Aria Adli

Universität Tübingen

Limits of variation or the dream of homogeneous data

"What do I mean by syntactic variation?"

This first question constitutes one of the important reasons why generative syntacticians and variational sociolinguists sometimes talk at cross-purposes with respect to syntactic variation. It is helpful to distinguish between (i) variation between languages, (ii) sentence-internal, structural variation, and (iii) linguistic inter-group (or inter-individual) variation due to external (social and/or cognitive) factors. Although syntax and sociolinguistics are not supposed to be close neighbours, a more thorough look reveals that they share at least a common problem: sentence-internal, structural variation is a difficulty in variationist studies of the Labovian style as well as in the Minimalist Program. The issue is the same: How can I assume two syntactic forms to be optional variants of each other?

"Why should I care about syntactic variation due to external factors?"

The Labovian definition of a linguistic variable, conceived as a linguistic unit with two or more variants that covary with other linguistic and/or social variables, requires the combination of the meanings (ii) and (iii). In contrast to phonology, this approach turns out to be particularly difficult at the level of syntax.

The generative syntactician does not, to start with, have any need to discuss the level of external factors due to partly the hypothesis of the autonomy of syntax, and partly the concept of the ideal speaker-hearer. However, another apparently independent issue in grammar research reveals such a need: the issue of the relation between theory and data, that is the problem of the empirical reliability of grammaticality judgments. This problem becomes visible at least in moments when a syntactician has to give intuitions concerning marginal data (in fact, one can discover surprising data-mismatches in the literature).

Syntacticians can profit from the knowledge about external factors, e.g. in order to develop a careful and conscious methodology of survey and data evaluation, whether they use personal intuitions or collect their data in controlled surveys. Variational sociolinguists, on the other hand, need the insights of syntax theory in order to be able to deal with meaning (ii) of syntactic variation.

"Does syntactic variation due to external factors exist ?"

Finally I present empirical data which show statistical correlations between quantitative measures of grammaticality judgments and specific social factors. Different types of French interrogatives sentences have been investigated in an experimental setting.

We do find a general effect of an external factor, i.e. meaning (iii) of syntactic variation. On the one hand, the socio-cultural variable lifestyle correlating with the grammaticality judgments, as long as the sentences are not marginal or ungrammatical. On the other hand, there is a specific cognitive effect of age (related to education) limited to suboptimal data. This pattern of results reflects, in a certain sense, the acceptability status of the sentences.

A social effect on the preference of structural variants, i.e. an approach in the style of the Labovian linguistic variable, shows up with the judgments of two licensed forms of French wh-questions, one with wh-movement, one with wh-in-situ.

The results reveal on the one hand systematic, i.e. explainable, variance in the grammaticality judgments and at the same time limits of these effects, beyond which the judgments seem to be resistant to the influence of certain external factors.