Andrew Radford - Linguistics An Introduction [Second Edition]

Supposing that there is some development of functional category systems (i.e. it is not the case that the child completely controls all aspects of all functional categories from the very earliest stages of acquisition), we can immediately pose another developmental question. We have already seen that even in a language like English, which is relatively impoverished morphologically, there is quite a variety of inflectional endings (third person singular present -s, past tense -ed, progressive -ing, perfect/passive -en, plural -s, comparative -er, superlative -est, etc.), along with a rather rich set of derivational and compounding processes and various other functional categories containing free morphemes (members of D, AUX, PRN, etc.). Are these items acquired in any determinate sequence? Indeed, what sort of evidence should we accept for these items being acquired at all? We turn to consideration of these questions.

Apprentices in morphology

Consider the plural morpheme in English. In section 10 (exercise 5a), we have suggested that with a number of well-known exceptions, the allomorphic realisation of this morpheme as /-s/, /-z/ or /-əz/ is predictable by taking account of the phonological characteristics of the final segment in the singular form of a noun. Thus, the plural form of cats (/kæts/) will not appear in the lexical entry for cat; adults, it is assumed, have access to this regular morphological process, i.e.

they control a morphological rule. Do we have means for ascertaining whether young children control this rule?

Firstly, note that the mere fact that young children produce appropriately

inflected tokens of cats, dogs and buses, while suggestive, does not provide conclusive evidence for them using the above rule. This is because there is

every reason to believe that they will have heard tokens of the appropriately

Children and words

189

inflected forms in which we are interested. Surely, they could simply have

committed them to memory and either include /kæts/ as part of the lexical entry for cat, indicating that it is the plural form (precisely what we would advocate for feet and men), or list it as a quite separate lexical entry, thereby failing to acknowledge any systematic relationship between cat and cats.

In a celebrated experiment reported in 1958, Jean Berko devised a technique

which enabled her to distinguish the above alternatives. Acknowledging that

existing forms could not be used to demonstrate the child’s command of rules, Berko invented some simple words, which she introduced to children in a specific context. For plural allomorphy, her technique was to show the child a picture of a single bird-like creature and say this is a wug (/wʌg/). Then, the child was shown a picture of two of these creatures and prompted with now there are two of them, there are two … The child was to supply an appropriate form. Now, if the mechanism for acquiring plurals requires children to be exposed to every specific example, they should be unable to complete the Berko test. However, the overwhelming majority of children tested responded with wugs (/wʌgz/). Note,

furthermore, that the form the children supplied contained the correct allomorph of the plural morpheme (/-z/). As well as plurals (for which there were several other items to test other allomorphs), Berko devised ways of investigating other aspects of inflectional and derivational morphology. While her results were not always as clear-cut as in the case of plurals, overall she established that children in the age range five to seven do exhibit creative control of a variety of morphological processes. In fact, evidence for this is available from a different source, the spontaneous speech of English-speaking children, and from a much earlier age (exercises 2 and 3).

In a seminal study of the 1970s, Roger Brown and his colleagues at Harvard

reported the results of their detailed longitudinal work with three children. This study had many aspects, but here we shall concentrate on what Brown referred to as ‘14 grammatical morphemes’. This set included a number of verbal inflections and here we shall restrict our attention to these. Within this group, Brown

distinguished between regular and irregular past tense inflections (as in jumped and came) and between regular and irregular third person singular present

(as in walks and does, where the latter involves a vowel change as well as the addition of -s). Completing his list was the progressive inflection -ing.

When we work with samples of naturally occurring production data, it is

necessary to formulate a criterion for acquisition. The point is that when children begin to use, say, past tense forms, they do not do so consistently, vacillating for some time between the appropriately inflected form and the base form. Brown

decided that an appropriate criterion was 90 per cent usage in obligatory contexts, the rationale behind this figure being that once the children in his study satisfied this criterion, they continued to do so; setting the criterion lower would have entailed that children moved from not having acquired a morpheme to having

acquired it, only to subsequently return to not having acquired it. With this methodological decision in place, it was then possible to determine the point at

190

words

which each of the verbal inflections was acquired. The ordering which emerged is in (181):

(181) 1.

progressive -ing

2.

past tense irregular

3.

past tense regular

4.

third person singular present regular

5.

third person singular present irregular

To begin with, we attend briefly to the fact that the progressive morpheme

comes first in this ordering. One possible reason for this is simply its regularity.

Unlike the past tense and third person singular morphemes, the progressive has no variant realisations as allomorphs (although, see section 16 on the sociolinguistic variable (ing)). As a verbal suffix, it attaches in a fixed form to the vast majority of English verbs, and this, coupled with its relatively transparent semantics in signalling on-going activities, may be sufficient to account for its accessibility to children. Of the remaining four items, the third person singular present forms will not delay us. There are very few irregular allomorphs of this morpheme (does, says [sɛz], has, is), and it is perhaps hardly surprising that these forms are relatively late in being acquired.

The surprise package in (181) is provided by the past tense allomorphs, with the irregular forms meeting Brown’s 90 per cent criterion before the regular forms. Of course, there are more irregular past tense forms than there are irregular third person singular forms, but they are far outweighed by the regular forms, and in these circumstances, intuition suggests that the regular pattern would prevail first.

There are two observations bearing on this order of acquisition. Firstly, the irregular forms, while relatively small in number, include some of the most

frequently occurring verbs in English (was, had, came, went, brought, took,

etc.). Secondly, the regular pattern does indeed prevail but only after a period during which the irregular forms are correctly produced. A consequence of this is the phenomenon of overregularisation, when the child incorrectly applies the regular past tense formation rule to a base form which, in the adult language, requires an irregular process. The result is a stage at which the child’s performance on such past tense forms as went and came deteriorates, as these forms are partially replaced by *goed and *comed. It is forms such as these, typically occurring in the child’s third year, which demonstrate that the child is operating in a rule-governed fashion. Such forms are very uncommon in the speech children hear (adults can be induced to overregularise in this way if, for example, they are asked to produce past tense forms under time pressure) and it would be fanciful to suggest that, having apparently successfully mastered the irregular forms, children abandon their mastery on the basis of a very unusual occurrence. It is more plausible to suggest that overregularisation is indicative of reliance on a rule system