Theory, Model, Method

Scientific standards

Linguistics is assumed to be an empirical science. The minimal standards for such a science are:

  • The inquiry must deal with perceivable data of a certain phenomenon,
  • the statements about the phenomenon (hypotheses) must be objective,
  • these statements must be logical and coherent (no contradiction in a statement and between statements),
  • they must be systematically ordered (in addition to coherence),
  • statements must be formulated so that they can be proven wrong or inadequate (if they are).

Thus, the quality of a theory and its methods must be judged according whether they lead to results that meet these test criteria; in general, a theory should be formulated so that it can be proven wrong (falsifiability) facing data on the dimensions above.

Other criteria concern:

  • the level of adequacy (see above),
  • generality,
  • simplicity.

A theoretical statement is adequate to the extent that it applies to all the known data which it is established to explain (cf. minimal observational adequacy). It is general insofar as it posits theoretical constructs beyond observed phenomena, and can therefore apply to the greatest amount of yet undiscovered data. The last criterion of simplicity is an internal one. It is used for the reliable and transparent decision between alternative and competing theories. The simplest theory is to be preferred , given that all other aspects are fulfilled in an equal way. A theory containing fewer theoretical terms and constructs and/or fewer rules (and no contradictions) in order to explain a given phenomenon is simpler (and better) than a theory with more content. This criterion is necessary because in all empirical sciences data always underdetermine the possible theories, i.e. there is never a sufficient set of data to decide only on one specific theory.

Summarizing, the adequacy of a theory (or model) is tested on several dimensions:

  • the extent to which the theory explains the phenomena that it is supposed to explain (empirical/observational and descriptive power, coverage of data),
  • the extent to which the theory makes correct predictions for old and new data of the phenomenon under examination (explanatory and predictive power), and
  • the extent to which it fits with other theories that deal with related facts (cross-theoretical coherence, independent motivation).

According to Chomsky's generative grammar, the justification statement includes any statement involving his notion of descriptive and explanatory adequacy of a theory.
Source of the picture:

Exercises on Scientific Standards