Professional Documents
Culture Documents
Chapter 15 - Variation
Chapter 15 - Variation
Variation
One problem in determining the relationship between form, function, and social
identity in language is that it is highly variable. The claim that language is
variable is relatively uncontroversial. Any short discourse will reveal variability.
For example, in 1) there are five words ending in -ing. Two are pronounced [ɪŋ],
two are pronounced [ən], and one is pronounced [in].
1)
Stocks and investments and stuff like that in math but no, they don’t do
nothen’. Like, I don’t do anything. We were talken’ about school and I don’t do
anything that I’ve learned from school. Notheen. (Alberta English Corpus; Cass
26, Spkr 023, 22.08)
Where fields of linguistics differ is in how this variation is to be
interpreted and even if it is to be interpreted at all. Most studies of the rules
governing phonology, morphology, and syntax attempt to factor out
unexplained variation. The variationist approach to language is unique in that it
takes as its starting point the tenet that language is inherently variable. As
noted in the section on discourse, the central focus of interest to variationist
analysis is the sociolinguistic variable. This is a set of forms or variants that can
all be used to say the same thing. A phoneme, a morpheme, or a set of syntactic
structures can each be considered a variable if at least some of its
manifestations cannot be explained using categorical, or invariant, rules. By
counting each possible variant of a variable and finding out what percentage
occur in each possible conditioning environment, patterns in the variation can
be discovered. For example, variationist analyses of the example in 1)
comparing the percentage of -in’ versus -ing in casual registers to that found in
formal registers in a number of communities have usually found that formality is
a factor in determining which variant is used.
Data culled from speaker intuitions is prohibited in variationist study of
the sociolinguistics of society because the researcher can never be sure what
speaker intuitions about variable patterns represent. Some people have very
strong attitudes about language. The “Letters to the Editor” column in the
magazine Organic Gardening provides an excellent example. When the
magazine hired a new editor who adopted a more casual style, a controversy
began that ran for well over three years with many readers voicing strong
objections, as shown in 2).
Contemporary Linguistics: An Introduction, 5th edition, Chapter 15: Variation, 2
2)
I suspect you view yourself as a breath of fresh air to this encrusted old
publication. I view your insistence on fracturing the written English language as
a blast of pollution. (January, 1992, p. 14)
Sociolinguistic studies have found that speakers who believe a particular
form to be “slang” or “incorrect” will often deny using it even though subsequent
examination of their recorded speech shows that they do. Thus, only corpora of
natural conversational speech are acceptable for examining variation.
Since the rarity of some variant may merely reflect the rarity of the structure
where it is most likely to occur, there is no infallible way of using conversational
data to determine when a form is completely prohibited. Instead, variation
studies compare overall rates of occurrence and co-occurrence of variants to
determine which linguistic or social contexts favor or disfavor any one form. To
obtain this information, a quantitative analysis is required.
Indexing analysis
Labov noted that the θ and ð sounds could be produced as the standard
fricatives or as nonstandard stops (the stereotypical dese and dose instead of
these and those). Sometimes, however, they were produced as affricates—an
intermediate stage between fricative and stop. Each possible pronunciation was
assigned a numerical value reflecting how closely it approximated the standard.
These values are shown in Table 1.
Contemporary Linguistics: An Introduction, 5th edition, Chapter 15: Variation, 3
Each speaker’s individual score was computed with the numerical values for the
possible pronunciations, using the formula in 4):
Thus, an individual who produced all stops and no fricatives or affricates would
have a score of 200, whereas an individual who produced all fricatives would
have a score of 0. Indexing scores for individuals and for groups can then be
compared. Table 2 shows the index scores for stopping of the interdental
fricative by two individual New Yorkers: Bennie N. was a truck driver who had
not finished high school, and Miriam L. was a practicing lawyer who had
completed college and law school.
Marginal analysis
In fact, this is probably the best way to determine rates and occurrences
of variants by context. However, since this variation may be subject to so many
possibly interconnected factors, more sophisticated statistical methodology
serves as a valuable tool for determining (1) if distinctions are significant and (2)
if cross-cutting factors are falsely inflating or obscuring variation. Variable rule
analysis is ideal for this purpose.
Variable rule analysis (VRA) is an important and frequently employed tool in the
modern study of variation. Using statistical procedures specifically adapted for
linguistic data, it indicates not only the probability that one variant or another
will occur in any context but also whether there is a statistically significant
difference between those contexts. More importantly, we can also find out which
Contemporary Linguistics: An Introduction, 5th edition, Chapter 15: Variation, 5
The speakers with the highest rate of nonstandard variants for the
variables in this component were found to be the older men in the community.
Contemporary Linguistics: An Introduction, 5th edition, Chapter 15: Variation, 7