Word-internal morphophonological processes

However, there is another class of segmental phonological processes. In contrast to the connected-speech processes discussed above, these do not apply across word boundaries, but are rather confined within words, where they tend to take place in response to the addition of a particular suffix – generally those suffixes identified as causing stress retraction . Forms with these suffixes are also prone to odd and irregular segmental processes. For instance, when the suffix -ity is added to electric, the final [k] of electric becomes [s] in electricity. The same suffix may also alter the stem vowel: when -ity is added to divine, sane, serene, the long stressed vowels of the stems are shortened in divinity, sanity, serenity. These changes are also unlike CSPs in that it is often hard to see why they take place where they do: while a fast speech reduction or assimilation is generally a response to speed of speech, and involves ease of articulation pressures, the word-internal type typically creates an alternation between two independent phonemes, not directly motivated by the phonological context (as in the /k/ and /s/ of electric – electricity).

Even where there does seem to be a reduction, as in the shortening of the stressed vowel in divine to divinity on the addition of the -ity suffix, it is not obvious why this particular suffix should have this effect; and it cannot be ascribed to speed of speech, since these morphophonological processes are obligatory, regardless of speed of speech or sociolinguistic factors: hence, the citation forms of electricity, divinity will also show these changes.

Although the affixes which provoke these segmental changes generally also influence the position of stress, this is not always the case. For instance, adding the past tense marker -t or -d to irregular verbs like keep – kept, sleep – slept, leap – leapt has no effect on stress, but does seem to cause a categorical shortening of the stem vowel. One of the most important jobs for phonologists, bearing in mind the focus discussed throughout on what speakers know about their language, and what they must be assumed to do in order to learn, produce and understand it, is to work out where to draw the line between productive processes which speakers apply regularly and which they will generalize to new forms in the language, and fossilized processes which might have started out as regular phonetic developments, perhaps CSPs, in the history of the language, but which are now simply associated with individual words or small groups of words. That is, perfectly natural phonetic processes may in time become less transparent, and less regular. In the case of keep – kept, or divine – divinity, we must ask ourselves whether the processes of vowel shortening, which perhaps were regular and phonetically motivated centuries ago, are still part of native speakers’ active knowledge of English, and still involve those speakers in actual processes of adding suffixes and shortening vowels; or whether children must learn that words like keep and divine have related, but different forms which are stored separately and produced on appropriate syntactic occasions. Since phonology, like all other areas of language, is consistently undergoing change and development, with new processes constantly arising and different accents diverging, our only definite conclusion can be that today’s connected-speech processes will present tomorrow’s phonologists with exactly the same problem.