Unifying entrenched tokens and schematized types as routinized commonalities of linguistic experience
Abstract Proponents of usage-based models of language acquisition, language structure and language change widely agree that the repetition of specific tokens of words and strings in language use (e.g. give me a break) is conducive to their entrenchment and has a stabilizing and conserving effect, while the repetition of different instantiations of a variable type or pattern (give me a kiss, give me a smile, give me an amen) fosters schematicity and productivity (give me a(n) X). In this paper, I will argue that token-entrenchment and type-schematization are subserved by the same repetition-driven cognitive mechanism. Commonalities observed in linguistic input and output become routinized by repeated activation of patterns of associations. Token-entrenchment and type-schematization do not differ qualitatively but only quantitatively with regard to the variability of what is noticed as being similar. I argue that any form of routinization requires an abstraction over differences between episodes in terms of pronunciation, cotext and context. Therefore, schematization is an inherent component of routinization, but routinization is clearly the more fundamental cognitive process and learning mechanism. I argue that routinized patterns of associations can do the job of constructions in a more flexible, dynamic and parsimonious way and illustrate the potential of this idea with the help of data and insights gleaned from Schonefeld (2015).