The Conlang FAQ

Morphophonemics

adapted from a 25 Nov 96 post by Mark Line

Timothy Miller wrote (responding to Jack Durst):

So you think that Morphophonemics should go with Morphology? Hmmm... well, as long as the reader is already familiar with phonology, this will not be a problem.
Morphophonemics (or more commonly, morphophonology) deals with the conditioned alternation of phonemes. Although the occurrence of particular morphophonemes is controlled lexically just like (non-morphophonologically alternating) phonemes, morphophonemic alternation can be described in terms of phonology alone, with no recourse to the morphological processes involved. You have to have a "layer" of description that handles the predictable alternation of phonemes -- this is morphophonology.

For example, most people here are probably aware of the vowel harmony processes in Turkish. These processes involve the alternation of _phonemes_ (not allophones) according to certain patterns. The same patterns recur in various, otherwise unrelated affixes, and can be described without recourse to the particular _morphological_ processes involved. For instance: plural is /ler/ or /lar/, locative is /de/ or /da/; and definite accusative is /i/, /I/, /y/ or /u/, genitive is /in/, /In/, /yn/ or /un/, copula is /dir/, /dIr/, /dyr/ or /dur/, and so on. It makes very much sense to say that Turkish has two "morphophonemes" (it has more than two, in fact), call them //e2// and //i4//, which always get realized as one of the two or four vowel phonemes according to very predictable patterns that involve the phonological, not the morphological, context of the utterance. Note that any description of Turkish will have to describe the gaps in productivity of vowel harmony which occur (to my knowledge) only in loan words (in which the source language, e.g. Arabic or French, can provide material that is vocalically disharmonious from the Turkish point of view). But the productive occurrence of vowel harmony in Turkish can be described morphophonemically pretty much in one fell swoop.

Languages such as English, that had an abusive childhood, often maintain relics of morphophonological processes that were productive in languages providing new lexical material (Norman French, for example), so it's important to be very careful about which alternations you think are really morphophonological and synchronically productive, and which (such as Modern Standard German verb stem ablaut) may best be described lexically, not (morpho)phonologically (except perhaps in a diachronic description).

I would therefore advise that morphophonology be seen as part of the linguistic code that deals with the predictable alternation of sounds, not of morphemes, and include it under phonology -- or at least not include it under morphology

The reason, by the way, why "morphophonology" is called such, reminding one of morphology in a way that is most certainly not warranted, is because traditional philologists and early structuralist descriptions would have listed /ler/ and /lar/, /dir/, /dIr/, /dyr/ and /dur/ and so on as different allomorphs of the (single) plural morpheme and of the (single) copula morpheme, respectively. Since there is very obviously something phonological going on in a phenomenon that was being described as morphological variation, the term "morphophonology" was used to describe such processes. Later structuralist descriptions, and most descriptions of any kind today, deal explicitly with the concept of morphophoneme -- thus capturing the phonological alternation as such, and allowing just a single allomorph (//le2r//, //de2//, //i4n// and //di4r//) to be listed as a sequence of phonemes and morphophonemes.

Well, the thing is, it isn't just nouns that can be an open category. I recall that in some languages, even pronouns are an open category.
I think that most languages use open-class nouns anaphorically, but I'd be hesitant to refer to any open class as "pronouns". I've certainly never seen a description of any language that would warrant this usage.


Return to Conlang-related topics|Back to FAQ page

Copyright © 1997, Jack Durst,
Last updated: 21 Jun, 1997