Andrew Radford - Linguistics An Introduction [Second Edition]

ringed is common in children’s speech, see section 13). Why might this be so?

7.

Take the verbs BE, HAVE, UNDERGO and SEND. Enumerate all

their inflectional forms and transcribe them phonetically. Then seg-

ment each word form into morphemes. How many distinct stems do

we need for each verb? How many forms show partial suppletion and

how many show total suppletion? How many stems are used for more

than one word form in each verb?

11

Morphology across languages

The previous section has concentrated almost entirely on English morphological phenomena. In fact, languages differ considerably in the extent and nature

of the morphological processes employed in their grammars. Vietnamese, for

example, has no bound morphemes, so that the only morphology in the language is compounding. By contrast, there are languages in which morphology is

extremely i ntricate and accounts for much of the grammar ’s complexity. I n this section, we will look at some examples of the types of morphological system that are found in the languages of the world, and the kinds of functions realised by that morphology. A range of the examples we consider will be seen to provide further support for the Separation Hypothesis introduced at the end of the

previous section.

The agglutinative ideal

In the last century, linguists introduced a classification of morphological

systems which is still often referred to today. This classification distinguished isolating, agglutinating and inflectional languages. We start with isolating languages. These, exemplified by Vietnamese, Chinese and a number of other Far

Eastern languages, as well as a number of West African languages, have few, if any, bound morphemes. Thus, in Vietnamese, there is no morpheme corresponding to

English -er in driver, this concept being conveyed by a compound with roughly the structure ‘drive + person’.

At the other extreme are languages such as Turkish, Finnish, Hungarian, the

Bantu languages of Africa, many languages of the Americas and Australasia and most of the languages of Russia. Here, words of great complexity, consisting of many morphemes, are formed. A (fairly typical) word from the classic example of an agglutinating language, Turkish, appears in (121) (note that this example uses the orthographic system of Turkish):

(121)

çalıştırılmamalıymış

‘apparently, (they say) he ought not to be made to work’

The segmentation of this word into its component morphemes is indicated in (122): (122)

çalış-

tır

– ıl – ma – malıy – mış

work

cause

passive negation obligation

inference

156

Morphology across languages

157

The root, the verb çalış ‘work’, comes at the beginning and the suffixes each add their own component of meaning.

Languages such as Turkish give the impression that every morpheme has just

one meaning and every meaning in the language is assigned its own unique

morpheme. This is often thought of as a kind of morphological ideal, with

the characterisation of such languages as agglutinating conveying the idea that morphemes are glued together one by one.

It is indeed the case that a ‘perfect’ isolating or agglutinating language would have the property that every morpheme would have just one meaning and every

individual component of meaning expressible in the language would correspond to just one morpheme. The difference between the two types would be that in an agglutinating language some of the morphemes would be bound, giving the

possibility of the construction of complex words like that in (121), whereas in an isolating language they would all be free. In practice, however, there are innumerable deviations from such ideals, and it’s unlikely that any language has ever met the ideal. Moreover, there are many languages which show, say, agglutinating tendencies in some areas of grammar and isolating tendencies in others.

For this reason, it is much more interesting to ask whether specific morphological processes are isolating, agglutinating or something else. Whether a language can be so categorised is something of a non-question. With this background, we can now ask more detailed questions about the kinds of inflectional systems we find in the world’s languages.

We begin by contrasting two languages, Latin and Turkish. In tables 16 and 17, we see sets of forms of the Turkish noun EV ‘house’ and the Latin noun VILLA

‘villa, country house’.

Table 16 Forms of the Turkish noun EV ‘house’

singular

plural

nominative

ev

evler

accusative

evi

evleri

genitive

evin

evlerin

dative

eve

evlere

ablative

evden

evlerden

Table 17 Forms of the Latin noun VILLA ‘country house’

singular

plural

nominative

viːlla

viːllae

accusative

viːllam

viːllaːs

genitive

viːllae

viːllaːrum

dative

viːllae

viːlliːs

ablative

viːllaː

viːlliːs

158

words

These nouns each have sets of singular and plural forms, but in addition they have case forms. A case form of a noun is a special form used to indicate various types of grammatical relationship. Roughly speaking, the functions of the cases are as follows: nominative – the basic form of the word; accusative – the form used when the noun is the object of the verb undergoing the action denoted by the verb (e.g. They painted the house); genitive – possession, of the house, dative – to/for the house, ablative – away from the house. However, the meanings are not so

important here; our focus is on the way the words are constructed.

The first thing we notice about the Turkish forms is that there is a single set of case endings which are used for both singular and plural: -i, -in, -e, -den.

Moreover, the exponent of the plural for all case forms is the suffix -ler. When we turn to the Latin forms, however, the picture is much less clear. Firstly, there’s no single suffix which expresses the property ‘plural’. Moreover, the case endings for the singular and plural don’t correspond to each other at all. In fact, it’s worse than this because the nominative plural form is identical to the genitive and dative singular forms. And yet, when we investigate the Latin noun system, it’s clear that we need to distinguish the five cases and the two numbers, because all nouns have sets of distinct forms for the various case/number combinations. The problem is that each Latin noun is only able to take a single suffix. Therefore, each suffix has to be simultaneously the exponent of two properties, number and case. When a single affix expresses more than one property within a word form in this way, we say that the affix cumulates those properties, and the phenomenon in general is called cumulation.

Latin nouns illustrate a further important feature of inflecting languages. The endings of the Turkish word KEDI ‘cat’ are essentially the same as those in table

16: kediden ‘from a cat’, kedilerin ‘of the cats’ and so on. In table 18, we see the forms of the Latin noun FELES ‘cat’:

There are only vague similarities between the endings for VILLA and those for FELES. Now, it turns out that there is a very large number of words which take the same endings as VILLA and quite a few which take the same endings as FELES,

so this is not just a case of isolated irregularity. Moreover, there are other patterns of endings for other groups of nouns (traditionally, five such classes are recognised). Distinct groups of words with different inflections to express the same sets of properties are called inflectional classes. The traditional term for inflectional Table 18 Forms of the Latin noun FELES ‘cat’