The
Creolization of Pidgin: A Connectionist Exploration
Javier Marín Serrano, Javier Valenzuela
Manzanares and Francisco Calvo Garzón
Universidad de Murcia
Pidgins are made-up languages which lack many features of natural languages.
Children exposed to a Pidgin environment develop as their first language
a Creole; a full-fledged language that exhibits the full variety of
features that the majority of languages share. According to Bickerton
(1984), Creole languages reflect innate Chomskian linguistic structures.
The process of expansion by means of which a Pidgin develops into a
Creole is to be explained on a genetically-based universal grammar,
and does not depend upon the "superstrate languages" that
children may be exposed toi.e., the natural languages that may be spoken
in their community. In this paper we propose to review the Pidgin/Creole
debate under the light of Connectionist theory. For methodological reasons,
we ran three sets of simulations, approaching the issue in a incremental
manner: First (i), we trained a number of simple recurrent networks
(SRNs)Elman, 1990on a prediction task. The SRNs were exposed to a pidgin-like
toy language. Unsurprisingly, no significant grammatical categories
emerged, given the empoverished pidgin input. In a second phase (ii),
the networks were trained on a mixture of Pidgin and different Superstrate
languages, respectively. The networks were then tested on novel Pidgin
and Superstrate
sentences. Creolization was assumed to take place when the network predicts
a Pidgin lexical item as a next item in a (Superstrate-like) grammatically
complex input sequence. In the final stage (iii), networks were exposed
to a mixture of Pidgin, Superstrate languages and Creole. Comparing
competency in prediction across networks, we suggest, may shed light
on the issue of whether Creole languages reflect universal linguistic
principles. By reading grammatical competency off the prediction task
in terms of the statistical regularities in the linguistic pool of data,
Bickerton's position may be challenged.
References
Bickerton, Derek (1984) "The Language Bioprogram Hypothesis"
The Behavioral And Brain Sciences 7, pp. 173-221.
Elman, Jeff (1990) "Finding Structure in Time" Cognitive
Science 14, pp. 179-211.
Marcus, Gary (2001). The Algebraic Mind. Cambridge (Mass): The MIT
Press