Capítulo 4. Desarrollo de la aplicación
4.2. Estructura de la aplicación
In addition to acquiring the set of categorical grammatical rules of a language, such as that articles go before nouns in English, learners whose goal is to sound as native-like as possible must also become competent with a plethora of variable linguistic features. In contrast to the previous section (‘type 1’ variation), here variable production is part of
23 Note that these are not exactly the same as the factor levels presented in the previous table. The motivations for
the ‘target’ of acquisition as they are grammatical features for which native speakers alternate between two or more forms, in certain linguistic or social contexts. Therefore, studies ‘type 2’ or ‘horizontal’ variation compare interlanguage and native speaker (i.e. ‘target’) data, however unlike with the studies discussed in the previous section, the native speaker data is also variable. This typically requires sampling the native-speaker as well as the learner population, since variable language features don’t tend to feature in standard descriptive grammars.
As Clark & Schleef (2010: 299) set out (as do other papers in the Polish studies cited in the following paragraph), this type of acquisition is quite complex, requiring attainment of the following for a given variable:
• similar frequencies of variation as found in the target language community
• similar social and linguistic constraints on variation as found in the target language
community, and
• similar social judgements on variation as found in the target language community.
To explore how this kind of type 2 variation has been approached in acquisition, I turn to a set of studies, by Meyerhoff, Schleef and Clark, (Meyerhoff & Schleef 2012; Schleef, Meyerhoff & Clark 2011; Clark & Schleef 2010) investigating the acquisition of
sociolinguistic competence by teenage Polish immigrants in the United Kingdom. Again, it is the general method, rather than the specifics of the studies that are relevant. They investigate pronunciation of the ING variable, as it alternates in unstressed positions between velar, i.e. walking /ɪŋ/, and alveolar, i.e. walkin’ /ɪn/. This involves comparing usage rates and underlying constraints in the speech of Polish-born
teenagers now living in London (N=21) and Edinburgh (N=16) with their locally-born peers (Edinburgh N=21; London N=24). They test a number of different linguistic and social factors, motivated from the set of literature exploring variation of this form in different native-speaker and learner contexts.
After Mougeon, Rehner & Nadasdi (2004), they propose four logical possibilities in terms of how the speech of Polish immigrants might relate to that of the locally-born teenagers (Schleef et al 2011: 207):
1. migrant adolescents could adopt the same distribution of variants as their locally-born peers;
2. migrant adolescents could show variation that reflects the same underlying constraints operating on the variation of their locally-born peers, but the strength of these constraints may differ or the strength of individual factors within those constraints may differ;
3. migrant adolescents could reinterpret the variation producing patterns of variation radically divergent from their locally-born peers; and
4. migrant adolescents could eliminate the variation and show categorical use of one variant or another.
Each group of learners resembled their respective native speaker community in terms of the baseline frequency of use of the variable: Edinburgh Poles, like Edinburgh natives, used the apical variant to a greater extent that London Poles. However, when the pattern of variation is addressed at a deeper level, the results show that in fact a mix of the first three outcomes is the case for both the London and Edinburgh data: both contained a only single shared constraint (between native-speakers and Polish immigrants) (hypothesis 1), plus several reinterpreted constraints (i.e. with different hierarchies within factors) (hypothesis 2) and several factors which were not shared at all
(hypothesis 3). This mix of factors suggests a complexity of processes operating on the acquisition and use of variable grammar.
The authors observe that this concurs with other research showing that teenage and adult L2 learners behave quite differently from child L1 learners who tend to “replicate variable input with variable output” (p226) from very early on in acquisition. They observe that L2 learners “in a situation of language contact do not have access to the same depth and breadth of information about the nature of the variable that an L1 learner does” (p227). They also propose that it could be related to differing capacity to extract variable rules from the input, such that children have a greater capacity to perform this task. I will return to discussion of these issues below (§3.3.2).
Moreover, while the learners did not particularly resemble native speakers, they also did not particularly resemble one another (i.e. London-based and Edinburgh-based Poles),
suggesting that there is not a universal strategy used in acquisition of this variable. A similar lack of generalizability, though across linguistic features rather than groups of learners is demonstrated in the work of Mougeon, Rehner and Nadasdi (2004), who looked at a number of sociolinguistic variables in a French-immersion context in Canada, and found results differed (from hypothesis 1 to 4 above) depending on the variable in question.
In a similar vein, but with a focus on the acquisition of L1 variable rules, Smith, Durham and Fortune (2007) compare the variable use of phonological and morphological
variables by young Scottish children and their caregivers. The aim of this study is to elucidate how variable rules in the input are replicated by children. In order to do so, the authors perform the same suite of analyses on both the adult and child data, and finally compare the results from both.
These studies are clearly methodological descendants of the studies of type 1 variation discussed in the previous section (§3.2.1.1). In order to make some comparison about the variable grammar, the authors cross-check several components including the overall frequency of variants, the set of constraints on variation (i.e. factor groups) and the extent to which these are shared, and the hierarchy (of factors) within shared constraints and the extent to which these are shared. I will now discuss the application of this
methodology to a rather different type of data set.