Объединенный доклад Михаила Даниэля и Полины Казаковой

“Who’s older than the old?” и “Are dialect features lost in a stable order?”

Both talks investigate dialect loss among speakers of the NorthRussian dialect spoken in Mikhalevskaya, a village in the south of Arkhangelskaja Oblast’ as represented in the Ustja River Basin Corpus(URB, Daniel et al. 2013-2016), a corpus of spontaneous speech from 62 speakers of different ages with more than 0.5 mln tokens.

I. We will discuss the problem of quantitative analysis of age groupoutliers, i.e. consultants whose linguistic behavior is significantly more or significantly less dialectal than that of their age peers in the condition of dialect loss. Isolating outliers may be useful forvarious purposes (see, for instance, Britain 2003), such as modellingage groups. Whereas the discussion on consistency in being innovative or conservative in sociolinguistics is mostly connected to the studyof gender (e.g., Maclagan et al. 1999), in this talk, we want to address the following issue: are outliers consistently ahead or below their age peers by all variables, or can a speaker be highly innovative in some variables while being highly conservative inothers?
An important issue we encounter in the analysis of the data consists in individual speakers showing statistical tendencies of use rather than categorical choices in respect to variables, and different degrees of representativeness of the data which is due to small numbers of observations per individual speaker. This makes the comparison of the speakers a difficult task. With these methodological issues in mind, we use an algorithm that sets up an ‘optimal’ order of speakers. This order is initially based on age and subsequently changed, so that it reflects not only the age of speakers but also their dialect preservation. The default assumption is that a younger speaker is less dialectal; whenever the opposite is statistically significant in our data (calculated as exact Fisher test based on occurrences of dialectal vs. standard realization of a variable compared in two speakers), we reverse the order. As a result, we obtain an ‘ideal’ order for each variable where for each pair of speakers it is true that the one who stands to the left of the othe rmust be either older or more conservative than the other. By comparingthis dialect age orders in different variables, we see whether theseorders are consistent. Preliminary results suggest that speakers are NOT consistently conservative or progressive. In the talk, we will discuss this method and similar approaches and present results based on the data in the URB.

II. As a dialect speaking community moves toward the standard, it gradually loses dialect features. In this talk, we ask whether
this happens along a fixed route, that is, whether certain features are consistently lost before others on the level of individual speakers (cf. Trudgill (1986: 20) on speakers of British English accommodating to American English who “will almost certainly accommodate phonologically by acquiring features in a certain order”, see also the discussion in Rickford (2003)). Such an order would shed light on relevant linguistic and sociolinguistic factors in the dialectattrition process. We consider the dynamics of dialect loss based on the data in the URB corpus.
The speech of the villagers in the corpus is highly heterogeneous interms of its assimilation to standard Russian. While the oldest speakers (born in 1920 to 1940) show a high degree of dialect preservation, those born in 1960 to 1996 have lost the dialect almost completely. We investigate a number of binary phonological and morphological variables that are well represented in the data and lend themselves to a clear interpretation of dialect loss. These include[e] for etymological [a] between palatalized consonants; dialectal realizations of the postfix -sja; dialectal realization of  the particle -to; absence of the initial n- in oblique forms of the third person pronouns; and others. The main research question is: are thevariables being lost in a specific order?
To answer this question, we analyze and compare dialectal vs. standard realizations of each pair of variables for each pair of speakers using standard statistical tests, and aggregate over the resulting matrix to establish a measure that indicates which, if any, features are lost before others, and to what extent this is a consistent phenomen on. The results support the hypothesis that there is a certain order in which(dialectal) features are acquired or lost in the community. While it thus seems that the fixed route hypothesis holds, note that individual variables follow different diachronic trends. For some variables, the curve of loss is steeper than for others, i.e., some are lost quickerthan others. The differences in variable preservation for each individual speaker may result simply from different rates of loss, making for a much less interesting phenomenon than if systemicrelations between the variables are the root cause of an orderedtransition to a new variety. In the talk, we will discuss the validityof this results and possible ways to expand on it.

References:
Britain, D. (2003). Exploring the importance of the outlier insociolinguistic dialectology. In: Social Dialectology. Amsterdam: JohnBenjamins.
Daniel, M., N. Dobrushina and R. von Waldenfels (2013-2016). Thelanguage of the Ustja river basin. A corpus of North Russian dialectalspeech. Bern, Moscow. Available at: www.parasolcorpus.org/Pushkino.
Maclagan, M.A., E. Gordon and G. Lewis (1999). Women and sound change:conservative and innovative behavior by the same speakers. LanguageVariation and Change, 11, 19-41.
Rickford, J. R. (2003). Implicational Scales. In: The Handbook ofLanguage Variation and Change. Oxford: Blackwell.
Trudgill, P. (1986). Dialects in contact. Oxford: Blackwell.