Competition and Selection in the Feature Pool1


      I got these ideas obviously from population genetics, which has inspired much of my recent work on language evolution. They are applicable both to contexts of “language acquisition” and to those of language evolution under any of its interpretation. The basic idea is that all speakers of a language contribute to a pool of features from which 1) each learner selects a particular subset that will form his/her respective idiolect and 2) a speaker can select new variants as he/she accommodates his/her interlocutors while they interact. In social ecologies where there are also non-native speakers, their xenolectal (i.e., ‘non-native’) varieties too contribute features to the pool.


      The extent of the xenolectal speakers’ impact on the varieties “acquired” by learners depends on various ecological factors. For instance, where xenolectal speakers do not constitute a critical mass, such as (integrated) immigrants and the learners interact regularly with native speakers, the influence of xenolectal features is usually negligible. There are, however, contact settings in which xenolectal speakers form a critical mass and/or in which non-native speakers interact more among themselves than with native speakers, such as on former plantation settlement and in exploitation colonies. Despite the non-negligible role of the founder effect, xenolectal features have exerted a serious influence on the varieties appropriated by the learners.


      Since learners do not have access to all variants in the feature pool and select features of their idiolects only from among the variants that are accessible to them, contact settings in which different dialects and/or different languages are in contact lead to different outcomes of “language acquisition,” as I explain below. Such variation, which is typically quite stable, was first observed in creole communities and identified as (post-)creole continua. Unfortunately they were also mischaracterized as symptomatic of “decreolization,” as if other speech communities in which a distinction exists between a standard and non-standard varieties exhibit less variation or have clearcut boundaries between standard and non-standard speech.2


      Ordinarily, when one acquires a language in a setting in which native speakers of the same language or dialect interact only among themselves or are the overwhelming majority, the variants that he/she acquires originate only from the subset of the pool that he/she has been exposed to. Differences among idiolects are a function of the fact that 1) every learner has been exposed to a different subset of utterances in the language or dialect, and 2) has recombined the selected variants into his/her idiolect in a somewhat different way, with his/her own individual peculiarities, such as weighting some of the same variants differently. For instance, one may favor help somebody do a job over help somebody to do a job, or it isn’t so over it ain’t so.


      In settings where the feature pool contains xenolectal features, the competition and selection processes are more complex. There is chance that xenolectal features will make their way into some learners’ idiolects and spread within the population of speakers. For instance, new lexical items may be “borrowed” into the language. This is the case of ganja and obeah in Jamaican English or tshipoy and bwana in Congolese French. New syntactic structures may become part of local parlance, such as can, cannot? for elliptical ‘can you?’ in Hong Kong and Singapore Englishes. (Note the absence of the subject pronoun and the conjunction or from the construction, consistent with the grammars of some of the indigenous languages.)


      In the vast majority of cases xenolectal influence either favors an otherwise minority/ recessive/stigmatized variant in the feature pool of “native varieties” or leads a variant to evolve into what appears to be an innovation ex nihilo. Instances of these may be found in the selection of tense-aspect markers and the extensive use of serial verb constructions in creoles.


      Letting the feature pool concept play a central role in language evolution does not negate the role of innovations, which occur every day as people communicate. What this approach suggests is that innovations are normal additions to the feature pool and they are subject to competition and selection, like other features. They may die immediately, when they are copied by no other speakers, or they may spread and remain in competition with other variants. What makes the feature pool idea more interesting is that it acknowledges individual speakers as the unwitting agents of changes in their language or dialect. If you are interested in more of this, click here for a copy my article “Competition and Section in Language Evolution.” (PDF File, Download PDF Viewer)


      1 Note that in this approach, as in biology, the notions of ‘competition’ and ‘selection’ confer no agency to the features themselves. ‘Competition’ is used in reference to the coexistence of variants in ways that make some of them more heavily weighted, positively or negatively, relative to their alternatives. The selectors are the speakers, who assign these values, variably, and select and reproduce them. ‘Selection’ has to do with the specific ways in which they favor some of the variants in their idiolects, and collectively in their dialect and/or language.


      2 The “decreolization” hypothesis partly assumes, incorrectly, that basilectal speakers aspire at speaking the acrolect as a vernacular or that “acquisition” of the acrolect entails loss of the basilect by the learner. Several speakers are multilectal.