Andrew Radford - Linguistics An Introduction [Second Edition]

(Apart from other considerations, as we have just noted, it is virtually impossible in English to form a compound by adding a noun to a verb.)

The situation with regard to inflection is more revealing. Thus, with noun +

noun compounds, we seldom find morphology on the first noun. A dog catcher

is presumably someone who catches more than one dog, yet we don’t say *dogs

catcher, and even if we had a cable for use with several printers we wouldn’t call it a *printers cable. The lack of plurals in this position even extends to words which only ever occur in the plural, so that although there is no noun *trouser, we do have trouser leg and trouser press. There are a few cases of plurals inside

compounds, e.g. systems analyst, arms control, but usually the plural form is more than just a simple plural and involves some change of meaning, suggesting that we have a different lexeme from that linked to the singular form. On the other hand, we do have dog catchers and printer cables. Here, the plural formation rule pluralises the whole compound (exercise 4).

Clitics

Another puzzle about words can be illustrated by the examples in

(117). How many words are there in each of these examples?

(117) a.

it’s

b.

they’ve

c.

she’ll

d.

wasn’t

It will come as no surprise that there are two correct answers. In one sense, it’s is a single word (indeed, it’s just a single syllable), homophonous with (that is, being pronounced identically to) its. However, while its means ‘pertaining or belonging to it’ (its name, its function), it’s means the same as it is or it has. Thus, there is a sense in which it combines two distinct words. The -s, -ve, -ll and -n’t components of the words in (117) correspond to the full words is/has, have, will/shall and not and can be thought of as words. However, they can’t stand alone in a sentence and they can’t be stressed – to be pronounced they have to be attached to some other word (much like an affix). For this reason, they are referred to as bound words.

A similar phenomenon is represented by the possessive -’s of Harriet’s hat. It is often thought that Harriet’s is a suffixed form of Harriet, just as the plural form

Building words

151

hats is a suffixed form of hat. However, this is misleading, because we can have expressions such as the man who Harriet met’s hat or the girl I’m speaking to’s hat. Here, the -’s ends up attached to a verb form (met) or a preposition (to). This is not the normal behaviour of a suffix. What is happening here is that -’s is added to the last word of a whole phrase, the man who Harriet met or the girl I’m speaking to. Unlike the bound word, this type of element never corresponds to a full word and hence it is called a phrasal affix.

Bound words and phrasal affixes are examples of clitics (from a Greek word

meaning ‘to lean’) and the word that a clitic ‘leans’ on is its host. Clitics such as -’s and -’ve appear to the right of their hosts, like suffixes. Such clitics are called enclitics. In other languages, we find clitics which attach to the left side of the host, as though they were prefixes, called proclitics. Pronouns in Romance

languages behave like this. Thus, in (118), the Spanish unstressed pronouns me

‘me’ and las ‘them’ appear immediately before the verb:

(118)

Me las

enseña

me them (he) shows

‘He shows them to me’

When the verb is in the imperative form, however, the clitics follow the verb (they are enclitics):

(119)

¡Enséñamelas!

show.me.them

‘Show them to me!’

Notice that in Spanish orthography the proclitics are written separately, while the enclitics are written as one word with the verb. However, once more this is merely an orthographic convention, which does not bear at all on the status of these items as clitics.

Allomorphy

We noted earlier that when -ity is suffixed to indecipherable, a change

occurs in the suffix -able. Specifically, there is a change in its pronunciation from

[əbl̩] to [əbɪl], a change which is reflected by a change in spelling to -abil-. To look at what is going on here in a little more detail, we will consider a similar, but more regular, case involving the pronunciation of the suffix -al. This creates adjectives from nouns, and its pronunciation also changes when such an adjective is

converted to another noun by the suffixing of -ity. So consider the sets of examples in (120):

(120) a.

nation, nation-al, nation-al-ity

b.

music, music-al, music-al-ity

c.

tone, ton-al, ton-al-ity

d.

origin, origin-al, origin-al-ity.

152

words

In each case, -al is pronounced as a syllabic /l̩/ at the end of the word and as /al/

before -ity. What is happening here is that -ity causes the word stress to move to the immediately preceding syllable. When -al is unstressed, it is pronounced as /l̩/ but when stressed, it is pronounced with a vowel /a/. This is a regular phonological alternation. Thus, we can say that the morpheme -al occurs in two shapes /l̩/ and

/al/ depending on stress. The shapes of morphemes as they are actually pro-

nounced in a word are referred to as morphs, and where two morphs are variants of one morpheme, we say they are allomorphs of that morpheme. The terminology here mirrors that of the phoneme, phone and allophone discussed in section 5.

We have said that the /al/ ~ /l̩/ alternation depends on stress. Since stress is an aspect of the phonology of a word, we can therefore say that the alternation is phonologically conditioned. This means that we can describe the difference

between the two in purely phonological terms. However, this is not true of all allomorphy. In some cases, a word form will be idiosyncratic in that it contains unusual inflections. Thus, the plural form of the lexeme OX is oxen. This is simply a peculiar property of this particular lexeme, and so we say that the plural allomorph -en is lexically conditioned here (exercise 5).

Awell-known irregular verb in English is GO. This has a base form /gou/ and a past tense form /wεnt/, which is completely different. This change in form illustrates the phenomenon of suppletion. Since there is no overlap at all in form between go and went, this is a case of total suppletion. The example of bring ~ brought to which we have already referred (115c) is also a case of suppletion, but as the form /brɔːt/

bears a partial resemblance to the base form /brɪŋ/ (they have the same syllable onset), we say that it is partial suppletion. In these cases, we can’t say that the allomorphy is triggered by some phonological factor such as stress. Again, we have idiosyncratic properties of the lexemes concerned and so further instances of lexically conditioned allomorphy. Of course, it is precisely such lexically conditioned allomorphs which must appear in lexical entries (exercises 6 and 7).

The concept of allomorphy pertains to morphemes, and it encourages the view

that complex word forms consist of strings of morphemes with the form of these morphemes (their allomorphs) being determined by either phonological or lexical factors. However, while this view is attractive in some cases, in others it proves difficult to sustain. We can illustrate the type of problem it confronts by considering again the exponents of the property ‘perfect participle’. These include the endings -ed (walked) and -en (taken), and perhaps in these cases, it is appropriate to suppose that there is a morpheme PERF(ect) which enables us to analyse

walked as walk + PERF and taken as take + PERF, with -ed and -en being treated as lexically conditioned allomorphs of this morpheme PERF. However, we also