Linguistics: An Introduction

By William B. McGregor



3.1 — Some difficulties with the notion of the morpheme

It is not all plain sailing with the notion of the morpheme. One difficulty is mentioned in the textbook (p. 73), namely the problem of zero morphemes. It is useful to recognize these morphemes in certain cases, as they can provide a way to formulate morphological generalizations that would be otherwise impossible, as discussed for the Hungarian verb in the textbook (p. 73). These have an identifiable meaning but no actual perceivable form. This is somewhat problematic for the idea of the morpheme as a Saussurean sign, since the signifier actually has no phonological form.

Other difficulties also exist. Here are a few:

  1. Lack of consistency in the contrasting forms In the examples discussed in the text, there is considerable consistency in the patterning of the forms of the morphemes in paradigmatic contrast, allowing you to identify a possible form for the morpheme. There may as well be some irregularities that do not follow this pattern, as in the case of the noun plural morpheme of English (e.g. the irregularity for ox). But sometimes no pattern is sufficiently recurrent to allow you to identify it as the regular pattern. Consider the forms of the pronouns of Gooniyandi, where two different case forms are shown, one for subject and object, the other for possessive (don’t worry about the meanings of restricted and unrestricted, this is irrelevant to our concerns here):
      singular plural
    restricted unrestricted
    1 nganyi ‘I’ ngidi ‘we’ yaadi ‘we’
    ngarragi ‘my’ ngirrangi ‘our’ yarrangi ‘our’
    2 nginyji ‘you’ gidi ‘you lot’
    ngaanggi ‘your’ girrangi ‘your’
    3 nhiyi ‘he, she, it’ bidi ‘they’
    nhoowoo ‘his, her, its’ birrangi ‘their’

    Although there are some regularities (e.g. plural number is marked by -di in the first row of forms, and by -rra in the second row), it is not possible to divide the pronouns easily into morphemes so that each can be described consistently in a formula such as we discussed in §3.6. (Try it!) And just four of the seven (a little over a half) of the possessive forms are formed by a pattern from the other forms. (How would you describe the pattern.) Trying to describe these forms in the way described in Chapter 3 is going to give us more exceptions than general patterns.

  2. The identifiable pattern is the removal of phonological forms. Consider the following data from Tohono O’odham (Uto-Aztecan, Arizona USA):
    Incompleted Completed Gloss
    hi:nk hi:n ‘bark’
    ñeid ñei ‘see’
    ñeok ñeo ‘speak’
    galon galo ‘rake’
    si:sp si:s ‘nail’

    If you try to identify a morpheme for Incompleted, notice that for the five words there would be four different forms for it. On the other hand, notice that the Completed forms are regularly formed from the Incompleted forms by removing the final consonant. In this case, you loose generalisations, rather than gain them, by identifying a morpheme.

  3. The pattern seems to be removal of meaning. In Russian there is a morpheme -sja, called an anticausitive, that apparently subtracts the meaning component ‘cause’ from the meaning of the stem it is attached to. Thus, compare the following two words:

    otkryt ‘cause to open’

    otkryt’sja ‘open, become open’

Difficulties like these have led morphologists to think of other ways of describing the morphology of languages than the item-arrangement style discussed in Chapter 3. See the Guide to further reading, p. 75 for some basic references to alternative ‘word-paradigm’ types of morphology.

Updated: Feb. 11, 2009