CHAPTER 1 : INTRODUCTION
3.2. Segmental phonology
Lewo has a total of nineteen segmental phonemes, comprising five vowels and fourteen consonants.3 In alphabetical order, using the orthographic representation adopted throughout this description, the vowels are
a, e, i, o, u,
and the consonants areg, k, I, m, m, n, p, p, r, s, t,
v, w, y.
In that the Lewo orthography is genuinely phonemic, these same orthographic symbols as used throughout this study are also taken in this chapter to stand for the phonemes they represent, eg.a
= /a/ etc. While the phonetic values of these segments are reasonably transparent, and are all discussed later, the reader’s attention is drawn to the following chart which indicates that orthographicg
is [g], andm
andp
are labio-velar4 nasal and plosive respectively.3.2.1. Consonant inventory
The fourteen Lewo consonants are given in the following chart (Table 3).
Table 3 : Lewo consonant inventory
L a b io -v ela r5 L abial A lv eo la r V ela r
P lo siv e P P t k N asal m m n 9 L ateral 1 T rill r C on tin u an t V s S em i-v o w el w y
The phonetic forms by which these Lewo consonant phonemes are manifested are now described. These phonemes derive from earlier stages in the development of Lewo, and correspondences
3 The known range for total inventories in the languages of the world is 11-141, with an average of 31 (Maddieson p. 7). Lewo is just outside what Maddieson considers the “typical” size of inventories (20- 37), which accounts for 70% of languages. Lewo’s 14 consonants are much fewer than the average of 23, and the five vowels less than the average of nine. However, the ratio of vowels to consonants in Lewo (0.37) is very close to the average of 0.4 (Maddieson p. 9).
4 Maddieson uses the term “labial-velar”, but I use “labio-velar”, the term more commonly used in Oceanic studies.
5 It is well-established convention now to “treat doubly-articulated stops as having a place of articulation distinct from that of either of their components” (Maddieson p. 18).
Chapter 3 Phonology
between them and the reconstructed POC phonemes they reflect are charted in Clark (1985:223- 5), which updates the first major systematic listing of correspondences by Try on (1976:11-50).
3.2.1.1. Plosives: p, p, t, k
On the basis of their phonetic salience, stops are considered to be the “optimal consonants” (Maddieson p. 25), and some variety of them is found in every known language.6 The most common stops are plosives, formed with egressive pulmonary (lung) air, and these are the ones found in Lewo. The four Lewo plosives, all taking voiceless phonation, are arranged at labio- velar, labial, alveolar, and velar positions on the chart of consonant phonemes.7
p, the labio-velar plosive, is phonetically a voiceless unaspirated labialised plosive with simultaneous release at bilabial and velar points of articulation, [£pw]. The phoneme has this manifestation before e and i, but before a, it has a very slightly implosive variant [£[>], and word finally, it occurs as [Icp]-8 It does not occur before o and u.
pa [ ® a | ‘emphatic evidential’
pig [ßpwirj] ‘sheath o f coconut flow er’
loperu [lo£pweru] ‘white fruit bat ’ kapkap [kafcpkalcp] ‘bent’
This sound in Lewo is used for borrowings that contain ba or pa sequences, eg. ‘bar’ is [Icjra:]. p, the labial plosive, has a primary phonetic manifestation9 as [p], a voiceless unaspirated bilabial plosive. It can occur as an unreleased variant [p1] word finally.
piga [piga] ‘cicada ’ lepepe [lepepe] ‘butterfly’
nop [nopn] ‘lizard’
6 “Stops are the only kind of consonants that occur in all languages” (Henton et al. 1992:65).
7 While only 16% of languages have just a single stop series like Lewo, the plain voiceless series as Lewo has (with more or less members) is found in 92% of languages (Maddieson pp. 26-27).
8 Note that here, and also for r?i, the allophone that is chosen to represent the phoneme is actually more complex phonetically than other allophones that occur. This is done on the basis of relative frequencies, but note that there is also historical evidence that the labio-velars have developed from labialised consonants.
9 In this and the following descriptions of phonetic manifestations, it is recognised that there will be a range of variation, occupying a certain region in articulatory space, of which the chosen primary phonetic manifestation is a convenient representative.
Although Lewo does have a bilabial fricative, which might be thought to be closer to English f and v than is this stop, p is used in Lewo for many borrowed words that contain f or v (as well as those with p and b of course).
t, the alveolar plosive. If it were not for Hawaiian, it could be said that every known language in the world has this sound (or a dental variety of it),10 and in Lewo it has a primary phonetic manifestation as [t], a voiceless unaspirated alveolar plosive. It occurs as an unreleased variant [ f ] word finally, but is only found in this position in borrowed words.
towa [towa] 'domestic fo w l’ kututusia [kututusia] ‘egg’
panet [p anef] ‘ship’s tender, barge (<BIS panitj’
k, the velar plosive, has a primary phonetic manifestation as [k], a voiceless unaspirated velar plosive. It occurs as an unreleased variant [\C] word finally.
kavika [kaßika] ‘Malay apple (BIS nakavika)’ kurki [kurki] ‘k. bush nut (BIS navel) ’
mak [rjmak'1] ‘le a f
3.2.1.2. Nasals: m, m, n, g
Lewo has four nasals, arranged at labio-velar, labial, alveolar, and velar positions on the chart of consonant phonemes.11
m, the labio-velar nasal, has a primary phonetic manifestation as [qmw], a voiced labialised nasal simultaneously articulated at bilabial and velar points of articulation. This allophone occurs before e and i, while before a, the phoneme occurs as the unlabialised variant [rjm]. Like p, it does not occur before o and u. m can be found word initially, medially and finally.
menta [rjmwenta] ‘k. termite’
ma [qma] ‘continuous aspect’
purmewa [punjm wewa] ‘k. coral trout (BIS lo s)’
mom [morjm] ‘your’
10Henton et al. (1992:79) state, with regard to stops, that “languages always have a voiceless dental/alveolar segment..., then /k/ appears in 89% of ...languages, and /p/ in 82%”. Hawaiian as an exception is noted in Maddieson (p. 35).
11 The three nasal system ([m, n, q]) is classic and common, but if a language is to add a fourth nasal to its inventory, this is nearly always the palatal nasal |ji]. In a few other cases, the fourth nasal is retroflexed [q], but the labio-velar nasal as in Lewo otherwise only occurs in systems with five or more nasals (Maddieson p. 64).
Chapter 3 Phonology
m, the labial nasal, has a primary phonetic manifestation as [m], a voiced bilabial nasal, which can occur in all positions in the word.
mene [mene] ‘red’
memaena [memasena] ‘green-blue’
mam [msem] ‘sweet’
n, the alveolar nasal, has a primary phonetic manifestation as [n], a voiced alveolar nasal, and is found in all positions in words.
ninlaia [ninlaia] ‘yellow or grey white-eye (bird) (BIS nalaklak) ’ kona [kona] ‘rotten ’
sirun [sirun] ‘brown wood borer'
g, the velar nasal, has a primary phonetic manifestation as [g], a voiced velar nasal, and can also be found in all positions in words.
ga [pa] ‘chew’
uga [uga] ‘gable o f ro o f
kolag [kolag] ‘k. fo o d baked as individual serves’
3.2.1.3. Lateral: I
I, the lateral, has a primary phonetic manifestation as [1], a voiced alveolar lateral. It can occur in all positions in words.12
levir [levir] ‘k. parrot (BIS nasiviru)’
kila [kila] ‘d o ’
mol [mol] ‘orange, citrus’
3.2.1.4. Trill: r
r, the trill, has a primary phonetic manifestation as [r], a voiced alveolar trill. The actual number of tongue taps that can be detected in the trill would best be subject to instrumental verification, but impressionistically varies between two to perhaps five for different speakers and in different environments.
roro [roro] ‘yellow fe ver’
12 Nearly all languages (96%) have one or more liquids, the most common arrangement (41%) being to have two such sounds. Of these languages that have two, 83% of them have one lateral, and one “r- sound”. The most frequent types of these two sounds are the alveolar lateral and trill (or its equivalent tap/flap; Maddieson pp. 80-84), which Lewo has, making it a fully typical language at this point of its phonology.
kulmaro [kulgmaro] ‘rabbit fish (BIS piko)’
mar [masr] ‘dry coconut; copra ’
A variant pronunciation of this consonant, [dr], can be set up to accommodate the situation where r occurs following n. This could be described as a process of phonological consonant epenthesis, but because o f its very limited distribution and occurrence, it is described here. It could also be described as an alternate of n preceding r, but it is considered that the allophone is a genuine phonetic variant of r, in the particular context provided by contiguity with n, rather than vice versa.13 Another reason for doing it this way is that under syllabification, speakers vacillate between breaking the sequence as Vn.drV or V.ndrV, but never consider Vnd.rV.
ve a-tol inu re [ßeatolindre]
NEG1 3pS-touch 1s NEG2 ‘Don’t touch me! ’
3.2.1.5. Continuants: v, s
v, the labial continuant, has a primary phonetic manifestation as [ß], a voiced bilabial fricative.14 Note that when it occurs word finally, it is devoiced. In other words, the sound is manifested as a rounded off-glide of preceding unrounded vowels (eg. maniniv ‘thin’ ends as [...niy]), or a more rounded off-glide of preceding rounded vowels (eg. ponov ‘nest ’ ends as [...no?]).
vio [ßio] 'Fijian asparagus (BIS naviso)’
vava [ßaeßae] ‘pandanus ’
purtev [purte0] ‘sugarcane ’
s, the alveolar continuant, has a primary phonetic manifestation as [s], a voiceless alveolar grooved fricative.15
suwe [suwe] ‘bivalve species ’ kumasua [kuqm asua] ‘triton shellfish ’
vis [ßis] ‘dig with fingers'
13 cf. Schütz (1985:552), describing the Fijian cluster phoneme dr, which sounds the same as the Lewo nr sequence, says: “the [d] is unusual phonetically in that we hear it ‘automatically’ when [r] is begun from closed position”. Ross (1988:31) also states that “[nr]...naturally becomes [ndr]”.
14 If a language has a fricative at this point of articulation, it is mostly voiced, rather than voiceless. The tendency that presence of a voiced fricative implies presence of its voiceless counterpart is well- supported, but the observation that bilabials are an exception is borne out in Lewo (Maddieson pp. 46- 47).
15 This is the most frequent fricative in the world’s languages, occurring in 83% of them (Maddieson p. 44). Languages with just two fricatives, like Lewo, will nearly always have [s] as one of them, but the pairing of [s] with [ß] is unusual (Maddieson p. 53).
Chapter 3 Phonology 3.2.1.6. Semi-vowels: w, y
These two approxim ants16 are called “semi-vowels” because of their vocoid characteristics.17
w, the labial semi-vowel, has a primary phonetic manifestation as [w], a voiced labio-velar approximant. This phoneme does not occur word finally.
wato [wato] ‘barracuda (BIS longmaot)’
kawe [kawe] ‘rat'
y, the velar semi-vowel, has a primary phonetic manifestation as [j], a voiced palatal approximant. It also does not occur word finally.18
yavia [jaeßia] ‘turtle’
ninuyu [ninuju] ‘orphan
3.2.1.7. Alternate structuring of consonant phoneme chart
In the chart of consonants above, conventional practice has been followed in assigning separate rows to continuants and semi-vowels. However, in that both of the semi-vowels are phonetically continuants, and in that there are only two cells filled for each of the semi-vowels and the continuants in the four-column matrix, it would be possible to conflate the last two rows of the consonant system and represent it as follows.
Labio- velar19
Labial Alveolar Velar
Continuant w V s y
However, there are no phonological processes for which this particular grouping of continuants represents a clear class of associated phonemes.
16“...consonantal sounds produced with relatively unimpeded flow of air through the mouth” (Maddieson p. 91).
17 These phonemes are frequently analytically problematic, but are nevertheless common, occurring in 86% and 76% of languages respectively.
18 This is the one Lewo phoneme that does not appear to reflect any reconstructed POC phoneme. Note however that reference is made later in this chapter to a sound change of v to y in some Lewo dialects. It is not likely however that this is the source for all occurrences of y in Lewo. In other instances, it has probably developed from intervocalic transition or prevocalic ongliding.
19 The w is formed from degrees of closure both at the front and rear of the mouth, and “falls into a clear class with other labial-velar consonants” (Maddieson p. 95). Maddieson also shows that there appears to be a dependence relationship between labio-velar sounds and [w]: if a language has some of the former, it is more likely to have the latter also.
3.2.1.8. More on the phonetics of the labio-velars
As noted, stops are the most frequently occurring kind of consonant in the world’s languages (Henton et al., 1992:65), but it is clear that labio-velar stops are rare. In M addieson’s hierarchy of preference of place of articulation for stops (Maddieson 1984:32), labio-velars are near the end of the list.20 They appear to be an areal feature of just two regions in the world, West Africa and Oceania.21 Ladefoged (1964:5-13) discovered that labio-velar stops such as [£p] were actually of three instrumentally-distinguishable sub-types, being produced by three different articulatory gestures. The first involved simultaneous articulation superimposed on an egressive pulmonic airstream, the second added a component of ingressive velaric air movement (as there is backward movement at the point of velar contact), while the third also involved downwards movement at the glottis, resulting in an ingressive glottalic air movement (and hence the production of a sound that involves all three air stream mechanisms simultaneously). He associates each of these with particular languages, but also shows that combinations of them do occur in the same language, and variation between them can be detected for different speakers. However, Ladefoged could only speculate whether this variation was responding to phonetic environment or not.22
The nasal labio-velar [qm] is not described in the same detail by Ladefoged, but he indicates that this coarticulation usually incorporates a velaric airstream component as well as the pulmonic air flow. The Lewo data is of interest firstly, because there is a close match between the variants of
20 Labio-velars of any kind are found in just 6% of languages (Maddieson p. 32), occurring less frequently than, in order, dental/alveolar, velar, bilabial, palatal/palatal-alveolar, uvular, and retroflex, and more frequently than very rare types of stops like pharyngeal, labio-dental, and linguo-labial.
21 The extent of their distribution in Oceanic languages can be seen in the data presented by Blust (1981). In that article, Blust gives Grace’s view that the labio-velars in Proto Oceanic “probably developed from simple labials next to a rounded vowel in Proto-Austronesian” (p. 232), but demonstrates the extremely difficult nature of historical reconstruction relating to the posited Proto Oceanic labio-velar protophonemes and their reflexes: “the great majority of labio-velar correspondences in Oceanic languages are contradictory” (p. 244).
22 Such instrumental measurement techniques have not been applied to Lewo labio-velars. Nor, one might add, have they been applied to the phonetics of hardly any feature of any language in the region. Over thirty years ago, Capell lamented that “Phonetics is another neglected branch of Pacific studies. In very few instances is there any satisfactory account of the phonetics of a language. No instruments for measuring any aspects of sound-formation have been used, and consequently all available descriptions are purely subjective. Until such instruments are available for field use they must remain so.” (Capell 1951:2). The CECIL speech analysis tool developed by SIL is now available, and I hope to carry out instrumental acoustic study of Lewo labio-velars at a later stage (CECIL does not deal with the articulatory components of speech production). Harrison (1981:191) has admitted that it has been his experience “in working with languages that have contrastive labialvelar consonants that the contrast is often difficult to hear before /a/”, and notes that Blust also acknowledges this. One of my own lingering doubts over some of the data presented in this description of Lewo is the status of some sounds as labio- velars or not, paticularly when these doubts are shared at times by native speakers.
Chapter 3 Phonology
the labio-velar stop [1$] and the labio-velar nasal [rjm], in that both have labialised manifestations ([£pw] and [gmw]) before the front vowels i and e, and secondly because this allophony parallels that reported for some West African languages.23 A further parallel between the phonetics of the West African and the Lewo labio-velars is that of implosion. While Lewo does not have implosives as such, it was noted above that very slight implosion of the labio-velar stop has been detected before a.24 This was based simply on observing speakers’ lips, but Ladefoged demonstrates instrumentally that this is true of some West African labio-velars, and gives an articulatory rationale for it.25
Note also that the restriction stated for Lewo, that labio-velars do not occur before the (back) rounded vowels o and u, is also observed for other Oceanic languages like Longgu (Hill 1992) and Nam akir26 (Sperlich 1993).
3.2.2. Vowel inventory
The Lewo vowel inventory presents the usual problem for five-vowel systems:27 how to demonstrate their relationship in a systematic way. The classic representation of five vowel systems is to arrange them in a two-way opposition with respect to the parameter of tongue position, and in a three-way opposition with respect to the parameter of tongue height, with the exception that at low tongue height, front and back are not distinguished. This “triangular” system (configured like that for three vowel inventories; Disner 1984:136) is an adequate representation for Lewo.28
23 Ladefoged (1964) notes “the well-known auditory resemblance between (labial-velars) and the corresponding velarized labials”, and notes that Araham’s description of the Tiv language has “kpw before e in complementary distribution with kp before all other vowels”.
24 Note that Schütz (1969:15) also detected implosion as a correlate of the articulation of these sounds in Ngunese.
25 Air pressures were measured in the mouth, behind the labial closure, and in the pharynx, behind the velar closure, as the labial-velar was articulated. Because the sound involves ingressive velaric air, it only takes a very slight delay in the release of labial closure, after velar release, for a negative pressure to result in the oral cavity. Ladefoged notes Siertsema’s description of this action as resulting in a sound that is “implosive at the lips and explosive at the back”.
26 Note that the name of this language is given by other scholars as Namakura or Namakira.
27 In terms of the number of segments found in vowel inventories, the five-vowel system represents the mode, occurring in 31% of languages (Maddieson p.126).
28 This representation is said to be “adequate” mainly because it has become the accepted way of representing five-vowel systems, and no-one has made any alternate suggestion that solves the problem of overspecification: there is a bi-dimensional matrix, with two values in one parameter, and three in another, giving a total of six cells, but only five are filled.
Table 4 : Lewo vowel inventory
Front Back
High i u
Mid e o
Low a
This representation does however run contrary to the well-established principle that vowel systems tend to develop and maintain configurations which reflect maximal dispersion in available phonetic (ie. acoustic) space,29 not to mention the somewhat counter-intuitive nature of the usual representation of the oral cavity as a box with lots of right-angled comers. An