The allo-emic classification of morphemes

"Allo-emic" theory was put forward by Descriptive Linguistics. In accord with this theory, lingual units are described by means of two types of terms: allo-termsand eme-terms. Eme-terms denote the generalised invariant units of language characterised by a certain functional status: phonemes, morphemes. Allo-terms denote the concrete manifestations, or variants of the generalised units dependent on the regular co-location with other elements of language: allophones, allomorphs.

The allo-emic identification of lingual elements is achieved by means of the so-called "distributional analysis". The immediate aim of the distributional analysis is to fix and study the units of language in relation to their textual environments, i.e. the adjoining elements in the text. The environment of a unit may be either "right" or "left", e.g.: un-pardon-able. In this word the left environment of the root is the negative prefix un-, the right environment of the root is the qualitative suffix -able. Respectively, the root -pardon- is the right environment for the prefix, and the left environment for the suffix. The distribution of a unit may be defined as the total of all its environments; in other words, the distribution of a unit is its environment in generalised terms of classes or categories.

In the distributional analysis on the morphemic level, phonemic distribution of morphemes and morphemic distribution of morphemes are discriminated. When allomorphs can be determined by phonological criteria, we say they are phonologically determined.Consistent phonological conditioning is rare. Even for the relatively simple English third singular present morpheme, some verbs follow a different pattern. The auxiliary can, for example, though phonologically identical with the verb can, makes its third singular present with no ending, as in She can go versus She cans apples. We can call this a zero allomorph. Since we can determine the distribution of the zero only from the type of verb, we say it is morphologically conditioned.

The study is conducted in two stages.

At the first stage, the analysed text (i.e. the collected lingual materials, or "corpus") is divided into recurrent segments consisting of phonemes. These segments are called "morphs", i.e. morphemic units distributionally uncharacterised, e.g.: the/boat/s/were/gain/ing/speed.

At the second stage, the environmental features of the morphs are established and the corresponding identifications are effected.

Three main types of distribution are discriminated in the distributional analysis, namely, contrastivedistribution, non-contrastivedistribution, and complementarydistribution.

Contrastive and non-contrastive distributions concern identical environments of different morphs. The morphs are said to be in contrastive distribution if their meanings (functions) are different. Such morphs constitute different morphemes. Cf. the suffixes -(e)d and -ing in the verb-forms returned, returning. The morphs are said to be in non-contrastive distribution (or free alternation) if their meaning (function) is the same. Such morphs constitute "free alternants", or "free variants" of the same morpheme. Cf. the suffixes -(e)d and -t in the verb-forms learned, learnt.

As different from the above, complementary distribution concerns different environments of formally different morphs which are united by the same meaning (function). If two or more morphs have the same meaning and the difference in (their form is explained by different environments, these morphs are said to be in complementary distribution and considered the allomorphs of the same morpheme. Cf. the allomorphs of the plural morpheme /-s/, /-z/, /-iz/ which stand in phonemic complementary distribution; the plural allomorph -en in oxen, children, which stands in morphemic complementary distribution with the other allomorphs of the plural morpheme.