Periodic Reporting for period 4 - STARFISH (Sociolinguistic typology and responsive features in syntactic history)
Reporting period: 2025-04-01 to 2025-09-30
The aim of the project was to find out whether adult second-language acquisition can lead to changes in the syntax of a language in historical time. In particular, we were interested in whether there are syntactic features that are particularly favoured or disfavoured in languages historically characterized by a high proportion of adult second-language acquirers independently of what their first language is, and - if so - what these "responsive" features are. Our aim to better understand language change and syntax falls under basic research, which is of inherent importance to society.
Our findings were mixed. In the case of null subjects, our results were compatible with the project's initial hypotheses: "responsive" features are semantically uninterpretable features in the sense of Minimalist syntax, and adult L2 acquisition may indeed lead to reductions in the frequency of null subjects. The same may be true of subjunctive mood morphology, insofar as it is mediated by uninterpretable features (that is, subjunctives that are syntactically rather than semantically specified), though more research is needed here. In the case of syntactic case marking, our empirical findings were compatible with the hypothesis that adult L2 acquisition may lead to reduction and loss of case marking (though our case study did not allow us to explore this in as much sociolinguistic depth as we wanted), but not with the hypothesis that this is driven by uninterpretable features. We also showed that there are plausible individual-level and population-level mechanisms whereby responsive features may change in frequency or be lost entirely depending on the makeup of the population in terms of L1 and L2 acquirers. Overall, we have shown that an approach that mixes historical corpus methods and mathematical modelling can shed light on questions of sociolinguistic typology.
Our findings were mixed. In the case of null subjects, our results were compatible with the project's initial hypotheses: "responsive" features are semantically uninterpretable features in the sense of Minimalist syntax, and adult L2 acquisition may indeed lead to reductions in the frequency of null subjects. The same may be true of subjunctive mood morphology, insofar as it is mediated by uninterpretable features (that is, subjunctives that are syntactically rather than semantically specified), though more research is needed here. In the case of syntactic case marking, our empirical findings were compatible with the hypothesis that adult L2 acquisition may lead to reduction and loss of case marking (though our case study did not allow us to explore this in as much sociolinguistic depth as we wanted), but not with the hypothesis that this is driven by uninterpretable features. We also showed that there are plausible individual-level and population-level mechanisms whereby responsive features may change in frequency or be lost entirely depending on the makeup of the population in terms of L1 and L2 acquirers. Overall, we have shown that an approach that mixes historical corpus methods and mathematical modelling can shed light on questions of sociolinguistic typology.
We have been carrying out original empirical research in four areas: null subjects in Latin American Spanish, case loss in Balkan Slavic, grammatical gender loss in Middle English, and change in the distribution of subjunctive mood in Ibero-Romance. For Latin American Spanish, diachronic trends are masked by the nature of the data, but a major finding is that orality plays a key role in regulating the distribution of null subjects. We found that case loss in Balkan Slavic proceeds broadly along the lines predicted by case hierarchies, with cases higher in the hierarchy (e.g. instrumental) lost earlier and faster than cases lower in the hierarchy. As regards grammatical gender in Middle English, quantifiers and adjectives are affected by the change differently, and there is a strong regional element to the distribution. On the modelling side of the project, we have established a threshold for proportion of L2 learners at the population level in order for responsive features to be lost, and we have engaged with the literature on quantitative modelling of L2 learning effects in linguistic typology. More broadly, we have established and articulated a historical-corpus-based approach for testing questions of sociolinguistic typology in syntactic change. An article in EU Research magazine aimed to disseminate our hypotheses and early findings to a broader audience. Our results have been published in a variety of journal articles, and a book-length publication providing an overview of the project's approach and main findings has been provisionally accepted for publication with Language Science Press.
The findings discussed above have not been remarked upon in earlier literature, so they go beyond the state of the art.