<<

A unifying explanation of the Great Shift, Canadian and Southern Monophthonging

Carlos Gussenhoven

1. Introduction

Phonological changes of the Neogrammarian type have a number of aspects. One of these is a dynamic social mechanism guiding the propagation of the new pronunciation trend (Labov 2001). The second is a phonological interpretation of the new sound(s) which differs from that of the earlier generation, the proper, known as reinterpretation or phonologization (Jakobson 1931, Hyman 2013). In the case of English, this core aspect figures prominently in early work by John Harris (1989a,b), which has revealed intriguing patterns showing the division of labour between lexical and postlexical . A third aspect is the beginning of the change, the gradual shift in the phonetic implementation of some sound or set of sounds, possibly restricted to or differentiated by a phonological context. This aspect is particularly intriguing, because of the lack of consensus over whether those beginnings are motivated, and if so, by what.

My interest is in this third aspect. In particular, I will argue that three well-known English sound changes, the GREAT , CANADIAN RAISING and SOUTHERN MONOPHTHONGING, were each motivated by the enhancement of a phonetic duration difference (.g. Stevens & Keyser 1989). Two articulatory policies are identified, Vowel Raising, which was applied in the case of the , and Off-glide Strengthening, which was involved in the enhancement of the laryngeal contrast in coda obstruents. Section 2 deals with the first policy and section 3 with the second. Section 4 offers a conclusion.

2. Vowel Raising and durational enhancement

At some point in his treatment of the GREAT VOWEL SHIFT (GVS) (Jespersen 1949), Lass (1976: 59) observes that explanations of its origin had never been of particular interest to earlier researchers. However, rather than taking up the challenge of identifying the motivation behind this complex , he restricted himself to formulating the question and giving it a name, the ‘inception problem’: Why did it happen in the first place? His concern was mainly with the cohesion of the individual changes as well as the identification of where it started, i.e. what the first phonetic shift was. For our purposes, the inception problem involves two elements:

1. What motivated speakers to begin the shift? 2. Why, given that motivation, is the phonetic nature of the shift a sensible policy?

The first question is addressed by Labov (1994: 118), who answered it in principle, adhering to the belief that sound changes are there for a reason, as seen in this quotation:

Functional analysis of phonological changes must involve the capacity of features and to distinguish words and thereby convey representational information. […T]he functional explanation of sound change will refer here to the effort to account for the rotations, collapse, or expansion of the sound system through correlations with the amount of referential information conveyed by the system.

However, like Lass, Labov steered clear of the second question. His focus, too, is on the cohesion of chain shifts like the GVS, on the way in which the various changes can be explained in terms of a preservation of vowel contrasts. Importantly, however, a more general theme of his chapter relates to the different directions in which long and short move when they shift, and thus by implication to the differentiation between English tense and lax vowels. Principle I, for which he presents fully consistent data from a number of , is that long vowels raise, while Principle II, for which he offers many examples but also some exceptions, says that short vowels lower. Principle I in particular suggests that the GVS might well have been a way of enhancing the contrast between tense and lax vowels. After all, it is unlikely that shifting the phonological identity of the tense vowels around was in itself a reasonable thing to do.

The second of the two questions above should therefore focus on why the initial move in the GVS was helpful in enhancing the quantity contrast for (ME) vowels, and so that brings us back to the issue of the first phonetic adjustment in the web of changes. Panel (a) in Figure 1 gives the separate changes as they took place in the sixteenth-century forerunner of the modern standard sound systems. Table I exemplifies these changes, along with the development of ME [æɪ] of day. As shown, the high vowels of the year 1450 diphthongized, the mid vowels moved up by one position or, in the case of most instances of [ɛː], two positions, while [aː] moved forward and up, merging with the independently monophthongized of [dæɪ] day. The dame-day group joined the cases of [ɛː] that didn’t make the second shift up. The resulting system of 1650 was enlarged with a new [aː], which arose from various lengthening processes affecting short [a]. Ignoring the addition of centring due to [r]-loss, this system has remained intact to the present day, though with adjusted phonetic realizations, as shown for 1950 (Gimson 1962).

Figure 1. Schematic vowel shifts in the English Great Vowel Shift in the south (top) and the north and (bottom).

The inception of the GVS continues to be a well-known controversy between adherents of the ‘pull-chain’ and ‘push-chain’ hypotheses. Advocates of the ‘pull-chain’ view assume that the high vowels diphthongized first, such that the space they vacated was occupied by the mid high vowels (Jespersen 1909). The ‘push chain’ hypothesis holds that a raising of the mid high vowels caused the high vowels to diphthongize in the interest of contrast maintenance (Luick 1898). Lass (1969, 1976: ch 2) convincingly argues that Luick was right, on the basis of the northern version of the GVS. North of a line stretching from Cumbria to Humberside, the starting position differed from that in the south, because [oː] had fronted to [øː], which later developed into other vowel qualities like [ɪə], before the start of the GVS (Wells 1982: 359). If diphthongization were the primary change, so Lass argued, there would be no explanation for the exceptional retention of [uː] in the northern and Scottish varieties. This Middle English vowel in fact still exists there, as in [huːs] house and in all the dialects in which [uː] has remained, the reflex of [oː] as in moon is a vowel other than [uː]. Lass’s point has not been disputed, as far as I know. The ‘push chain’ scenario is also the one that is consistent with Labov’s discussion (1994).

Table I. Exemplification of the southern English GREAT VOWEL SHIFT.

1450 1650 1950 time tiːm tɛ̈ ɪm taɪm meet meːt miːt mɪit beak bɛːk biːk bɪik steak stɛːk steːk steɪk dame daːmə deːm deɪm day dæɪ deː deɪ house huːs həʊs haʊs moon moːn muːn mʊun goat gɔːt goːt gəʊt

The second aspect of the inception question can therefore be reformulated as follows:

3. Why does a raising of the long mid high vowels aid the perceptual salience of the difference between the long and short vowels?

A compellingly reasonable answer to question 3 is that higher vowels sound longer, to speakers of English and perhaps to speakers of all languages. Perception research confirms this conjecture for Dutch. In Gussenhoven (2007), I reported longer perceived durations for [i u] than for [ɛ œ o], respectively, for identical acoustic durations of single-vowel stimuli. So, reformulating the question once more:

4. Why does Vowel Raising increase perceived duration?

It is suggested that the answer lies in a compensatory mechanism that listeners call upon in order to partly cancel out natural side effects of speech production. Two classic cases involve the rate of vocal fold vibration. First, there is the perception of higher pitch in later f0 peaks, a compensation of the tendency for f0 to decrease towards the ends of phrases (Pierrehumbert 1979). That is, phonetically identical up and down movements are heard as higher as they come later in the utterance. Evidently, listeners here subtract a hypothesized effect of declination from the acoustic peak value before deciding on the linguistically relevant percept of pitch height. Second, Silverman (1984) found that listeners hear f0 peaks as more prominent when they occur during a low vowel like [ɑː] than during a high vowel like [iː]. It is reasonable to assume that if Silverman’s task had been to rate the vowels for pitch, [ɑː] would have been judged as higher-pitched than [iː]. The effect is explained by Silverman as a compensation of intrinsic f0. High vowels, particularly [i], naturally have a higher rate of vocal fold vibration, due to the raising of the forward part of the tongue, which will exercise some pull on the hyoid, the horseshoe- shaped bone to which the base of the tongue is attached. In its turn, the hyoid will pull up the thyroid, the large front cartilage of the larynx on the inside of which the vocal folds are attached and to which the hyoid is connected by a ligament. The resulting stretching of the vocal folds will cause some increase in their rate of vibration. But again, our brain takes this fact into account, and will take off some of the f0 before deciding how high the pitch of the high vowel is.

Back to question 4. If raising vowels makes them sound longer, this may again be a compensation effect. Observe that in addition to intrinsic f0, vowels have intrinsic durations. Lower vowels will by their nature be longer than higher vowels, because of the tongue and jaw drop they require for their production. Raising a vowel reduces the drop and thus the time needed to reach the vowel’s target. If the acoustic duration of the vowel is retained, a compensatory effect will result from the fact that our brain will now take off some time from the acoustic duration before deciding how long it is.

5. Is the compensatory effect of Vowel Raising on the perception of duration exploited for other contrasts than vowel quantity?

Independent evidence for the exploitation of the compensatory effect of vowel raising should preferably be found for some other contrast which is heavily dependent on a phonetic duration difference. There are in fact two such cases.

The first is the raising of long vowels as a function of a contrast in , as reported for the tonal dialect of Maasmechelen (Belgium) and reproduced in Table II (from Verstegen 1996). Quite generally in this area, there is a salient durational enhancement of the binary tone contrast, whereby syllable rhymes with ‘Accent 2’ are longer than those with ‘Accent 1’. Table II shows how the phonetically lengthened vowels with Accent 2 have become segmentally distinct from those with Accent 1 in the expected direction, i.e. with closer vowels for Accent 2. Maasmechelen is some 20 kilometers removed from the toneless enclave described in Cajot (2006) in which new vowel contrasts have largely replaced the former tone contrast, and quite possibly, therefore, the tone contrast may be lost in the Maasmechelen minimal pairs in Table II.

Table II. Raised long vowels in syllables with Accent 2 in Maasmechelen.

Accent 1 Accent 2 ɣɛɛl ‘yellow’ ɣeel ‘yellow’ (attributive) (predicative) wɛɛx ‘road’ weex ‘roads’ ɣɔɔn ‘(to) go’ ɣoon ‘(I) go’ nɔɔl ‘needle’ nool ‘needles’

For the second case, we return to English, where the contrast between fortis and lenis codas is, phonetically, served largely by a durational difference between the sonorant portions of the syllable rhyme (cf. [iː] in neat - knead, [aɪn] in pint - pined, etc.). We may expect that this enhancement of the laryngeal contrast by duration is aided by a vowel quality difference, such that (tense as well as lax) vowels before lenis obstruents are higher than their counterparts before fortis ones. Gussenhoven (2007) shows how English [iː ɪ uː ʊ] have a higher tongue body before lenis codas than before fortis ones in a speaker of British English, while for a speaker of , all differences are in the same direction, with just the data for [iː] failing to show statistical significance. In earlier literature, further evidence can be found for this behaviour. The effect is small, however, and no perception research has been undertaken to show that it helps the hearer recognize the laryngeal categories.

Because of the important role of the difference in duration between rimes before fortis codas and other rimes in English, the question arises why vowel raising isn’t exploited more vigorously to enhance that difference. The answer may lie in the size of the set of vowels whose duration might need enhancement. In the case of the GVS, the concern was the set of tense monophthongs, which moved up, squeezing out the high vowels. By contrast, the phonetic duration difference for the fortis-lenis contrast applies to the complete set of rhymes, diphthongs as well as long and short monophthongs. In the next section, we will see that this second duration-loaded contrast is aided by a different ploy, one that actively seeks to exploit a potential side effect of a diphthongal pronunciation.

3. Off-glide Strengthening as an enhancement of duration differences

Shorter durations of vowels before voiceless than before voiced consonants are found in almost all languages and are widely recognized as providing a perceptual aid to the difference between the long articulatory constrictions of voiceless consonants and the shorter ones of voiced ones (Kluender, Diehl & Wright 1988). Preglottalization, which accompanies English pre-fortis in the case of fortis plosives (Wells 1982: 260), has accordingly been claimed as an enhancement of the general shortening of sonorant rimes before fortis obstruents. An additional ploy applied in English is Off-glide Strengthening. When instead of [aɪs], a rhyme like ice is pronounced as [ajs], an impression of a shortened vowel is created by the suggestion that the second element of the diphthong is a . In Gussenhoven (2007), I could only show a significant effect for the comparison of back rounded [ou] vs [ow], but the front unrounded and front rounded results were in the same direction.

The functional element in CANADIAN RAISING, typically described as the central realization of the first element of the diphthongs [aɪ aʊ] before voiceless coda consonants, as in rice and house, is therefore not so much the raised beginnings of these diphthongs, but their high endings. The raised first elements are side effects of the raising of the second elements. These high endings must also occur for [eɪ] and [oʊ], as in race and roast, and possibly for the slightly diphthongal [iː uː], too. Evidence for this is presented in Moreton (2011). Because this policy is inapplicable in the case of short vowels, which do not offer the same scope for this manipulation, short vowels are likely to be affected more by the vowel raising policy discussed in section 2.

My explanation suggests that the counterpart of CANADIAN RAISING should be the lowering of the second element. In an extreme form, this behavior is found in the South of the USA, where SOUTHERN MONOPHTHONGING has been widely reported. The original context would appear to have been the absence of a fortis coda, but the change was apparently generalized to pre-fortis contexts in working class speech. Wells (1982: 537), citing McDavid (1967), writes about the allophones of [aɪ] that ‘[…] although [a(ː)] was used by all classes finally or before a voiced consonant, before voiceless consonants there was a social distinction in that educated speakers had a diphthong but many uneducated speakers a monophthong.’ Assuming the usual change from below for Neogrammarian sound changes, the generalization to fortis contexts is an extension of the more restrictive non-fortis context which hadn’t yet reached beyond the working class. SOUTHERN MONOPHTHONGING signals the opposite of Off-Glide Strengthening, an exaggerated absence of the shortening cue. This explanation may also hold the key to yet another phonetic change, which is applied to lax vowels in the same dialect area, the opening up their final portion. This may be interpreted as a safe way of enhancing their duration. The Pronunciations like [bɪɛ̈ d for bid, [beəd] for bed, [gɤəd] for good, etc. would in that view count as hyper-monophthongs, with tongue glides in the opposite direction from what would be required to make them sound short. However, they take their place among a number of ‘breaking’ processes, and it would be hazardous at this point to attribute this effect to the enhancement of the duration difference between vowels before fortis and non-fortis codas.

Conclusion

The inception of regular sound changes is frequently motivated by the desire to increase the distinctiveness of phonological contrasts. This contribution has argued that this is true for three well-known sound changes which took place in English, the GREAT VOWEL SHIFT, CANADIAN RAISING and SOUTHERN MONOPHTHONGING. I have argued that in all three cases the motivation was an increased salience of a phonetic duration difference, as used to express a vowel quantity difference in the case of the GREAT VOWEL SHIFT and a laryngeal coda contrast in the case of CANADIAN RAISING and SOUTHERN MONOPHTHONGING.

Earlier explanations of the GREAT VOWEL SHIFT include that by Samuels (1972), who assumed that vowels naturally acquire higher targets due to their being stressed, and that to prevent a build-up of vowel targets in the higher zone of the vowel area, the highest vowels diphthongize. He does not however offer an explanation of the presumed effect of stress, neither does he provide a demonstration that there in fact is such an effect. Samuels (1972: 144ff) discusses and convincingly dismisses two further attempts. One is an explanation based on a disassociation of the lax high vowels from the tense high vowels and their reassociation with mid tense [eː oː]. The other is one which assumes a neutralization of the tense-lax contrast, an explanation that can be dismissed on the basis of any evidence for the neutralization. More recently, an explanation of CANADIAN RAISING was offered by Moreton (2004). He interpreted the high off-glides of pre-fortis diphthongs as a hyperarticulation triggered by the fortis nature of those obstruents. Observe that his theory does not explain why short vowels are lower before fortis consonants, unless some independent reason is found for why hyperarticulated monophthongs are lower, and that it predicts that vowels are generally more peripheral before voiceless plosives than before voiced ones, while no such data have been presented.1 A fuller evaluation of Moreton’s proposal is given in Gussenhoven (2007).

The three sound changes discussed here provide evidence for a functional view of their origin. Specifically, their motivation was found in aiding a durational enhancement of phonological contrasts whose phonetic salience needed to be boosted.

Acknowledgement I would like to thank two anonymous reviewers for their helpful comments.

References Cajot, José (2006). Phonologisch bedingter Polytonieverlust; Eine tonlose Enklave südlich von Maastricht. In Michiel de Vaan (ed.), Germanic Tone Accents. Proceedings of the First International Workshop on Franconian Tone Accents. Stuttgart: Franz Steiner.11-23.

Gimson, A.C. (1962). An introduction to the pronunciation of English. London: Edward Arnold.

Gussenhoven, Carlos (2007). A vowel height split explained: Compensatory listening and speaker control. In Jennifer Cole & José I. Hualde (Eds.), Laboratory Phonology 9. 145- 172. Berlin/New York: Mouton de Gruyter.

Harris, John (1989). Towards a lexical analysis of sound change in progress. Journal of 25: 35-56.

Harris, John (1989) Derived phonological contrasts. In Susan Ramsaran (ed.), Studies in the pronunciation of English: a commemorative volume in honour of A.C. Gimson. 87- 105. London: Routledge.

Hyman, Larry (2013). Enlarging the scope of phonologization. In Alan C.L. Yu (ed.) Origins of sound change: Approaches to phonologization. Oxford: Oxford University Press. 3-28.

Jakobson, Roman (1931). Principles of historical phonology. In Allan R. Keiler (ed. 1972), A reader in historical and comparative linguistics. New York: Holt, Rinehart & Winston.121-138.

Kluender, Keith R., Diehl, Randy S., Wright, Beverly R. (1988). Vowel-length differences before voiced and voiceless consonants: An auditory explanation. Journal of 16: 153-169.

Labov, William (1994). Principles of Linguistic Change: Volume 1: Internal Factors. Cambridge MA: Blackwell.

Labov, William (2001). Principles of Linguistic Change: Volume 2: Social Factors. Malden MA: Blackwell.

Lass, Roger (1996). and Phonological Theory. Cambridge: Cambridge University Press.

Moreton, Elliot (2004). Realization of the English postvocalic [voice] contrast in F1 and F2. Journal of Phonetics 32: 1-33

Samuels, M.L. (1972). Linguistic Evolution, with Special Reference to English. Cambridge: Cambridge University Press.

Stevens, Kenneth N., Keyser, Samuel J. Keyser (1989). Primary features and their enhancement in consonants. 65: 81-106.

Trigo, Loren (1991). On pharynx-larynx interactions. Phonology 8: 113-136.

Vaux, Bert (1996). The status of ATR in feature geometry. Linguistic Inquiry 27: 175- 182.

Verstegen, V. (1996). Bijdrage tot de tonologie van Oostlimburgse dialecten. In H. van de Wijngaard (ed.), Een eeuw Limburgse dialectologie. 229-234. Hasselt/Maastricht: VLDN/Vereniging Veldeke Limburg.

Wells, John C. (1982) Accents of English. Cambridge: Cambridge University Press.

1 A reviewer points out that after voiceless plosives vowels have been found to be lowered and back vowels to be more back than after voiced plosives (Trigo 1991, Vaux 1996).