The role of linguistic input in adult grammars: Modelling L1 morphosyntactic attrition

Abstract

Theoretically modelling L1 grammatical attrition – defined in this thesis as the modification of fully acquired L1 grammatical representations due to a significant and prolonged change to a speaker’s linguistic input – remains a considerable research challenge. This thesis seeks to contribute to the modelling of L1 grammatical (specifically morphosyntactic) attrition by investigating the role of linguistic input in this process. To this end, it tests a recent and hitherto untested Minimalist model of grammatical attrition in which input plays a key role: The Attrition via Acquisition Model (Hicks and Domínguez, 2020a, 2020b). In the model, the possibility of attrition is determined by the availability of intake (defined as processed input) and the potential for Feature Reassembly (FR) of a previously acquired L1 structure. Two broad predictions are derived from the model’s assumptions and architecture. Firstly, attrition is in principle only possible for an L1 structure which has an analogous/equivalent L2 form which nonetheless differs in its behaviour due to differences in the respective L1–L2 functional feature specifications. Secondly, attrition is facilitated in linguistic environments in which the L2 is holistically more similar (i.e., in terms of morphosyntax, semantics, phonology, phonetics, and lexis) to the L1. To further develop the predictive power of the model, this thesis also formulates and tests the novel hypothesis that attrition is further facilitated for L1 structures which would need to undergo less complex FR to match a corresponding L2 structure due to greater overlap in the relevant L1–L2 feature specifications (in other words, this predicts that structural similarity at the level of individual morphosyntactic structures modulates the likelihood of attrition). Three groups of L1 German late-sequential bilingual speakers of either Dutch, English or Spanish are tested. These participants (n=85) have at least 15 years of residence, starting in adulthood, in either the Netherlands, UK or Spain. As well as a native German control group (n=44), this study also uses three further control groups consisting of native speakers of Dutch, English or Spanish (n=94). Two grammatical structures, which are argued to differ in the relative complexity of FR required for them to attrite in the L1, are investigated per L1–L2 pairing. Grammatical properties are tested by means of bimodal Acceptability Judgement Tasks (AJTs). There is a German and equivalent L2 version of each task. Potential attriters completed both an L1 and L2 version of the relevant AJTs. As such this is the first study to compare attrition of the same L1 under the influence of three holistically different L2 inputs and to investigate the role of FR complexity in attrition. This study also adds to the very small number of attrition studies which test potential attriters in their L2 as well as their L1 in order to further investigate the relationship between L2 acquisition and L1 attrition. Though results from a series of Culminative Link mixed effects models reveal no significant group-level attrition, attrition is found for a number of individual participants on five out of the six properties tested. Comparison of these individual results within and across the three L1–L2 groups does not suggest that either L1–L2 holistically similarity nor L1–L2 structural similarity – at least as formulated in this thesis – modulate the likelihood of grammatical attrition. The L2 acquisition results of these participants are very varied and indicate a rather complex relationship between L2 acquisition and L1 grammatical attrition. In light of these findings, some modifications to the Attrition via Acquisition model are proposed.

More information

Identifiers

ORCID for Lewis Baker: orcid.org/0000-0001-8348-4677

ORCID for Glyn Hicks: orcid.org/0000-0003-4126-8655

ORCID for Laura Dominguez: orcid.org/0000-0003-2701-2469

Catalogue record

Export record