Tonal features and the development of vietnamese tones 越南话的声调的发展 sự phát triển thanh ̣̣điệu tiếng việt

TONAL FEATURES AND THE DEVELOPMENT OF VIETNAMESE TONES MARK ALVES In describing the development of tones and register in Southeast and East Asia, several phonetic features need to be

From: (1995) Working Papers in Linguistics: Department of University of Hawaii at Manoa, Vol 27: 1-13

TONAL FEATURES AND THE DEVELOPMENT

OF VIETNAMESE TONES

MARK ALVES

In describing the development of tones and register in Southeast and East Asia, several

phonetic features need to be considered: various dimensions of pitch (e.g height and

con-tour), phonation type features (e.g breathiness and creakiness), and length (i.e intrinsic tonal

duration) Vietnamese as seen as a link between two linguistic areas with different types of

suprasegmental features, namely Chinese tones versus Mon-Khmer register shows complex

bundles of features in its tonal system, features that interacted to generate the conditions for

tonal contour, if not tonogenesis This study will add to Haudricourt’s proposal (1954) for

Vietnamese tonogenesis by looking at how feature enhancement played a role in Vietnamese

tonogenesis This study also modifies Haudricourt’s proposed order of tonal development and

shows how—historically, genetically, and phonetically—another order of development

obtains involving three stages of tonal development followed by a tonal split Another aspect

of this study is to formalize the phonological relationship among tones, phonation features,

tonal length, and the tone-bearing units

1 INTRODUCTION This study attempts to support the following points:

(a) Tones, in this case Vietnamese tones, consist of a number of phonologically significant phonetic features In addition to tonal contour and height, which are considered the more salient features of tone (and thus the more noted in the linguistic literature on tone), varieties of tonal phonation and length are important in understanding both the synchronic representation and diachronic development of tone

(b) Vietnamese tones could only have arisen through the interaction of a number of glottal and pharyngeal phonetic phonation features (as opposed to simply ‘borrowing’ tones through language contact), which, over time, ‘enhanced’ each other to the point of creating phonemically lexical pitch which still contained the phonation features

(c) There has to be a phonetically logical sequence of tonogenesis and tone splits For Vietnamese, a Mon-Khmer language that had creaky vowels before it had tones, it makes sense both phonetically and historically to posit that tonogenesis occurred in more than one stage, in which both creaky vowels and final stops independently developed one tone category before syllables with final fricatives developed another category Some evidence from Ruc, a Vietic language, will be provided

All of these claims are discussed with some focus on the relation between tonogenesis and length as well as the concept of feature enhancement and the features involved in the tonogenesis process Thus, this study is a preliminary attempt to formalize the features associated with Vietnamese tones Since these features are endemic to the region of East and Southeast Asia, they could be used to describe the tonal systems of other languages as well

2 TONAL AND REGISTRAL FEATURES Tones are typically considered to be suprasegmental phonemes based on pitch height and contour This can be seen in works such

as Pike’s (1948) analysis of tone, in which he discussed tone in terms of pitch, height (what

he called ‘register’), and combinations of them Y.R Chao’s five-number system to represent tonal pitch height (5 as the highest and 1 as the lowest) is commonly employed by

linguists in discussing tonal languages and in dealing with phonetic description of tones Wang (1967) analyzed Chinese tones in terms of seven features that all pertained to pitch

height and contour Those features include the presence or absence of contour; the three heights high, central, mid; and the three contour types rising, falling, and convex In terms of

tone height and contour, there are other factors that can be included, such as (1) average pitch, (2) direction, (3) length, (4) extreme endpoint, and (5) slope (Gandour 1983)

Clearly, the term ‘tone’ does have to do with pitch contour and height, but in a discussion

of the phonetic and phonemic features of ‘tones’, then other nonpitch features are involved, either in allophony or even in lexically distinctive features Consider some of the relevant features seen in acoustic phonetic studies of tone

The physical concomitants of prosody include fundamental frequency,

duration, intensity, harmonics and spectral energy distribution from which

linguistic categories, such as tone, stress, vowel length, phrasing and

(linguistic) intonation, as well as expressive categories, such as (affective)

intonation, timbre, tempo and melody, are constructed (Ross et al 1986)

Hereafter, the term ‘tone’ refers to a suprasegmental lexically distinctive complex of pitch height and pitch contour, but also length and a variety of emic and etic manifestations of phonation Listening to most any tonal language, one can easily perceive other sorts of phonetic features (besides height and contour) of a language’s tonemes that help distinguish the tones The Mandarin third/dipping tone is the classic example of a tone with phonation, a glottal creakiness that, at least in isolation (excluding tonal sandhi), distinguishes it from other tones Wu dialects have tones which are sometimes described as ‘breathy’ (Ramsey 1987), another kind of glottal phonation Thai is said to have tones that differ primarily in phonation rather than contour (Uri Tadmor, personal communication)

These tonal phonation features, such as glottalization and breathiness, are some of the same features that are common in the vocalic register systems of Mon-Khmer languages The division between the first and second register (usually considered to be conditioned by the voicing of the initial consonant either synchronically or diachronically) is seen in a variety of segmental phonetic effects, including the same sorts of features which are included among the suprasegmental tone features in tonal languages Consider table 1, taken from Matisoff 1973

Tense-larynx Lax-larynx

higher pitch/rising contour lower pitch/falling contour association with - / association with -h

retracted tongue-root advanced tongue-root

‘creaky’ laryngeal turbulence ‘rasping’ laryngeal turbulence larynx tense and/or raised =

reduced supraglottal cavity

larynx lax and/or lowered = distended supraglottal cavity

T ABLE 1: Laryngeal features

The kinds of description for these include seemingly impressionistic terms such as

‘breathiness’, ‘creaky’, and ‘raspy’ Other seemingly more objective descriptions include pitch and contour, tongue-root position, larynx tenseness, and voicing Similar features are

seen in table 2, taken from Gregerson 1976 In addition, there are features that include vowel qualities such as open/closed and ongliding or diphthongs

Register Initial Consonant Voice Quality Vowel Quality Pitch

first (original) surds normal head, clear,

tense

more open, onglided

relatively higher pitch

second (original) sonants deep, breathy,

sepulchral, relaxed

close, centering diphthongs

relatively lower (larynx also lowered)

T ABLE 2: Registral features

What needs to be acknowledged here is that, despite some of the so-called

‘impressionistic’ descriptions, these phonation features need to be taken quite seriously in terms of their use in phonological systems and their effects on the development of the phonology of languages with register Though they are not as easy to define as point or manner of articulation, these kinds of glottal and laryngeal sounds need to be regarded as phonemically significant So ‘breathy’, for example, is not just a description, but is also a distinctive feature An attempt at formalizing these features is seen in the study by Headley (1976) that showed how pharyngeal features could be used to analyze Sre vowels and other Mon-Khmer and West African languages with pharyngeal expansion

Formalized phonation features can thus be used in a phonological analysis of register languages that show numerous significant phonation features Moreover, the features in question can be shown to have interacted diachronically, resulting in various phonological consequences The transference of register and tone from one language to another through language contact can also be better understood when the phonological and phonetic features are clearly distinguished Section 5 provides a phonological analysis of these features Before that, sections 3 and 4 will further discuss register in Mon-Khmer languages and Vietnamese tonogenesis

3 PHONOLOGICAL EFFECTS OF PHONATION AND REGISTER There are numerous phonetic and phonological consequences of register that can be seen both synchronically and diachronically Though these effects vary according to language, there is a general tendency

of change Table 3 (Huffman 1985) shows five stages of registral effect in Mon-Khmer languages, with a sample language for each category Some languages have not changed or show intermediate stages of change to a register system Others have developed full register systems Still others have undergone changes of the vowel system due to the complete or partial loss of register Finally, at the end of the continuum, tones may develop

Voicing Proto-lg Conservative Transitional Register Restructured Tonal

example Stieng Souei Mongkong Khmer Vietnamese

T ABLE 3: Stages of registral effects

The particularly direction of the splits (such as first register showing clear voice in one language and creaky in another) in each case is not as important as the fact that the splits

occur at all.1 The Austronesian Chamic languages, in contact with Mon-Khmer languages in Vietnam, have shown the same kinds of changes (Thurgood 1996), suggesting that these sorts of changes are naturally motivated and, thus, can be phonologically explained Regardless of the theoretical view, register and related phonation clearly have had a tremendous impact on both the vocalic and tonal systems of the languages in question

4 TONOGENESIS AND REGISTER Haudricourt (1954) was able to break down the barrier between Vietnamese and Mon-Khmer languages (as opposed to genetic affiliation with Tai, Maspero 1912) by positing tonogenesis in Vietnamese He saw numerous correspondences between the tones of Vietnamese and certain sounds in Mon-Khmer languages These correspondences have been fairly well supported by later studies (cf Gregerson and Thomas

1974, Huffman 1977) Haudricourt’s hypothesis, whereby pitch contour was derived from the final of the syllable and the pitch height then split into high and low based on the voicing

of the initial, is seen in table 4 (Note: The numbers 1 to 6 as seen in table 4 will be used hereafter when referring to Vietnamese tones, 1 to 3 being the high series and 4 to 6 the low series.)

final initial

zero/nasal stop fricative

T ABLE 4: Haudricourt’s tonogenesis hypothesis

Haudricourt posited that Vietnamese tonogenesis began with the generation of two nonlevel tones which resulted in three tone categories Then there was a tonal height split, creating six tones Eventually, the nonphonemic suprasegmental features became phonemic tonal distinctions as other sound changes occurred, such as loss of finals and changes in voicing of the initials However, Haudricourt did not, at that time, have much access to information about the possible effects of phonation on Vietnamese tones and registrogenesis

Diffloth (1989) somewhat tentatively but persuasively showed that he had evidence to place vocalic creaky voice at the level of proto-Mon-Khmer or perhaps Proto-Austroasiatic This dating is important, since the presence of phonetically and/or phonemically significant phonation in Vietnamese at the time of contact with Chinese could have had significant consequences on Vietnamese tonogenesis Assuming that Haudricourt was correct (that Vietnamese is a Mon-Khmer language), and that Diffloth was correct (that Proto-Mon-Khmer had creaky voice which shows reflexes in Vietnamese tones), then these phonation features were most likely present in Vietnamese when the great wave of tonogenesis surged throughout East and Southeast Asia Phonation seen in Vietnamese tones should be originally from phonation features present more than two thousand years ago, as they are not likely to have come from contact with Chinese

Phonation has had an even more significant effect on Vietnamese than previously imagined As noted by Gage (1985) and Diffloth (1989), a supposed problem in

1 Marc Miyake pointed out to me that Khmer itself shows what could be considered reversed features in comparison with other Mon-Khmer languages He also suggested prosodic factors in terms of syllabic stress as being another factor besides initial voicing David Thomas (1995 SEALANG e-mail discussion on tone and register) summarized various registral effects, and in some cases, these effects were the opposite of what the are

in Vietnamese Nonetheless, the point to be made here is that there is a difference, typically two-way splits

Haudricourt’s hypothesis of Vietnamese tonogenesis was, in fact, explained in the light of reflexes of Mon-Khmer languages According to Haudricourt (1954), final stops resulted in tones 2 and 5, while open and nasal-final syllables showed tones 1 and 4 There are quite a number of counter-examples where tones 2 and 5 occurred with open or nasal-final syllables Diffloth (ibid.) showed that in a number of Austroasiatic languages (including Talan, Chong, Song, and Muong), such syllables occurred with creaky voice, which is reconstructible to Proto-Austroasiatic Examples are shown in table 5

T ABLE 5: Correspondences

This early form of phonation in Austroasiatic provided the phonetic conditions later for tones

2 and 5 to develop, separate from conditioning by the final In terms of the final, those tones appear to be in the category for tones 1 and 4 (i.e occurring in open or nasal-final syllables), but phonetically and phonemically, they appear as tones 2 and 5 (i.e syllables with final stops) This modification of the tone category in terms of syllable form (splitting syllables with clear versus creaky vowels) is shown in table 6

-voice 1 (a) 2a (á) 2b (á) 3 (ä)

+voice 4 (à) 5a (Õ) 5b (Õ) 6 (ã)

T ABLE 6: Diffloth’s modification

The evidence for creaky voice in proto-Austroasiatic makes a significant contribution to the understanding of phonation features in Vietnamese: it appears that (1) creaky voice led to the creation of a tonal category, while (2) the register distinction, related to the F0 and breathiness, led to the tonal height distinction.2

Thus, the earlier Moh-Khmer creaky vowel quality took part in Vietnamese tonogenesis Vietnamese is not the only language to show this Sagart (1988) showed that in Chinese, glottalization and breathiness had a role in the development of tones, in particular, the Chinese departing tone category That phonation took part in Vietnamese tonogenesis is seen

in the tones themselves The phonation features that helped create Vietnamese tones in the first place have not been lost, though they have shifted from the vocalic tier to the tonal tier

to create tones with a complex of features, as will be discussed next The register distinction that led to pitch distinctions as features of certain vowels (segmental) in many Mon-Khmer languages led to tonal (suprasegmental) height distinctions in Vietnamese and other Viet-Muong languages

2 Discussions with Marc Miyake helped clarify this distinction between early Austroasiatic vocalic creakiness and the later development of register that may be a southeast Asian areal feature rather than features from the Austroasiatic proto-language

5 VIETNAMESE TONES AND AUSTROASIATIC FEATURE COMPLEXES In considering Vietnamese tones, there seems to be what can be called a ‘feature complex’ (LaVaughn Hayes September 1995, SEALANG e-mail discussion on tone and register) Vietnamese tones have registral phonation features like those in Mon-Khmer and the pitch contour features like those of Chinese and Tai In addition to these features, I posit that length, like that seen in vowels, needs to be considered an inherent feature of Vietnamese tones

Consider table 7 (from Thompson 1965) The final column ‘short’ is my own addition to indicate tonal length Though it is largely my own impressionistic perception of the difference of tonal length, there is evidence from an acoustic study of Northern Vietnamese tones (Han 1966) that a real difference in length is present in tone 5 A plus sign indicates an intrinsically short tone (tones 2 and 5), a minus indicates a long tone (tones 3 and 6), and both signs mean that the tone has no inherent length (tones 1 and 4) Though it could be a low-level phonetic factor conditioning the length (e.g shortness due to the presence of final stops), there are historical reasons for the difference In the next section, I will discuss how the concept of intrinsic tonal length enters into the development of Vietnamese tones as well

as the modern phonology of Vietnamese

Tone high tense glottalic short

T ABLE 7: Vietnamese tone features

In table 7, height is the primary distinction (though tones 3 and 6 have switched height in modern Vietnamese) Tenseness is not clearly defined by Thompson, but it may have had to

do with the tenseness of the glottis The feature glottalic indicates the presence of a glottal break in the tone itself, a significant phonetic feature Thompson omits the actual tone contour from this particular description, yet the tones are still distinguished by other features

I posit that tone contour in Vietnamese may be somewhat overrated A phonetic study (Earle 1975) showed that the so-called ‘level’ tone (‘level’ according to 13 analysts) is never pronounced level The so-called ‘low-falling’ tone was closest among all of the tones to being ‘level’, but only from onset to midset In any case, there is a sufficient number of other tone features, besides contour, to distinguish them It would be an interesting experiment to see, when tone contour is removed from pronunciation, whether the tones could still be differentiated by native speakers

There are some differences between varieties of Vietnamese The dialects of Vietnamese have already been divided into south and north for centuries Two tones merged in southern Vietnamese, while northern Vietnamese has preserved six distinct categories There does appear to be a difference in length Table 8 gives an impressionistic view of the length of tones (and phonation features) on different syllables

open4 , breathy , breathy

open2 long short

open5 short, glottal short, breathy/glottal

open3 long (or short, glottal) long

checked5 short, glottal short, breathy/glottal

T ABLE 8: Northern versus Southern Vietnamese tones

Northern and southern varieties of Vietnamese do differ, mainly in the type of phonation Northern Vietnamese tones tend to have glottal breaks A particularly notable tone in northern Vietnamese is tone 6 with a glottal break in the middle of the contour In south Vietnam, the most typical tonal phonation feature is breathiness In northern Vietnamese, tone 2 is partially unspecified, since it is longer on open syllables, even to the point of making it similar to tones 3/6 in southern Vietnamese

6 LENGTH IN VIETNAMESE TONES Moira Yip (1989:151) made the claim that ‘ contour tones must involve association to one TBU, rather than to two vowels or moras’ What she saw in Chinese was that a tone occurred with one syllable, and that the whole syllable, not just a vowel or a mora, is the tonal-bearing unit upon which the tone is dependent This makes phonetic sense; tones are phonetically dependent on other sonorant segments for their realization Phonemically, the TBUs need not be separated into sections (i.e vowels or moras) for realizing contour tones However, this model does not allow for any interaction between tone and the TBU, placing tones at the complete mercy of the TBU

in terms of length and toneme phonation features This view of the tone’s dependency on the TBU does not explain how Vietnamese tones influence their TBU or even suggest that as a possibility This influence of tones on the TBU has to do with length

Han (1966) showed that tone 5 was pronounced 50% shorter than any other Vietnamese tone Though in northern speech, tone 2 is longer with open syllables that it is with checked syllables, in the south, it is pronounced short in either closed or open syllables Minimal sets can show how the length of the syllable and the tone it carries interact The vowel /Q/ in Vietnamese is considered to be /-short/ and has two moras, but can be shortened by tones 2 and 5, reduced to one mora Figure 1 illustrates the distinction between vocalic moras and tonal ‘moras’ on an open syllable with a nonshort vowel

F IGURE 1: Samples of Vietnamese Tonal Length

What has happened is that the length of tones 1 and 4 is controlled by the TBU; thus, on long syllables, the tones are long, while on short vowels, they are short Tones 2 and 5 shorten the length of the TBU, and 3 and 6 lengthen the TBU, regardless of the lexical vowel length In general, most of the interaction between tone and TBU results in shortening since

9 of the 11 vowels of Vietnamese are nonshort (i.e containing two moras), though the aspect

of phonation is of course another salient effect of the tone on the TBU

The tonal length just discussed is not an accident, but must have been conditioned by the environment in which tonogenesis took place

mora: µµ µµ µ µ µµ µ µ µ µ µ µ µ

| | \/ | | |/ |\ | | | |\ | |

forms: par → ba1 reh → re3 cç/ → cç2 po-n → bon5 or bon5

F IGURE 2: Historical derivation of Vietnamese tones

Figure 2, which contains boxes 1 through 4, shows hypothetical earlier Mon-Khmer forms based on modern conservative Mon-Khmer languages, the realization of those forms in modern Vietnamese, and the relation between the moras of the syllables to the ‘moras’ (to parallel with the concept of vowel length) of the tone

Tones 1 and 4, the more level tones, are simply the remnants of a register distinction having to do with the initial, and are thus neutral in respect to their TBU That scenario is shown in box 1 Tones 3 and 6 are long because of a preservation of the original length of final fricative syllables Box 2 shows the original form, the loss of the final fricative, and the long tone resulting from compensatory lengthening There may have been tones with the final fricatives, but once those finals were lost, the length of those tones became phonemic Tones 2 and 5, the short tones, had developed as a result of final stops Thus, due to those tones’ phonetic environment, they could only be short, as vowels followed by final voiceless stops tend to be phonetically short and because the final does not carry moraic weight Their shortness is maintained even in open syllables, seen in box 3 In box 4, we can see the effect

of the creaky voice as posited by Diffloth (1989) Southern Vietnamese shows a short tone, and in northern Vietnamese, tone is nonshort Again, due to the loss of certain finals or the creaky voice, the length of those tones became distinctive

7 FEATURE ENHANCEMENT None of the features mentioned above could have singly developed tones in Vietnamese, but rather they must have developed through natural processes of phonetic ‘enhancement’ The concept of enhancement (San 1991) is that features that tend to enhance each other, thereby creating other types of phonetic realizations with phonological consequences, tend to co-occur Combinations of features (including phonation, glottal, and pharyngeal features) have resulted in new features in other Austroasiatic languages In Chong (Huffman 1985, Diffloth 1989) a combination of glottalized/nonglottalized and creaky/breathy vowels gave a four-way distinction These features’ interaction has led to phonation shifts, pitch contours, and vowel breaking (Diffloth 1989:145), quite similar to the phonetic features of Vietnamese tones

Language contact has been considered the single source of tone in Vietnamese, such as contact with Thai and/or Chinese Clearly, language contact with Chinese had something to

do with the development of Vietnamese tones, as the tonal system of Vietnamese corresponds quite directly to the eight-way system of Middle Chinese, but Chinese was not likely the sole trigger for tonogenesis Registral influence on Vietnamese tones is implied by the presence of the phonation features of Vietnamese tones that could not have come from Chinese Vietnamese had register at the time of contact with the Chinese, and potentially had the similar kinds of contours or vowel breaking seen in Chong It may be just as

presumptuous to presume that Vietnamese did not have tones before contact with the Chinese

as it is to say there were already pre-existing incipient tones due to the interaction of phonation features at the time of contact

Phonetically, the F0 distinction between vowels after voiced and unvoiced initials is minimal Voiced consonants were shown to be lower than voiceless ones, a difference which could have caused the difference in the following tone or vowel quality (Hombert 1978), though not with great phonetic consequences A spectrographic study on register (Gregerson 1976) showed how vowels’ frequencies due to the voicing of initials differed from fifty to a few hundred hertz Still, other phonetic (excluding questions of language contact) factors had to have applied Only if the various features in question were to enhance each other could register/tone complexes come to have phonemic standing, though ultimately, some of the features were highlighted (e.g pitch contour) at the expense of others (e.g phonation)

Stage 1 Stage 2 (Registrogenesis) Stage 3 (Tonogenesis)

F IGURE 3: Three stages of development

As the case in Chong suggests, pitch or tone contour may have only been a consequence

of the other phonetic factors Clearly, register and phonation features must have come first, resulting in pitch contour Figure 3 shows a hypothesis for the three stages in Vietnamese from the lack of distinctive phonation, to register, to tonogenesis with the features shifting from the segmental to the suprasegmental level

This order of tonal development is better seen in tone reflexes in Ruc, a Viet-Muong language with presyllables (a typical feature of many Mon-Khmer languages but not of Viet-Muong languages) and tone on the main syllable The tonal correspondences between Vietnamese and Ruc are as shown in table 9 (Nguyen 1988)

Ruc -a1 -a2 -ap3 -ap4 (ah? ah?)

Viet -a1 -a4 -ap2 -ap5 (a3 a6)

T ABLE 9: Vietnamese and Ruc tones

It is not clear whether Ruc tone is a development shared with Vietnamese, thus coming from

a protolanguage, or an independent innovation The languages spoken around Ruc, such as Bru and other closely related Muong languages, do not have tone, supporting but not proving the former hypothesis In any case, the phonetic environment for the tones is the same in both languages, namely final stops for Ruc tones 3 and 4 corresponding to Vietnamese 2 and

5, while open or nasal final syllables show Ruc tones 1 and 2 for Vietnamese 1 and 4 As Ruc has not lost the final fricative /h/, it has not developed the final tone category seen in Vietnamese, which lost the final fricatives and gained a separate tone category Those tones

do not appear to have any governing conditions

Does monosyllabism matter? It would seem that two possibilities exist Vietnamese tone developed either before or after the reduction from sesquisyllabic forms seen in other Mon-Khmer languages However, there is a third possibility Ruc and Bulang (Paulsen 1992) do have tones in disyllabic words The tones occur only on the primary syllable, not the

unstressed presyllable Vietnamese tones could have developed before Vietnamese became monosyllabic, and perhaps not all the categories arose at once In addition to chronological ordering of tonogenesis, the Ruc and Bulang evidence suggests that the collapse to monosyllabic forms is not the sole impetus for tonogenesis as some have claimed (Benedict

1996, Matisoff 1973), though it could have heightened chances for the features’ interaction in the process of tonogenesis Register and the resulting phonetic features along with monosyllabification and language contact have led to the modern state of Vietnamese tones

In any event, there needs to be a way to show how these numerous features may have interacted Section 8 provides a list of relevant features and discussion of their mutual effects

8 TONAL AND PHONATION FEATURES In order for linguists to account for phonological and phonetic phenomena in languages, distinctive features must be utilized with as much restraint as possible to account for as wide a range as possible At this point, the relevant phonological features and their interaction are listed and discussed, with the hope that both synchronic and diachronic analyses and studies of tonal languages can benefit As this is a preliminary study, the goal is to be clear and complete, but also to be flexible enough to be useful to various theoretical frameworks

Table 10 is lists the features relevant to tone and register The table consists of the feature categories and the specific features needed to describe tones and register A binary distinction could also be used, in contrast to the multiple distinctions in the feature column

A binary split would allow the general grouping of sets of tones, such as the distinction between level and nonlevel tones, which makes sense diachronically in viewing nonlevel tones as a single group of phonological innovations, but this also represents a traditional view

by Chinese and Vietnamese for purposes of poetry in which those two categories are distinguished This two-way distinction in tone categories is also seen in Vietnamese morphological reduplication in which the tones of the two syllable forms alternate between level (tones 1 and 4) and nonlevel (tones 2, 3, 5, and 6), regardless of the specific tone (Hoang 1979) However, since not all of the logical possibilities of the combinations of the features has been worked out, a multiple rather than binary feature description is used The following is a general discussion about the interaction between the various categories of features

An important distinction to keep in mind in analyzing tones both synchronically and diachronically is the difference between tonal height and tonal contour in tonogenesis Tonogenesis shows both ‘contour-genesis’ and ‘pitch-height genesis’, in which the kinds of related phonation features differ In general, breathiness tends to be associated with low tones (e.g Vietnamese tones 4, 5, and 6) and dipping tones (e.g Northern Vietnamese tone 3, Mandarin tone 3) Southern Vietnamese no longer has much of its tonal creakiness, but rather has breathiness as seen in tones 4 and 5 Height tends to be the result of the phonetic shape of a syllable, though pre-existing creaky voice may also be a factor

Tone length is conditioned by the phonetic shape of a syllable Extremes in the length range (long or short) leads to other phonation features An example of a long tone is the Mandarin third tone, which is a dipping tone with creakiness Northern Vietnamese tones 3 and 6 both are long and have associated phonation features The extreme effect tones 5 and 6 have on the TBU in terms of length and glottalic quality is an example of how these related features interact and can enhance each other for maximal perceptual distinction