Ohala 1990, Halle (1954) and a few other very notable scholars have suggested that phonetics and phonology should be unified, denying an interface between the two.

I naively imagine there are three logical possibilities for such a view, and if I understand Ohala correctly, I suppose he argues for (2):

  1. All phonetic phenomena is actually phonological
  2. All phonological phenomena is actually phonetic
  3. All phonetic and phonological phenomena would best be described as something else altogether

Regardless, I wonder if linguists who promote such views share the same/similar vision as to what scientific subdiscipline the two disciplines should be unified as?

2 Answers 2


You need to provide a reference to "Hale (1954)": it is virtually a certainty that the year is wrong, since this would have been an entirely different issue in 1954. There is, however, Morris Halle, who has written many things on the topic. The idea of having unified phonetics and phonology is, in fact, the theoretical starting point of pre-generative phonology Praguian and became a fundamental assumption of generative phonology as represented by Chomsky & Halle 1968, Postal 1968. We can contrast Ohala's view with that of Chomsky & Halle. Both deny that there are separate components of grammar: actually, it is most probable that Ohala does not have a concept of "grammar", in the sense that generative linguistics uses it.

The C&H view is that all language-specific sound rules are implemented by a set of phonological rules, which operate on a universal set of features that can have integer coefficients (though which on can say that a stop in French is "more voiced" that a stop in Swahili). These rules produce a set of instructions to the articulators, and the production of physical sound procedes thereafter in a non-linguistic way. That is, languages do not differ in how nasality is produced in [ana], once the specific integer values of these phonemes are set by rule.

Ohala on the other hand does not accomodate sound-system processes that aren't part of what have been called "postlexical phonology", thus there are no phonological rules of palatalization in Polish, there is no vowel alternation within the root in "dream ~ dreamt", Classical Arabic does not have rules deleting intervocalic glides. It is entirely unclear what his theory of language production is, since it appears that he holds that speakers have learned all of the words of their language and do not generate them from parts.

The basic dichotomy between Ohala and Chomsky & Halle is that for Ohala, everything is historical change, and for Chomsky & Halle, everything is synchronic grammar.

  • Just to make sure I understand correctly: 1) Ohala and C&H (and for that matter Postal) agree that phonetics and phonology are actually one discipline---this is a core tenant of pre-generative phonology; 2) While C&H believe(d) that phonological rules operate on universal features, as you describe; 3) Ohala believes that there is no phonological rules (and possibly no universal features?), presumably, for him, the pronunciation of each word is "stored" in the lexicon (?). Is this accurate?
    – Teusz
    Aug 29, 2016 at 6:16
  • (1) is true of the Jakobsonian thread of PGP. As for (3), he rigorously avoids dealing with phonological alternations, so we can't know what he thinks about "features", although he does believe in phonetic descriptions. Apart from storing words in a lexicon, he also believes in behavioral analogy (his response to wug tests).
    – user6726
    Aug 29, 2016 at 16:11

IMHO O'Hala's problem is that splitting the disciplines introduces a too limited view onto the whole matter. The option (2) is maybe too strong, I would reformulate it that vast majority of phonological phenomena can be better understood or even explained through using phonetics, while phonological outlook is necessary to preserve the connection to the language.

For example, my background is in historical linguistics, which is dominated by phonology, but with the caveat that very soon you will reach a point where you work with just a very abstract structure of the language without any tangible values behind it and when you use e.g. "*d", it means nothing else than "a reconstructed phoneme based, which evolved later mostly into something working like /d/ but in reality could have been anything.

This has two major disadvantages:

1) even people with a linguistic background may find it difficult to detach themselves from the notation and stop realising that *d actually might not have any d-like qualities.

2) you do not have any control of your reconstruction, i.e. whether the system you are reconstructing is plausible/possible

As such, it can be useful sometimes to distinguish between phonetic and phonological approaches, most of the times they are and always have been intrinsically interconnected.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.