Course materials for/by Peter L. Patrick. Contains copyright material used for educational purposes. Please respect copyright.

Some Basic things that Variationists do or assume

by Peter L. Patrick

University of Essex

Language is heterogeneous, both in its structure and its use.

This variation is not generally random or free, but patterned/highly regular.

There are both internal (linguistic) & external (social/extra-linguistic) causes for this.

These patterns cannot be understood without looking outside langue (the abstract, autonomous system of language) to parole (speech behavior) & to the social world at large-- systematically examining the vernacular speech of communities.

Variation analysis must then be at least socially realistic (add social context & categories to normal linguistic practice), if not socially constituted (driven by social functions and placing language in a broader communicative context of social action-- Hymes 1973), since language is inescapably a social phenomenon.

The identification, definition & description of linguistic elements (variables) is central; also,

...the environment they occur in must be tightly constrained (envelope of variation).

The consideration of linguistic context is thus always a primary means of explaining variation,

...but linguistic elements may be extracted from context and compared for analytical purposes.

Linguistic variables are relatively autonomous instances within a closed set of choices.

Linguistic levels are permeable: elements at one level constrain/affect variation at another.

Such effects are not to be ruled out a priori due to lack of an explanatory theory.

Variation Analysis is to be empirical, explicit, replicable, and objective.

Descriptions are to be (initially) rich and full rather than spare and idealized.

For contemporary language, high-quality recorded natural data should be the norm;

...elicited, introspective, experimental, and attitudinal data are secondary;

...and these different data-types require different types of analysis/interpretation.

Only vernacular speech reveals full regularity/structure underlying surface variation.

Data from a range of styles iare needed to establish limits & norms of speech behavior; while from a range of social positions (speakers) are needed to map social distribution.

Ideally, investigate several data-sets, which should have complementary weaknesses.

(This last principle derives from, and applies also to, data-collection methodology.)


A typical sequence for variationist analysis:

1) Establish which forms alternate with one another - i.e., which are "the same".

2) Delimit the environments in which this alternation-with-sameness occurs, and classify the factors within those environments exhaustively.

3) Propose hypotheses for contextual factors which might constrain the variation.

4) Compile a data set that allows for investigation and (dis-)confirmation of the alternations and co-occurrences predicted by hypotheses in (3).

5) Compare the frequencies/probabilities with which the different variants co-occur with the different (environmental) factors.

6) Typically, place primary emphasis on internal linguistic factors, and only secondary importance on external social explanations.

7) Typically, consider analysis primarily exploratory rather than confirmatory (due to lack of precisely predictive sociolinguistic theories).

The text above consists primarily of independent observations of my own, though informed by the literature in language variation and change (esp. the work of Gillian Sankoff and William Labov). Please feel free to quote and use this material for educational purposes but respecting authors' intellectual rights and copyright laws.

LG554 Sociolinguistic Methods page

Peter L. Patrick's home page

Last updated 05 December 2007