In this study, I present a new model for analyzing sociolinguistic variation within the framework of Optimality Theory (OT) and the Gradual Learning Algorithm (GLA). This model contributes to the advancement of sociolinguistic methodology...
moreIn this study, I present a new model for analyzing sociolinguistic variation within the framework of Optimality Theory (OT) and the Gradual Learning Algorithm (GLA). This model contributes to the advancement of sociolinguistic methodology as well as to OT and the GLA, unifying both linguistic and social factors. I propose a number of social constraints and incorporate them with linguistic constraints in the GLA. I show that incorporating social constraints yields variant-usage percentages that match real life occurrences. I follow the evaluation process of stochastic grammar in which social constraints are treated like linguistic constraints on a continuous scale of ranking strictness. Their ranking values may differ from one speaker or group of speakers to another. These differences in the ranking values may result in intra- and inter-speaker, intra- and inter-group, intra- and inter-community, or intra- and inter-dialect variation. The number and type of social constraints involved influence the percentage of occurrence of each variant and affect the ranking values of other constraints and the grammar chosen by a speaker or a group of speakers at a particular time and place.
This model enables us to show the effect of linguistic and social constraints simultaneously. It allows us to investigate individual as well as group grammars and shows that those grammars may act independently from each other. It also enables us to project expected percentages of variation by manipulating either the ranking values of constraints or the pair distribution of the output. Evidence comes from the study of the naturally occurring speech of migrant rural speakers of Colloquial Arabic to the city of Hims, Syria. These speakers show different degrees of variation, particularly regarding the use of the two variants [q] and [ʔ], based on various social factors, such as age, gender, residential area, and social class.
The aim of the study is to show how rural migrants adopt the new phonological system of the urban dialect to appear prestigious. The study will focus on the sound change of [q] to [ʔ] and the four stable variants [t] and [s] and [d] and [z]. It will show that variation may be influenced by different factors. The variable use of [q] and [ʔ] is attributed to prestige and social factors. The use of [t] and [s] and [d] and [z] is not social in nature; it has developed historically from the Standard Arabic (SA) [Ɵ] and [ð] respectively as a response to markedness constraints. Today, this variation is stable and each of the four variants is used in specific lexical items. Faithfulness constraints play a major role in maintaining the pronunciation of the input as [t] and [s] and [d] and [z] in the output. This stable phenomenon is further explained in terms of the two opposing effects of frequency (Bybee 2001).
In a situation where prestige plays a role in adopting a new variant, as in the case of rural migrants to the city of Hims, social constraints are viewed as the motivating force behind changing a grammar at a particular time and place. However, one should take into account that the activation of social constraints depends on the speaker’s selection or choice to activate them or not. Hence, these constraints may rank very high in the speech of a speaker who is highly aware of the social values attached to a certain sound. On the other hand, they may rank very low in the speech of a person who, for example, does not care to adopt a new form.
A theory that can take into account all the factors that lead to variation is a better theory than one that takes into account social factors in isolation and ignores grammatical factors or vice versa. Integrating social constraints into formal theory and showing that social constraints can have equal weight or even more weight than grammatical constraints in conditioning variation and change is essential in this study. Including social constraints in the computation provides explanation of the observed sociolinguistic variation between [q] and [ʔ] among members of the same social group or among different speakers or social groups. Interacting social constraints with linguistic constraints in the same framework is a simple comprehensive method to depict and explain the mental process of a speaker at a certain time and/or place. Feeding the GLA with the right output distribution gives the specific ranking values or grammar of each speaker or group of speakers. In other words, the model takes advantage of the stochastic grammar embedded in the GLA to generate grammars that match real life output percentages without trying all the possible rankings of constraints and counting the times those rankings give a certain output. Furthermore, where a statistical analysis fails to indicate the interaction between one social factor and others, the specificity implemented in the GLA by dividing a social factor into a number of social constraints enables us to see interaction among the same social constraints that emerged as insignificant in the statistical analysis. In this sense, implementing social factors as constraints and accounting for sociolinguistic variation within the framework of OT and the GLA has an advantage over other theories.