Andrew Radford - Linguistics An Introduction [Second Edition]

(43) show, this dialect has not been affected by this process: (43)

toe [tʊu]

tow [tʌu]

rose [ɹʊuz]

rows [ɹʌuz]

moan [mʊun] mown [mʌun]

So far, we have looked at a number of essentially independent sound changes.

In the case of many vowels, however, linguists have noticed that a change to one vowel can have a knock-on effect for others in the neighbouring area of phonetic space, where we understand this notion in terms of the vowel quadrilaterals from section 2. Sometimes cases arise in which one vowel will change and leave a

‘space’ into which a second vowel moves. It is not uncommon for several vowels to be linked together in this way in a series of changes known as a chain shift.

As we saw briefly in the main introduction, while our knowledge of the

linguistic changes that have occurred over time is largely based on diachronic research – a detailed analysis of the gradual historical development of a particular linguistic feature – methods which can accurately chart language changes as they take place within a community of speakers have recently been introduced. These so-called apparent-time methods involve the simulation of a historical dimension within a synchronic study, and apparent-time researchers collect recordings of the language varieties used within a particular community and compare the

speech of people born at different times. By comparing the speech of those

born in 1920 with that of those born in 1970, it is claimed, we are comparing the language acquired by children at two distinct points in the history of the language. The language of the older speakers should therefore reflect an earlier stage in the development of the language than the varieties spoken by the younger age groups.

Apparent-time studies have enabled linguists to observe some quite complex

examples of chain shifting in progress. For example, William Labov and his

colleagues have carried out extensive research on a series of vowel shifts, known as the Northern Cities Chain Shift, which is under way in American cities such as Chicago, Detroit and Buffalo. Some shifts in the chain are almost complete and others are in their infancy, but overall the chain forms a complete ‘loop’ in phonetic space. The oldest change in the chain is the raising of [æ] in words

Sound change

67

I

e

V

O

{

A

Figure 26 The Northern Cities Chain Shift

such as hat, pack, last, bath and man. In these words, the vowel is shifting from

[æ] to [eə] or [ɪə] (the raised [ə] indicates a very weak second component to a diphthong). The space left by [æ], a low front vowel, has been filled by a fronting of [ɑ] (in words such as got, not and pop) to [æ]. Similarly, the space vacated by [ɑ], a low back vowel, has been filled by the lowering of [ɔ] to [ɑ] in words such as caught, talk and taught. We thus see a sequence of changes with vowels taking over the ‘space’ vacated by other vowels. Furthermore, something like the converse of what we have just described has also occurred as part of the Northern Cities Chain Shift. Specifically, the change of [æ] to [eə] or [ɪə] produced a

‘congested’ area of mid closed/high front vowels. As a result, these have also begun to move. In particular, [ɪ] (in words such as pip, tin and sit) is moving from

[ɪ] to [e], and [e] (in words such as pet, lend and spell) is moving back to the position of [ʌ]. Finally, [ʌ] – in cup, butter, luck, etc. – is moving slightly further back and rounding, to fill the position vacated earlier in the process by /ɔ/.

From the above description and figure 26, it should be clear that the chain involves a series of changes which constitute a closed ‘loop’ in phonetic space.

Now notice that some of the changes in this chain have been caused by one

vowel moving and pulling other vowels behind it. This is the case with the [æ] –

[ɑ] – [ɔ] chain: [æ] moved first and the others ‘followed’. Such chain shifts are called drag-chains. Sometimes, however, a vowel may move towards the position of another vowel, causing that vowel to move itself. This is the case with the

[ɪ] – [e] – [ʌ] part of the chain: [ɪ] lowered to the position of [e], which backed into the position of [ʌ] which, consequently, had to move back itself. This sort of shift is called a push-chain (exercise 1).

The transition problem: regular sound change

versus lexical diffusion

Having observed a number of different types of sound change, we can

turn to the question of how, more precisely, these changes affect the words in which they occur. Does a sound change affect every word which contains that

sound at the same time, or are some words affected before others? Are vowel

changes phonetically gradual, taking small steps in phonetic space on their route

68

sounds

to the new vowel, or are they abrupt, ‘jumping’ from one vowel to another without going through intermediate phonetic stages?

Two hypotheses have been put forward to account for the way sounds change.

The first was initially proposed in the nineteenth century by the Neogrammarian group of historical linguists and it regards sound change as regular. Two important principles underlie this hypothesis. The first of these is that if a sound change takes place, it will take place in all words with similar environments at the same time.

There will be no exceptions. The outcome of this is that sound changes must be phonetically gradual, but lexically abrupt. A vowel shift, adhering to this

principle, would move through phonetic space towards its new destination in

small steps, rather than in one step, and the change would apply to every word in which that vowel occurred. If, for instance, we take the change from [ɛ] to [e] in the Southern Hemisphere varieties of English of Australia, New Zealand and

South Africa, we would expect to find (a) small phonetic changes to gradually shift [ɛ] to [e]; and (b) every word which contained [ε] to move to [e]. In the case of South African English, this appears to be correct with all words with

[ɛ] passing through a stage where they had a vowel intermediate between [ɛ]

and [e].

The second Neogrammarian principle elaborates on the notion of ‘similar

environment’ which appears in the first principle. Specifically, it states that if a sound change takes place, the only factors that can affect that change in any way are phonetic ones, such as the phonetic characteristics of the segments which surround the feature undergoing change. These changes, then, may be phonetically conditioned: the changing sound in some of the words may shift faster than in others because it is surrounded by a phonetic environment which particularly favours the change. Conversely, in some words the phonetic environment may

hinder and slow down the change. However, according to the Neogrammarians, it is impossible for a sound change to operate, say, in nouns but not in verbs, since this would be an example of a change being subject to non-phonetic conditioning (i.e. grammatical category membership). An example which appears to be consistent with this emphasis on phonetic environment appears in Labov’s studies of the Northern Cities Chain Shift, which we have just described. He found that the change from [æ] to [ɪə] was most favoured when the vowel preceded a nasal

consonant, as in aunt, dance and hand, but hindered when the vowel preceded a velar consonant, such as in black and track.

Despite the predictive success of Neogrammarian principles in some cases, a

number of historical linguists, particularly those working on dialects of Chinese, became unhappy with the hypothesis that sound change always displayed regularity. This was because they discovered examples of changes which did not