Investigating Obsolescence: Where are we now?

Anwyl-Irvine, A. L., Massonnié, J., Filtton, A., Kirkham, N. & J. K. Evershed. 2019. Gorilla in our midst: An online behavioural experiment builder. Behavior Research Methods, https://doi.org/10.3758/s13428-019-01237-x.

Bates, D., Maechler, M., Bolker, B., & Walker, S. 2016. Lme4: Linera mixed effects models using “Eigen” and S4. https ://cran.r-proje ct.org/web/packa ges/lme4/index.html (accessed 20 June 2023).

Dresher, B. E. 2000. Analogical levelling of vowel length in West Germanic. In: A. Lahiri (ed.), Analogy, Levelling, Markedness. Berlin: Mouton De Gruyter, pp. 47-70.

Favre, S. 2009) Qu’en est-il du subjonctif présent ? In: C. Fréchet (ed.), Langues et cultures de France et d’ailleurs. Lyon: Presses universitaires de Lyon, pp. 163-176.

Hill, J. 1989. The social functions of relativization in obsolescent and non-obsolescent languages. In: N. C. Dorian (ed.), Investigating Obsolescence: Studies in Language Contraction and Death. Cambridge: Cambridge University Press, pp. 149-164.

Lopez-Beltran, P. 2021. Heritage speakers’ online processing of the Spanish subjunctive: A comprehensive usage-base study. Unpublished PhD Dissertation. Penn State University.

Nützel, D. 2009. The East Franconian Dialect of Haysville Indiana: A Study in Language Death. Regensburg: Vulpes.

Tsitsipis, L. D. 1984. Functional restriction and grammatical reduction in Albanian language in Greece. Zeitschrift für Balkanologie 20: 122-131.

Sasse, H.-J. 1992. Language decay and contact-induced change: Similarities and differences. In: M. Brenzinger (ed.), Language Death: Factual and Theoretical Explorations with Special Reference to East Africa. Cambridge: Cambridge University Press, pp. 59–80.

Stich, D. 1998. Parlons francoprovençal: Une langue méconnue. Paris: L’Harmattan.

White, Y. & G. Roberts. 2022. Variability in speaker expectations of morphosyntactic mutation in Welsh. Glossa 7(1): 1-44.

Since the 1980s at least, it has been the receive view that linguistic variation is ‘atypical’ in language obsolescence, which is often characterised in terms of deficit, and where labels such as ‘defective’, ‘incomplete’, or ‘random’ are commonplace (e.g. Sasse 1992, Dresher 2000). Subjunctive subordination is an oft-cited example where this variation is reported, and where processes such as a laxing of categorical rules, paradigm levelling, and complete loss are described (e.g. Tsitsipis 1984, Hill 1989, Nützel 2009). I argue that these observations, though frequent, result from an anecdotal evidence base and out-dated methods.

This pilot study reports on the production and perception of the subjunctive mood among 36 speakers (balanced for age/gender) of obsolescent Francoprovençal in Savièse (Valais, Switzerland). Previous descriptive work (e.g. Stich 1998, Favre 2009) suggests that the subjunctive is undergoing a weakening of present and imperfect paradigms, leading to variation in the system, with the imperfect replacing the present (cf. 1-2, dataset). However, little empirical work exists on this under-documented and severely endangered language.

Production data were elicited using variationist sociolinguistic protocols, modified for the specificities of the language context, as all participants self-report to be illiterate in their language. Participants attended sociolinguistic interviews which took place in two parts. First, an oral-translation task was administered in order to capture more monitored speech. Second, participants attended group discussions, guided by a speaker and community insider, capturing more informal speech. Both parts yielded n=176 subjunctive tokens across styles, which were then subjected to generalized linear mixed-effects modelling in R using lmea4 (Bates et al. 2016). In follow-up meetings, perceived-acceptability data were elicited using a novel auditory judgement task built in Gorilla.sc (Anwyl-Irvine et al. 2019). A small sub-sample (n=9) of participants provided ratings on a 5-point scale for 21 randomised matrix+embedded-clause audio samples. Participants, who undertook the task in their homes on an iPad, did not demonstrate or report difficulty in undertaking the task.

Results from the judgement task indicate that participants do not discriminate between present- and imperfect-subjunctive forms in subordinate clauses where present-tense matrix clauses contain a subjunctive governor (Fig 1, dataset). However, limited evidence for this tense mismatching is observed in production, where just 5% of matrix clauses governed by a present-tense subjunctive trigger contained an imperfect form (Table 1, dataset). The subjunctive is however more likely to vary with the indicative (i.e. as is observed in ‘healthy’ languages), where grammatical category and speech style tend to significance (Table 2, dataset).

The findings from this severely-endangered-language context, rare in experimental sociolinguistic work, augur not only with heritage-language research, where subjunctive forms may be more accepted under controlled conditions than they are produced in elicitation tasks (e.g. Lopez-Beltran 2021), but also with experimental work where bilingual speakers in minority-language settings accept much more variability in complex structures (White & Roberts 2022). Taken together, evidence from production and perception point to predictable patterns of variation as observed in non-obsolescent settings.