How abstract is phonology?

On the one hand we have argued for abstract analyses of Matuumbi, Yawelmani, Maltese, and other languages; but we have argued against abstract analyses of English. The reason for this apparently inconsistent view of abstractness is that abstractness per se is not the issue; the proper question to be focusing on is what motivates an analysis. Thus we conclude that the formal theory of grammar imposes no constraints on the relation between underlying and surface forms, though the theory does state what kinds of elements can exist in underlying representations: phonetically interpretable combinations of features, i.e. segments.

This does not mean that highly abstract underlying representations can be gratuitously assumed. Underlying representations require motivation: they must be acquired by children learning the language, and the best assumption to make is that in lieu of evidence to the contrary, underlying and surface forms are identical. The question that needs further investigation is, what constitutes valid “evidence to the contrary”? Phonological alternations in the shape of a morpheme provide very powerful evidence for abstractness. It remains an open question whether other considerations are also valid in constructing an underlying form.

Although we have focused on the relation between underlying and surface forms, the larger question which this debate raises is, what counts as valid evidence for testing a phonological theory. It has proven extremely difficult to resolve questions about the psychological reality of theorized linguistic constructs. Two approaches, both valid, have been taken. One is the “domain-internal” approach, where formal constraints are proposed to the effect that (for example) underlying forms should be a subpart of an actually pronounced word in the language, or underlying forms should only contain segments actually pronounced in the language. We cannot show that these claims are literally “wrong”: what we can do is show that such a position renders us incapable of capturing important generalizations about the phonologies of Maltese and Yawelmani, for example.

The other approach, the “domain-external” approach, seeks evidence from outside the domain of synchronic phonological grammars themselves, in an attempt to find independent evidence that answers the question of what is actually in the mind of the speaker. Any number of such approaches can be imagined – neurosurgery, psycholinguistic testing, language games, historical change, the study of language acquisition, and so on. Such evidence is extremely hard to find in the first place: virtually all relevant experimental work is conducted on a tiny handful of commonly spoken languages, which typically do not have internally well-motivated abstractness. Additionally, the experimental methodology must be critically evaluated, which is usually very difficult to do outside one’s own discipline. Finally the evidence must be interpreted against a general theory of, for example, child developmental psychology. The question of how to empirically validate theory-internal hypotheses remains very much an open question in phonology, as it is in all scientific domains.