ conditions tend to be more susceptible to the effects of reduced exposure to the L1 and increased exposure to another language, at least in adults (on the difference between anaphora resolution and DOM in L1 Spanish attrition, see, for example, Chamorro, Sorace and Sturt, 2016; Chamorro, Sturt and Sorace, 2016). Conversely, grammatical structures conditioned by more ‘internal’ conditions are resistant to attrition, at least in adult speakers (much less in children; see Flores, 2010).

In sum, we are working on a better understanding of the relationship between L1 attrition and adult L2 acquisition. H&D raise some interesting questions but we need a wider interdisciplinary framework to address them effectively.