Looking at the ability to shift from AAE to SAE in formal situations in a quantitative way is vital to answering the questions posed in this dissertation. But in order to study shifting
behavior, it is necessary to define “style shift” itself. This term is largely discussed in the literature in a descriptive way, talking about how speakers alternate their speech among multiple dialects. A definitive method for approaching the study of style use in a way that is useful for quantitative analysis has not been agreed upon, however. Just as how the best way to define
style has been debated widely in the variationist literature, there are many possibly ways one might define “style shift” operationally. In §2.4.2, it was noted that one of the problems often cited with respect to large-scale DDMs is the fact that they only track overall shifts in language use. This likely oversimplifies the true nature of shifting behavior, which may exist on
numerous levels given the intrinsic variability of AAE and all language varieties. For example, as discussed in §2.3, AAE not only overlaps considerably with SAE and other nonstandard varieties like Southern English, but also has a great deal of grammatical and regional variation within itself. Additionally, there is variation surrounding individual features, like copula absence, where in many cases it is grammatical within AAE to either include or omit it. While these concerns are valid, it is somewhat unavoidable at this stage in the study of language. As I will discuss in this chapter, as well as in §5.1, other complementary methods of quantifying language use should be employed to add to the information garnered through more
comprehensive measures of AAE. As this and other research work toward this end, however, it is necessary to operationalize style shift in some manner. In this dissertation, then, I will first look at style shift as a difference between the DDMs calculated using language data from the informal contexts and the DDMs computed from the formal data (§4.2.1); in subsequent analyses, I will look at the ratio of the informal DDM to formal DDM for each speaker at each time point to gauge style shift (§4.2.2). The rationale for utilizing each method will be discussed in its respective subsection.
4.2.1 Style Shift as a Difference Score
Although style shift might be defined in a variety of ways, one recent operational definition reduces it to a “difference score” (Craig et al., 2009). Craig et. al. conducted an analysis of the
unstandardized DDM scores from oral and written contexts using the following calculation: Oral DDM – Written DDM = Individual DDM Shift Score (849). In their analysis, a positive
individual shift score meant that a speaker had used more AAE in the oral context and then shifted to a lower DDM in the written context. They considered a speaker with a negative or no difference in individual shift score as a nonshifter. They found that speakers with a positive difference score, i.e., those who shift to SAE during reading tasks, outperformed their
nonshifting peers on standardized measures of reading assessment. While this work compared the use of AAE in oral and written language contexts, this type of comparison is similar to the formal vs. informal dichotomy used in this dissertation. In §4.3, this technique is utilized to provide a basic description of individual speakers’ contextual style shifting behavior at all three temporal data points by comparing speakers’ DDM difference scores in informal and formal situations.
4.2.2 Style Shift as a Ratio
Another way to consider shifting is as a ratio of two DDM scores; in the case of this work, style shift was assessed as a ratio of the informal DDM to the formal DDM. While this method is similar to the idea of a difference score, it has some added benefits. First, a ratio accounts for the fact that a difference of 10 AAE features may in some cases be a large difference but in others a small one. For example, if one speaker uses 10 AAE features in formal situations and 20 in informal ones, there is a 100% increase in nonstandard feature use from the formal to informal context. Another speaker might use 100 AAE features in the formal and 110 in the informal context. As with the first speaker the difference in the feature count is 10, but there is
only a 10% increase from the formal to the informal context. Thus, one could argue that a ratio offers a more precise method of capturing the extent of change in linguistic behavior.
A related benefit is that a ratio is an interpretable value that has a clear meaning. For instance, a ratio of 1.6 would indicate that the speaker used 60% more AAE in informal scenarios versus formal situations; thus, the researcher knows not only that the speaker uses more AAE in informal situations, but he or she also has an indication of how much more AAE is used. A difference score, however, is not as easily interpreted. A difference of 0.02, for
example, only signifies that the speaker uses more AAE in informal contexts; it unclear whether this is a large change in language use or a small adjustment. As a result of the reasons discussed here, the ratio score was used for the statistical analyses in this project.