|
Article Excerpt
Abstract. This paper seeks to further establish the connection between phonological vowel harmony and phonetic vowel-to-vowel coarticulation by examining the role of stress in each. A surprising number of languages have vowel harmonies in which unstressed vowels assimilate to stressed vowels. It is argued that these vowel harmonies are the result of the phonologization of a stress asymmetry present in the vowel-to-vowel coarticulation of precursor languages. Experimental results indicate that unstressed vowels have a tendency to undergo greater vowel-to-vowel influences than their stressed counterparts. The results of a perception test confirm that the effects observed on the unstressed English vowel /i/ are great enough to be perceived, and a model of how a language would develop a stress-dependent harmony from a stress asymmetry in vowel-to-vowel coarticulation is put forward.
I. INTRODUCTION. Many of the world's languages have a phonological process known as vowel harmony, which in its most general form, involves a long distant process of assimilation where certain features of the vowels within a word are restricted to one or another set of values for a given phonological feature. Languages differ greatly in which phonological feature is restricted and in precisely how that feature value is determined. Very often, vowel harmony proceeds from word roots and spreads onto affix vowels so that the vowel of a given affix is determined by the quality of the root vowels. A surprisingly large number of languages contain harmonies that are triggered by a stressed vowel and manifested on unstressed vowels. Such STRESS-DEPENDENT HARMONIES (SDH) can be found in languages representing a wide range of language families as well as differing phonological feature restrictions. To demonstrate the range of these types of harmonies, a few examples are given below. Languages have been chosen to represent VII with varying phonological feature restrictions.
1.1 Pasiego Montanes Height Harmony. The first example of a SDH comes from the well-known height harmony of the Pasiego Montanes dialect of Spanish spoken in northwestern Spain (Penny 1969a, 1969b). In this harmony, non-low, unstressed vowels agree in height with the stressed vowel of the word. Data illustrating this harmony is shown in (1).
(1) /beb/ 'drink' bi' b-i:s 'you (pl.) drink' be'b-er 'to drink' be'b-amus 'we drink (sub.)'
These forms contain variants of the morpheme /beb/ 'drink'. The stressed vowels of each of these words appear on a suffixal vowel and alternations appear on the root. If the stressed suffixal vowel is high, as in bi' b-i:s, the root vowel is also high but if the stressed vowel is mid as in be'b-er, the root vowel is also mid. The third example contains a stressed vowel that is low and in this case, the harmony does not apply so the underlying form emerges (Penny 1969a).
1.2. Eastern Cheremis Palatal Harmony. Eastern Cheremis, spoken in eastern Russia, contains a stress-dependent harmony in which word final mid vowels agree in the features [round] and [back] with the stressed vowel of the word. The alternations in (2), taken from Sebeok and Ingemann (1961), illustrate this phenomenon.
(2) -[??]e ~ -[??]o ~ [??]o 'third singular possessive' (a) 'kit-[??]e 'his hand' (b) '[??]p-[??]o 'his hair' (c) 'surt-[??]o 'his house'
These examples all contain a form of the third singular possessive suffix, whose mid vowel alternates according to the frontness and rounding of the stressed vowel. The form in(a) contains only front, unround vowels, (b) contains only front, round vowels, and(c) contains back, round vowels. This harmony is not limited to suffixal vowels as monomorphemic words abide by the same harmonic restriction. One interesting characteristic of the Eastern Cheremis vowel harmony is that the extremely reduced vowel [[??]] appearing in word medial position is reportedly unaffected but also fails to impede the harmony triggered by the stressed vowel and observed on the word final vowel. (3) illustrates monomorphemic roots in which the word-final mid vowel agrees with the stressed vowel while (4) shows root forms in which the harmony is present on the word-final mid vowel despite the intervening [[??]] vowels.
(3) 'bate 'wife' 'imne 'horse' 'jolko 'lazy' 'j[??]s[??] 'difficult' (4) 'k[??]reltemo 'unceasing' 'puld[??]r[??]o 'quail' (3) 'erge 'boy' 'jumo 'god' (4) 'teg[??]de 'little'
(5) shows that the harmony is limited to word-final vowels while unstressed medial vowels tend to reduce to [[??]].
(5) 'tu[??]-[??][??]to 'at the bottom' 'kid-[??]st[??]-[??]e (5) 'tu[??]-[??][??]to 'in his hand'
The last vowel of the locative morpheme is reduced when word-medial (-[??]st[??]), but Obeys the harmony when in word-final position as in 'tu[??]-[??][??]to.
1.3. Koya Total Harmony. Tyler (1969) reports a vowel harmony present in Koya, a Dravidian language spoken in the Andhra Pradesh region of India, which can be interpreted as SDH. Stress in Koya is quantity sensitive, so all long vowels are stressed. As shown in (6), in unstressed syllables, the short vowels /i/, /a/, and /u/ harmonize with respect to all vowel features with a following long, and therefore stressed, vowel.
(6) (a) iru--'vu:ru 'both' (b) ira--'va:y 'yae:ydu '25' (c) vani 'k-i:t-e 'it will bend' (d) vanak 'to tremble' (e) migi 'l-i:t-e 'it will remain over' (t) migal 'remain over'
Forms in (a) and (b) show an alternation in the morpheme meaning 'numeral' [iru]![ira] in which the last vowel of the morpheme is an identical copy Of the following stressed vowel. The forms in (c)-(d) and (e)-(f) show alternations in root morphemes in which the final vowel surfaces as low when no stress follows as in (d) and (f), but surfaces as a copy of the following stressed high vowel in (c) and (e).
Each of the languages above varies in the feature involved in the VII, but they share one basic fact: the harmony appears to be triggered by stressed vowels and manifested on unstressed ones. The development of SDH in these languages is most likely not due to contact or inheritance since they hail from vastly different language families and geographical regions. Instead, we can safely assume that SDH developed independently in these languages and is a result of natural speech processes becoming crystallized in these very different phonologies.
Vowel harmony is assumed by many phonologists to arise from a phonetic assimilation known as vowel-to-vowel (V-V) coarticulation. Simultaneous examination of the role of stress in vowel harmony and V-V coarticulation will help establish this connection more firmly. Therefore, having reviewed a few of the languages that have VH triggered by stressed vowels, I will now turn to V-V coarticulation and the role that stress plays in the phonetic domain. I will then examine perception of V-V coarticulation and propose a model of how stress asymmetries in V-V become phonologized in SDH systems.
2. Phonetic Origins of SDH: V-V Coarticulation. V-V coarticulation differs from vowel harmony mainly in the extent of the triggering vowel's influence observable on the target vowel. V-V coarticulation tends to be more gradient in nature; the vowel triggering the coarticulation tends to exert its influence only on a small portion of the affected vowel, whose identity is not compromised when listeners attempt to recover the phonemic information from the speech stream. Because V-V coarticulation is a phenomenon characteristic of connected speech, it can be affected by extra-linguistic factors such as speaker idiosyncrasies (some speakers are prone to more careful and therefore use less coarticulated speech, than others), speech rate, and social context.
Ohman (1966) first reported the discovery of V-V coarticulation in VCV sequences when he found that in English and Swedish, the acoustic transitions from vowel-to-consonant and consonant-to-vowel are significantly influenced by the quality...
|
