Levels and strata in linguistic modeling: Cross-domain considerations
Louise McNally
August 2023
 

Linguistic phenomena are complex, and, as a result, appeals to distinct forms of representation are inevitable in modeling distinct domains of language such as phonetics vs. syntax. However, there is also a long tradition in linguistics of using multiple representations to analyze phenomena assumed to fall within a single domain of language. The positing of “deep” and “surface” structures starting from early Generative Grammar is just one example. These sets of multiple representations have been variously referred to as “levels” or “strata”, often without it being clear what, if anything, was at stake in the choice of terminology. However, the decision to use one or more representations to model a single domain of language has theoretical, empirical, and methodological consequences. Ladusaw 1985 highlighted an especially crucial, but underdiscussed, aspect of this decision: whether multiple representations involve the same theoretical vocabulary (multiple strata) or different theoretical vocabularies (multiple levels). In this article I update this discussion by showing how decisions about multiple syntactic representations look rather different once one sees a reason to posit multiple semantic representations. The discussion highlights the importance of not just within-domain but also crossdomain considerations when making representational choices in linguistic modeling: Decisions about the deployment of representational levels and strata in one domain can influence our understanding of (and modeling decisions about) another.
Format: [ pdf ]
Reference: lingbuzz/007715
(please use that when you cite this article)
Published in: to appear in the Oxford Handbook of the Philosophy of Linguistics
keywords: level; strata; linguistic representation; syntactic theory; semantic theory, semantics, syntax
Downloaded:261 times

 

[ edit this article | back to article list ]