Suppletion
Consider the paradigm of inflection for degree in English adjectives, which is illustrated in (5). The irregular comparative and superlative forms for good and bad seem totally unrelated to the basic (positive) form. Better and best do not contain any trace of the root form good, unlike such irregular plural forms as ox/oxen, child/children, criterion/criteria, etc. in which the original root can still be seen in the plural form.
An alternation like good/better/best, in which the inflectional paradigm for a certain word involves more than one root form, has traditionally been referred to as SUPPLETION. Other examples in English include the irregular verb forms go–went and am–is–are–was–were. These suppletive forms of the root cannot be derived or predicted by any regular phonological rule, but must be listed in the lexical entry of the word.
Total suppletion (the existence of two unrelated roots or stems for the same word, as in go–went) occurs primarily as a marker of inflectional categories, and rarely (if ever) in derivational processes. Since derivational morphology is often semantically irregular and does not form paradigms, it would actually be very difficult to identify suppletive stems as belonging to the same morpheme. For example, on semantic grounds we might be tempted to say that kill is the causative form of die, or that drop is the causative form of fall. But there is no evidence that these pairs of forms are related morphologically; the best analysis seems to be that they are distinct (i.e. unrelated) lexical items which share certain components of meaning.
Modern linguists extend the term SUPPLETION to apply to affixes as well as roots.1 For example, the two allomorphs of the Huichol possessive marker in (4) stand in a suppletive relationship because neither form can be derived from the other by a phonological process (one is a prefix, the other a suffix). Since there is no way to predict which allomorph occurs with which noun, the choice must be indicated in some way in the lexical entry of each noun. This is an example of LEXICALLY CONDITIONED SUPPLETION.
The two forms of the Korean nominative case marker in (2) also stand in a suppletive relationship. However, in this case it is possible to predict which form will occur with which noun:-ka occurs after a vowel, while-i occurs after a consonant. This is an example of PHONOLOGICALLY CONDITIONED SUPPLETION, meaning that the choice of allomorph depends only on the phonological environment.
A third possibility is that affix suppletion may be MORPHOLOGICALLY CONDITIONED. This means that the choice of allomorph for a particular affix depends on what other specific affixes are present in the word. For example, Muysken (1981) reports that in Quechua verbs, the normal third person agreement suffix–n is replaced by a zero allomorph–Ø when the verb also bears the past tense suffix–rqa.
Morphologically conditioned suppletion appears to be less common than the lexical or phonological conditioning discussed above; at least it is harder to find cases where this is clearly the correct analysis. As another example, the future tense prefix in Lalana Chinantec has two allomorphs: /rɨ 2-/ when the verb is marked for first or second person vs. /rɨ23-/ when the verb is marked for third person. (The number 3 represents low tone, 1 is high tone.) This change in tone is not phonologically conditioned. The subject agreement suffixes appear in several different allomorphs, which are lexically conditioned. However, the tone pattern associated with the suffix does not affect the tone of the prefix; this seems to be determined strictly on the basis of person, as illustrated in (6).2
Haspelmath (2002) cites Welsh pluralization as another example of morphologically conditioned suppletion. Welsh has about a dozen different ways of marking plural number on nouns. In derived nouns, the form of the plural marker depends on which nominalizing suffix is used, as illustrated in (7).
We will discuss the kinds of rules we could use to represent each of these kinds of suppletion. But before proceeding further with our study of suppletion, it will be helpful to discuss the other major type of allomorphy, namely MORPHOPHONEMIC change.
1. The term is sometimes also applied to cases where some similarity of form remains between the allomorphs but the relationship is phonologically irregular, as in the plural forms man/men, goose/geese, mouse/mice, etc. So, in this extended usage of the term, any allomorphy which cannot be analyzed as morphophonemic can be considered to be a case of suppletion.
2. This analysis is based on Bill Merrifield’s unpublished teacher’s guide for Merrifield et al. (1987). Thanks to Cal Rensch for additional clarification.
