Hewa phonemes: a tentative statement

2. Outline of Phonemes

2 . 1 Chart of phonemes

2.1.1 Work chart of contolds*

p

[p ph pi

bl

t
[t th t^ ]

[U

P

[p b]

[s zl

X

■

n

1

[I T f]

w y

2.1.2 Work chart of vocolds

t u

[t t] [u u]

e o

C A 3

a

^ Notice that there Is no /h/ phoneme, even though the people and their
language are called [*heiia]. This is a Duna word meaning 'sun*. (The Hewas
all live at lower altitudes than the Dunas.) All significant government
contact has come through the Lake Kopiago post in the Duna language area.

< . Ir-tfii"pretation ;f /ixbiva." f.Tit Seginents and Sequences

; . 1 status of vocoid segments

!.:.! Word-flrjal vocoids- -lost or non-existent?

The question of whether some words were losing their final voweis or
whether there were in fact closed syllables was a major concern throughout
the early analysis. After much reading, especially Hooper (1976) and
Lehiste (1970). a pattern emerged from the data. Native speakers also
recorded comparative data which was analysed with the aid of an
oscilloscope and "Its printed record. A pattern became clear showing Hewa at
present still has word final vowels, although phonetically some words
regularly end in a consonant, even in isolation.

Basically, the data showed that /;/ and /t,/ were being dropped
word-finally following /« n I s/. Since /i/ and /c/ are two of the shortest
vocoids in intrinsic duration ( Swadesh 1947:139; Lehiste and Peterson
1961:275; Lehiste 1970:18; dt Chene 1979:21), any process that shortens or
absorbs any of their duration would tend to cause their complete
disappearance. lax allophones Ti u] already cannot occur word-finally in
Hewa. The vowel .'a/' is already a lax vocoid and is easily dropped
word-finally. Section 3.1.2 attests to this loss of features as a general
pattern word-finally in Hewa. (Hooper (1976:235-40) discusses vowel
deletion as a historical process of losing features until finally becoming
unnecessary altogether.)

Two additional items serve as supporting evidence. One is that Hewa
as a stress language would tend toward such a pattern of deletion (Hooper
1976:227, 236). Regular word stress on the first syllable aids such a
word- final deletion tendency. Being far from one stress , but needing to
appear weak in relation to the stress on the next vowel spoken, the
word-final vowel devoices, or becomes aspiration on the preceding
consonant, or deletes altogether. The other item is that some of the most
common vocabulary appeared to be most consistently af fected--words such as
[ 'Bcan ] 'boy' , I 'en an ] 'girl ' , [ ' warn] 'wild animal ' , [ * 1 1 1 i o« ] 'cassowary* ,
['Kip] 'pig', l'cI: 'he', ['Lnicll 'sickness' , [ '■ Is] 'long, tall' , [ 'y is]
'smoke*. Hooper (1376:104) states, "Phonological change works its way
through the lexicon, moving from item to item.... the frequency of use of a
lexical item plays a part, i.e., more frequent items undergo change before
l^ss frequent items." She indicates that an erdeletion word-finally has
occurred in Spanish in this way.

Part of the early confusion was caused by /[/ and /c/ very rarely
being lost at the end of the past tense verbal suffixes /-ac/ [-bc] 'far
past, realis' and /-yc/ [~yc] 'recent past, realis'. It wasn't until the
pattern of vocoid sequences began to be clarified in the verbs, and
limitations seen on sequence length there (section 3.2.4), that I realized
each vowel or vowel sequence in the verb carried meaning, even the final
vowel. Therefore, it could not be dropped, as occurred in other parts of
speech being composed of single morphemes as words.

The consonants mentioned above ( /m n ! s/') also pattern to facilitate
this deletion word-f inally . As cont iuuauts they more readily are capable
of absorbing a very short vowel's voicing or duration into their own. This

54 Workpapers In Papua New Guinea Languages

is seen to be a universal tendency by Hooper (1976:206-29) as she details
her Universal Strength Hierarchy for consonants .

This loss of word-final vowels occurs with all vowels in the normal
flow of speech, especially when the preceding consonant is one of those
mentioned above, or when the next word begins with a voweI--especially one
which is phonetically close to it in its formation in the mouth, or when
the last consonant of one word is phonetically similar to the consonant
beginning the next word.

1. /'ana '30u/ ['an *pau] 'I have none.'

I neg.

2. /'clc a'pu/ E'cI a'3u] 'Where is he?'

he where

3 . / *napa ' atapa ' nasi / [ ' nap^ ' atapa ' was] 'Let 's all go ! '

we.pl all hort .go

4 . / 'napa 'no 'dIo '■otoyai ' nc le/

we, pi place there stay .cont .past that

['napa 'nolo 'moto'yai 'nelcj 'That is where we were staying.'

5. /'ho pi 'pile 'wi A'pe/ [ 'mob 'pile * wi a ' pc ]

man that name what 'What is that man's name?'

As vowel loss is predictable, all words shall be interpreted as having
a final vowel .

3.1.2 Voiceless vocoids and aspirated stops

Voiceless vocoids occur in the word-final position particularly when
the preceding consonant is a stop. This is due to the normal stress
pattern, as mentioned above. In addition, as pitch drops lower and lower
following the highest pitch on the stressed syllable, it compounds the
problem of speaking or hearing final vowels. All the vowels have been
recorded as voiceless word-finally. They will be represented as the voiced
vowels with which they vary, but never contrast.

/ '«as I / [ *«asl ] - C * mas] 'must go'

/wipe/ [ ' wi pC] - [ ' wi pc] - [w i p] 'domestic pig'

/'wase/ ['wasE] - ['wase] 'Let's do it!'

/'tiuBA/ i'tiuwA] - E'tium] 'place name'

/' wapa/ [ 'aapQ] - [ '«apa] ' forehead '

/' wawa/ [ 'wauwO] - [ ' wauwo] ' intend to eat '

/'topo/ ['topO] - ['topo] 'Put it (there)!'

/'topu/ ['tapUl - [*tapu] 'clouds'

Aspiration of voiceless stops [C^] in the word-final position occurs
infrequently, but varies with silent vocoids or voiced vocoids. The
aspiration will be represented by the voiced vowel with which it varies.

/'wolapi/ ['wolaph] ^ ['wolapl] - ['wolapi] 'all right'
/'lati/ ['ioth] - [-lotl] " ['lati] 'arm'

Hewa 55

/•titc/ ['tith] - ['tite] - E'titc] 'this'
/•Ris€tc/ ['wiscth] - ['wisctc] 'cold'

/•noke/ E'nokh] - ['nokei 'bird'

All voiceless vocoids and aspirated word-final stops will be
represented phonetically hereafter as voiceless vocoids, but phonemically
as voiced vowels.

3.2 Status of vocoid sequences

Although eight vowels are being proposed in this analysis, only seven
co-occur regularly in sequences. The vowel /a/ has not been found to occur
in sequences other than [Ia] and [ua].

Seven vocoids can theoretically combine into forty-two pairs (not
including geminates). Thirty-four of those pairs have been recorded in
Kewa speech and are listed in Figure 1 in descending order of the number of

actual sequences occurring .

Following Vocoid

-i

-u

-0

-a

-e

-c

-0

p

i-

—

iu

io

ia

ie

ic

io

r

u-

ui

—

uo

ua

ue

uc

e

o-

oi

ou

—

CO

oe

0€

a-

oi

au

ao

--

V

e-

ei

eu

60

ea

—

ee

eo

o

c-

c\

cu

CO

CO

—

CO

c

D-

oi

ou

00

OQ

—

Figure 1. Vocoid Pairs Heard in Hewa

3.2.1 Sequences of mid and low vocoids

No more than two vocoids articulated with mid and low tongue height
are ever found together at any one time within a word. Thirteen of the
twenty possible combinations from Figure 1 have been recorded. Each pair
takes the timing of two syllables. Primary or secondary stress may apply
to either member of a pair.

'male name*

'affirmative response '

'vomit * (noun)

'the Raggiana Bird-of -Paradise'

'must regularly do'

' therefore '

'female name'

•have not done'

'Wake (him) ! '

' Come 1 '

'already assembled'

'knows how to carve'

'from the centre, in the middle'

[oa]

/ * lol oapa/

[Molo^apa]

[oe]

/o'e/

[o'e]

toe]

/'koe/

['koc]

[ao]

/' yao/

[ ' yao]

[eo]

/' Baseo/

[ * MQseo]

[eo]

/'keleoBc/

[*kcle2a»G]

[ec]

/' lecBo/

[* Iec«a]

[eo]

/•peo/

[■peo]

[co]

/ ' tawc I co/

[ • ta"wclco]

[ea]

/sc'a/

[sc'o]

[CO]

/uc ' anco lopi /

[no ' onc^o lopI ]

[do]

/'«oo/

[•■00]

[oa]

/'poapo/

['poapQ]

56 Workpapers in Papua New Guinea Languages

These pairs are interpreted to be two syllables phonemically.

3.2.2 Sequences beginning with high vocoids

Other vocoids which combine regularly in pairs are those where the
first is articulated with high tongue height and the second may be any
vocold. Eleven of the twelve possibilities from Figure 1 have been
recorded. The only two sequences with [a] found to date are of this type.

[lu]

/•tlU«A/

E'tiUBA]

'place name'

[to]

/'■€kio/

['■ckio]

'customarily si

[ial

/t i ' aacne/

[t i ' aacne J

' slowly'

[Icl

/• 1 ieopa/

[» I ie^'opa]

'male name'

CiG]

/■sic/

['sic]

' (I) don't know'

[lol

/'ti3/

C^tio]

'ours [dual ) '

[ui]

/' su i»o/

[ * suiaa]

' female name '

[uo]

/■a' na luo/

[■A* na luo]

'customarily speaks'

Cua]

/• luo/

[' lua]

'wife'

[ue]

/* DDuenoHc/

[ ' * op'e^naaC]

•dragonfly'

Cue]

/'■aluenla/

[ 'aal u^cn ia]

'six (adj. ) •

[1a]

/'OpItlAPA/

[ *DpLtiAPA]

'male name*

[ua]

/*nukuApe/

['nuk"Ape]

'morning'

Stress occasionally falls on the second vocoid of such pairs, showing
the Independence of each of the vocoids, as for example,

/kopt'uaa/ [kopl'uaa] 'male name'
/tn I 'cpl/ [ttl I 'cpl ] 'afternoon'
/u'asi/ [u'osl] 'owner'.

In addition, most words of four syllables or longer have a secondary
stress which falls on the third syllable when counting the primary stress
as the first syllable. Often this secondary stress splits pairs of
vocolda, as narked above. On this basis, these vocoid pairs are also
interpreted as two vowels being peaks of two syllables.

3.2.3 Sequences ending with high vocoids

All ten pairs of a mid or low vocoid followed by a high vocoid have
been recorded (Figure 1). Those written phonetically as [yv] generally
take the timing of only one syllable, but those written [VV] generally take
the timing of two syllables. {The key difference between ['ao"] 'descend'
and [a3*u] 'swim', or [*ya"] 'dog' vs. [ya'u] 'sun', is not contrasting
tialng, but contrasting stress.)

[o*] /*axol/ ['agoi] 'sweet potato'

[ow] /'aopou/ ['aopou] 'will wrap/bind'

[ai ] /'palkoo/ [ *pa)koa] 'tapioca'

[a"] /■tonuBOune/ [* tanu*ao"ne] 'Pull!'

[e»l /'Icl/ [Mel] 'Jungle vines'

[eu] /'aaneu/ [*aaneu] 'will get'

Cc>] /'aelwD/ ['ac'so] 'intend to see'

Ccul /'acu/ ['acu] 'will have'

[3 » 1 / • to i ae/ [ ' to lac ] ' strong '

[o"! /'ouno/ ['auno] 'dip food in sauce'

57

As these are basically the reverse patterns of the pairs discussed
immediately above (3.1.2), and since no two-vocoid pairs having the same
stress pattern contrast their timing as one syllable vs. the timing as two
syllables (e.g. [o'j vs. [a.i]), these pairs ending in high vocoids shall
be regarded as the phonetic manifestations of the two vowels in sequence.

3.2.4 Sequences of three or more members

There seems to be a limitation on the duration of vowel sequences,
causing them to be pronounced as closely as possible to the timing of two
r.id or low vowels in sequence as two syllables, e.g. [e-o]* (See 4.3.3.)

When three or more vocoids are heard together, at least one is a high
vowel. Its duration is always brief and is always tied to the preceding
segment. When it follows a consonant it causes the consonant to sound
labialized or palatalized, e.g. /'kueala/ ['k^e'^alo] 'male name' (3.2.3).
When the high vowel follows another vowel, it is pronounced as the
off -glide, e.g. /'tiliau/ ['tili^a"] 'bat ' and /'mouanu/ [* «o"^anu] 'deep,
well-anchored (roots)'. Secondary stress also breaks the sequence as a
consistent feature.

When a sequence of three or more vocoids is potentially formed by
morphemes Joining within words, a semivowel is inserted to break up the
string or to replace a high vocoid. Since all Hewa verb roots end in a
vowel or vowel sequence, and since six vocoids are each a verbal suffix
morpheme, this insertion occurs regularly.

/'■ia/ 'is going' [ ' m i a ] < {•i-a}

go continuous

/'■iu/ 'will go' ['Miu] < {ai-u}

go future

/ ' mi you/' 'will be going' [ ' m iya" ] < (mi -o -u }

go cont fut

/'■jyc/ 'recently went ' I'miyc] < {ini-i -c }

go nr.pst realis

/* wi Hj, ' intend to go ' [ *■ i w j ] < (ni i -u -3 }

go fut irrealis

/'miyanj.' 'Intend to be going' I'miyo^uj] < !mi-o -u -o)

go cont fut irrealis

Examples where geminates are created across morpheme boundaries are given
in Section S . : .

Since the neaning is carried heavily In the vowels of the adjoining

morphemes, Hewa speakers separate poter.tlal squeezes and possible loss of a

vowel. The insertion of /; / fits a ;:niversal pattern in handling such
cases (Hooper, 242) .

58 Workpapers in Papua New Guinea Languages

3.3 Status of contoid clusters

No unambival'snt contoid coasters are to be found within the onset or
coda slots of Hewa syllables. Therefore, all phonetic clusters heard in
such positions are to be reinterpreted. (Clusters across morpheme
boundaries are discussed below in Sec. 6.2).

3.3.1 Word-Initial clusters

Contoid clusters are found only when the first syllable is unstressed,
and consist of a stop or fricative plus lateral. Such contoids will be
separated by re-inserting [l]/ which appears to have weakened in the
unstressed syllable to the point of being lost. (Hooper (1976:235,39)
states, "Stress languages which usually have vowel reduction processes,
insert or delete the minimal vowel.")

/'pile/ L*pLle] - ['pile] - ['pcle] ' mouth '

{The Sepik Hill Stock root for mouth/ teeth is [p i -b 1 ] . )
/p i 1 e ' ka/ [pi e ' ka] 'teeth' (mouth's appendages)
/p i J i ' pa i apa/ [pi i ' pa ' apM 'male name '
/t i ' I i / [ • tf I ] 'the Blue Wren Warbler'

3.3.2 Labialized stops, fricatives and nasal

All stops and fricatives (except /x/) and the nasal /■/ can be found
in /Cu/ syllables when not followed immediately by another vowel. Since
pairs of vowels occur in all environments, these [C'V] will become /CuV/.

/' Dpuenawc/ [ 'op^enamc ] 'dragonfly'

/' nukuApe/ [ ' nuk^Ape] 'morning'

/'tutuapo/ ['tutWapAJ 'from that far (place)'

/'kui/ ['k"i] 'the Huli people'

/'pue/ ['p^c] 'outer wrapping'

/'■ucBc/ [ '»"c«g] 'wove'

4. Description of Phonemes

4.1.1 Description of consonantal phonemes

/p t k p s x/ Apart from /t/ the obstruents vary freely with their voiced

allophones in initial and intervocalic positions.^ The stops also vary
freely with their aspirated allophones in both of those positions.

2 Some secondary evidence for [b] not being a full phoneme was noted when
words from Tok Pisln which begin with /b/ were generally pronounced by Hewa
speakers with /■/ or /k/ or /p/.

C'»csLkLs] biskit 'biscuit, cracker'

['■LRAtaq] binatang 'insect'

[ ' kusbo * ] bosboi ' government representative '

[ ' po* } boi 'boy, male'

[ ' p iS ] bis ' beads '

and/or pis 'tinned fish'

5e««a

59

/' puko/
/' napa/
/ ' pu ( u /

[•puka] - t'buka]
[ 'napa] - [ 'naba]
[ 'pul u ] - [ 'p^u lu]

'male name'
'we (plural) '

'hunting bow'

/* tapu/

/'otcte/

/•tIto/

[ 'tapu]

[•otete]

I'thito] - ['thitho] -

'clouds'
'work' (n.)
['tito] 'here'

/*ka«u/

/'nikat/

/'kDkopa/

[ 'k*>a«u] - C'ko»u3
['nlkai ] - [ 'nigai ]
[ 'kokopQ] - ['kogop^ ]

'frog'

•eye'

•nose'

/'pcne/
/'■opi/
/a'pu/

[ 'pcne]

[•■opi ] - [■■obi]

[a'pu] - [A'bu]

'you (dual) •
'married male'
'where?'

/' aaslo/
/' sunc/
/'si sale/

[ 'masi o] - [ 'muz io]
[ 'sunc]
[ 'sLsale]

'must go'
'Snap it! '
•ferns'

/ ' xo i pa/
/' axa/
/' saxD/

[ ' xo *pa]
[*axa]
[ "saxo]

' gourd '

'mountain range'
' bone spoon '

The velar consonants are backed further when preceded or followed by a
low or back vowel.

/' nokc, [ ' nok^ ] - [ ' no9c] 'bird'

/'kona/ ['kono] - ['^tona] 'banana'

/'pakAma/ ['bakAao] - ['bakAao] 'heel of foot'

/'axol/ ['oxoM - ['ago*] 'sweet potato'

/• 3X3 1 one/ [ ' Dxolone] - [ 'oicolane] 'new'

The stops may be unreleased word finally when the final vowel is

dropped.

/'wolapi/ ['wo lap I] - ['wolap^]
/ 'wi pG/ [ 'wipe] - [ 'wip"" ]

/' iwitc/
/* tiata/

[• iwLtc] - [• iWLt^ ]
['tiota] - I'tiot^ ]

'It's all right'
'domesticated pig'

'down lower on mountain*
'immediate vicinity*

/'papakuku/ [^papa^kukL] - [ » papa^kuk'' ] 'area name'

It is difficult to find many pairs of words where /x/ and /k/ phonemes
contrast clearly, yet there are a number of other reasons, including the
pattern pressure and my own "feel" of the language, for positing /x/ as a
full phoneme with limited distribution. Some of my reasons are:

-. [k k^ g k 9 X g y] as allophones of one phoneme is to overload the
s i ng 1 e phoneme /k / .

2. To group [k k^ q \f. g] as /k/ and [x g ^] as /x/ follows the clear
pattern of the allophones of /p/ vs. /p/ and /t/ vs. /s/.

60 Workpapers in Papua New Guinea Languages

3. Cochran (1968) proposed multiple phonemes /k x g/ for the western
dialect of Hewa.

4. Hepburn (1980) proposed multiple phonemes /k "^ h/ for Sanio, the
only other language within the Sanio Family to have been analysed.

5. Dye and Dye (1965) proposed /k g *> h/ for Bahinemo (Gahom) , the
next language eastward from Sanio in the Sepik Hill Stock to have
been analysed.

6. Bruce (1979:36) proposed /k g h/ for the far eastern end of the
Sepik Hill Stock. He also proposes a proto-Sepik Hill sound system
including /*k *g *h **?/ (1979:494).

On the basis of all these factors supporting an analysis of there
being more than one back consonant phoneme, I am positing /x/ as a
full phoneme.

/« n/ The voiced nasals occur in all positions in a word.

/'■cone/ CaeanC] 'unmarried male'

/'pisa«a/ ['pLsaMQ] 'sugar cane'

/' one/ [ * and 'me'

/*nok€/ [ 'nokG] 'bird'

/I/ [f] (infrequently) and [T] vary freely with [1] in intervocalic
position.

/'ale/ ['ale] ~ [afe] 'shoulder'

/o'lapcni/ [a'Tcpcnl] - [a'fapen] 'upper arm'

/ti ' I 1/ EtL' I i ] - [ 'tf i ] 'Blue Wren Warbler*

/tA'nalu/ [lA'nalu] * [tA'noTu] 'Speak! '

/'niliac/ ['nLliaaj - ['nLliaa] 'two days away from today'

/' lua/ [ ' lua] 'wife'

fm y/ The voiced semi-vowels occur word Initial and medial in non-peak
slots of syllables, [w] tends to follow a back vowel ( 69i»s of word
medial examples in the data); [y] tends to follow a front vowel (80%
of word medial examples in the data) ; this is characteristic of Sepik
River basin languages (Pike 1964:130; Hepburn 1980:7).

/*wolapi/ [•polapl] 'all right, O.K.*

/'wiwa/ ['wiwa] 'bark rope'

/'towai/ ['topa»] 'Hang (it) up!'

/'ylllwa/ ['yLllwa] 'male name'

/a«e'ya/ [a«e'ya] 'fruit bat'

4.1.2 Free variation between full phonemes

Some words have been heard where there appears to be free variation
between pairs of full phonemes in certain words, not as a general pattern

Hewa

across all words. These variaticDns are generally intervocalic, but
occasionally word initial. The predominant form is listed first in each

example ,

61

/p/ ' >/

/'pokole/ - /pokole/ 'grass'

/'vaiapa/ - /yaiapa/ 'male name'

/t/ - Is/

/'takupa,' - "sakupa/ 'white haired possum'

/ ' yakasonc/ - /' yak a tone/ 'area name '

(This place is at the edge of the language area, towards the Duna
language. Some people there are bilingual with Duna. The government
from Kopiago uses the /s/ form of the name. Duna has no /t/ vs. /s/
distinction. (Cochrane 1966) )

Part of the reason for the /l/-/s/ variation may be the influence from
Duna, where the three major areas of that language correspond to a split
partially along a /t/, /s/, /ts/ variation. The Lake Kopiago dialect,
which borders Hewa, uses /s/. Many loanwords from Tok Pisin with /t/ in
them are pronounced by Hewa speakers with /s/ or varying /t/ - /s/
because those words entered via Duna speakers.

/I/ - /y/

The choice of /!/ or /y/ seems to be a matter of personal preference.

'yati/ ~ /'lati/ 'male rope belt'

/ ' iiDp i 1 i / - . ' nop i y i / 'dark- fur red possum ( pha 1 anger ) '

/k/ - /x/

/'kaipa/ - /'xaipo/ 'breadfruit'

/'poxo/ - /'pake/ 'sugar cane pith'

4.2 Description of vowel phonemes

All vowels are found word initial, medial, and final.

\\.\ is an allophone of [i] and occurs when followed by a nasal or when the
vowel of the following CV syllable is a high vowel.

luj is an allophone of [u] and occurs word medially in the environment of

/p/.

"iwcli/ ['iwcli] 'sleeping mat'
.''niapo/ ■ 'niapa] 'age mate of the same sex'

/ ' i I i pa/ [ ■ L 1 ipa J ' bone*

/'sinu/ ['smL] "^ -' --7 i into the house)!'

/ ' i ncpt:/ [ * LHcpe] 'weapon '

62

Workpapers in Papua New Guinea Languages

/'enapa/ ['enapa] 'heart'
/'mewo/ ['mew3j 'intend to do'

/'maneu/ ['maneuj 'will get, hold'

/ • Gp imc/ L ' cb LinG ] 'already announced '

/pc' tcnc/ [pc ■ tenC] 'young'

/«c 'ancDiopi/ [mG^anc^oIopi] 'already assembled '

/'Ana/ ['Anal 'do like that, is like that'

/'tAiDCla/ ['tAmcljj 'roof'

/*»AXAli/ ['mAXAli] 'to hollow out, bore a hole'

/'alualu/ [*alu^alu] 'fruit dove'

/'atapa/ ['atapa] 'all'

/'yao/ C*yao] 'Raggiana Bird-of-Paradise *

/'ule/ ['ule] 'head covering'

/'kusiau/ ['kusia"] 'Ecclectus Parrot'

/'pupete/ ['pupGtc] - ['pupctc] 'pain'

/'wupc/ ['wupc] 'yesterday, tomorrow'

/'opa/ ['opaj 'hole/den in a tree'

/•oto'to/ [mato'to] 'must gather'

/'oxolonc/ [3xol3nC] 'new'

/■yDkole/ [ 'yokale] 'testicle'

/' l3p3/ ['13P3] 'theirs (plural)'

4.3 Suprasegmental Items

4.3.1 Pitch

Pitch is the most important factor of the suprasegmental features in
recognising the stressed syllable, although intensity and length are
interrelated. Only one syllable per word carries the highest pitch and
normally it is the first syllable. The final syllable has a low, fading
pitch. All other syllables have an intermediate pitch regardless of
whether they follow the high pitch or occasionally precede it. Absolute
pitch is not important. What is significant is the relationship of syllable
pitch within one word.

'SS / 'wi .pc/ 'pig'

/•yoT^/ 'Raggiana Bird-of-Paradise'

/*a7Ti/ 'shoulder'

' mar r i ed woman '

•SSS / ' aio - I u . a/

/' i . I i .pa/ 'bone'

'SSSS /'ku.Uu.tG.tc/ 'spider'

/' pi • pa . I a . le/ 'a hawk species'

63

S ' SS u . ' tu who/
S'SSS" /si .^ la.ku.m^/

SS'SS /ya.pQ.'si.nn/

'elbow, knee'
' i^zards '

-T

"U

SS'SSSS /nu . p i . ' sa .u . na . no / 'the Honeyeater tird'

SSS 'S /ma . to . po . ' 1 a/ 'must dispose of (something) '

Contrastive pitch patterns occur only when location of the stressed
syllable contrasts within a single string of phonetic segments. These
patterns are overlaid on the next examples below (Sec. 4.3.2) to avoid
repetition of data and to show the close relationship between the
suprasegmental features .

4.3.2 Intensity

Within a word there is one syllable which carries slightly greater
intensity than the rest. In the data under analysis more than 88X of the
words have that syllable first. The other syllables within the word are
heard with an intermediate intensity, generally decreasing as the word
progresses. The final syllable has the least intensity, often being
decreased to the point that distinctive features of the final vowel are
lost. (See Sec. 3.1 above.) This is particularly true of final syllables of
the shape

Im-l-i i

in- i-€ i

1 1 - ! -A I

Is-l-al

where the vowel has disappeared almost completely in the speech of a large
percentage of the Hewa speakers.

Even with the high predictability of the first syllable having greater
intensity, some words contrast only on the basis of which syllable carries
both the greater intensity and highest pitch. This combination of greater
intensity and highest pitch creates the stressed syllable and will be
symbolised by / * / phonemically .

/ ' matano/ ['ma. la. no]

/ma* tone / [«a . ' to .no]

'bean'

'must listen'

/' tcic/
/tc' \c

/ ' a^o ! c /
/a ' ^^I c /

'that distant (thing)'
'Scrape it! '

' grave '
'Where? '

64

Workpapers in Papua New Guinea Languages

/ 'yau/
/ya' u/

/'sic/
/si 'c/

[si .'c]

'dog'
'sun'

' (I) don't know'
'Shoot! '

Of those words where stress is not on the first syllable, it is more
often on the second than on another syllable. Examples of this as well as
stress on the third or fourth syllables are given in section 4.3.1.

A secondary Intensity {^) may be heard in words of four or more
syllables, and falls two syllables behind the one with primary intensity
(M- Being predictable, this secondary intensity will not be written
phonemically.

/ 'pipal ale/
/ke* ol UApA/
/'pi picme/

4.3.3 Length

[^pipa^lale]
[ke^alu^APA]

[1 pLpi ^CUC]

' a hawk spec i es '

'a male name'

' sharp , pointed , tapered '

Length has been recorded with vowels but not with consonants.
Lengthened vowels occur only in the first syllable and only when that
syllable is stressed. Such lengthened vowels vary with regular length
vowels in the same word and will not be represented phonemically. Only two
words have ever been heard consistently with length and not the variation
mentioned. They are both personal names and will be written without length
for the present.

/'tlpo/
/'kapi/

L'ti.pa]
[•ko.pi]

*a male name'
'a male name'

Non-contrastive Features

5 . 1 Onomotopoea

The words heard to date are related to animals and the natural
elements. These include the following:

/*p3polo/ 'bubbles, boiling action, gurgling stream'
/'nopupu/ 'a strong wind (associated with rain)'
/*u/ 'high-pitched howl of a dog'

(Hewa dogs don't bark.)
/'eya/ 'a bush fowl which makes this cry'.

/*mc\a/ 'cicadas, the time when they screech (approx. 6:30 pm)

/'wolawolo/ 'a bird which makes this cry (Black-headed Pitta)*

Hewa

65

5 . 2 Reduplication

Examples include single morphemes which have no meaning associated
with only the portion which is repeated. Others that do have meaning will
be broken down for clarity.

/ ' a i u a I u . '
/'walcHioIe/
f ' lopc I ope/
/ ' noma/
/' 3p i Dp i /
/" 3 i i 3 1 i /
/'kukueme/
/* itutuapa/

a fruit dove*

pleasant '

a draft or gentle air movement'

dance leader ' ( /bo/ 'beat drum' )

small portions or pieces' (/opi/ 'part')

on top ' ( /d 1 i / ' on ' )

chatted' (/kue/ 'words')

from very far down the mountain' (/tua/ 'far'

5,3. Nasalisation

Nasalisation has been heard consistently only on the words which are
the last three examples in section 5 . 1 above, all animal sounds . It is
interpreted as non-phonemic,

6 , Unsolved Issues

6.1 Consonant clusters in Duna names

Less than thirty words with consonant clusters of apparent Duna origin
have been recorded. These clusters occur word medially, and consist of a
nasal plus a stop of the same point of articulation. In Duna, they are
prenasalised stops, but are pronounced by Hewa speakers as consonant
clusters with a syllable boundary between the two members.

Of these, twenty-one are personal names. These individuals are of all
age groups and were born or raised on the edges of the Hewa language area
bordering the Duna language area. They have at least one Duna grandparent
and/or a parent bilingual with Duna.

[mb]

[ ' kcmba ]
[ ' amb i I a J ]

Lfid ] [ ' t Ltid i QKo]

[ ' anduku ]

Lag J

[ ' paw i oqgo 3

[ ' I opangi 1

male name'
female name'

male name'
female name'

male name'
female name*

Of the remaining words, one is a food which comes only from altitudes
higher than the Hewa language area, i.e., the Duna and Ipili language
areas, ['oogal ' pandanus nuts'.

Another plant name is [kampe] 'reed'.

One word of unknown origin is a beetle name, ['ko»batio] 'rhinocerous

beetle' .

66

Workpapers In Papua New Guinea Languages

The last words are related to seeing:

[ ' Loga' ] - [ * niko '
[tq'kikal ]

' eye '

'vision, eyesight '

Hewa speakers appear to have no problem pronouncing these names for
two reasons. One is the adaptation of making voiced stops voiceless or
accepting the voiced form as the variant allophone normally found between
voiced segments (see 4.1.1). The other reason is that nasals word-finally
(perhaps It is morpheme-f inally) regularly lose the following vowel,
especially in the flow of speech (see 3.1.2 and 4.3.2). That process
regularly brings consonants together across word boundaries or morpheme
jTinctions (6.2).

/'wa«G tili'oBc/ ['wam ti.li'o»] 'the wild cassowary'

wild. animal cassowary

/ * waac ' sua/
wild. animal bandicoot

[ * warn ' sua ]

'the wild bandicoot'

/'eaanc pe ' tene/
girl young

/ ' one tD*n3/

me give!

[ ' e«on pc 'ten]

[ 'an to' no]

'young girl '

'Give it to me!

As it appears that Hewa speakers sense a break coming between such
adjoining consonants, it will be assumed that the borrowed names from Duna
have been Hewa-cised and can honestly be represented with Hewa phonemes.

/■p/ from [«b]
/' kcwpa/
/' QKpi I ai /
/'koBpat io/
/'koMpe/

/nt/ from [na]
/'tintlawa/
/* antuku/

/nk/ from [Hg]
/*pa»i anko/
/* lapank i /
/' anka/
/in'klkai/

[kcaba]
['aabi laH
[koabat io ]
C 'koabe]

'male name'
' female name'
'rhinocerous beetle'
' reed'

['tLudiaaa] 'male name'
['onduku] 'female name'

[ ' pawiaqgo]
[ ' lapaqgi ]
[ 'aqga]
[iQ'kikai ]

'male name'
' female name'
'pandanus nuts'
'vision, eyesight'

6.2 Consonant clusters across morpheme boundaries

Hewa has very few words which are composed of two or more morphemes.
These appear to be limited to the pronouns, where a final vowel changes;
the verbs, where all roots end in a vowel; and the locatives. Only in the
locatives do consonants appear to cluster across morpheme boundaries, as
some 'location' -morphemes end in a nasal and all 'direction' -morphemes
begin with consonants, as shown below.

67

hoc Morpheme Dir Morpheme

OIB- -SU

upstream towards

Lii- -tut^apQ

downstream far- from

won- - p i 3 1 C

across there nearby

3- -pulapQ

higher from

i- -wtti

lower at

Any ' location' -morpheme of the first column must join with any one of
the 'direction' -morphemes in the second column to form a locative word.
All combinations are possible. Of all the morphemes indicated the first
three 'direction' -morphemes have been found to stand free. There is no
known association or borrowing from Duna here.

/ C? ) / L'amsu] * in the direction of upstream'
/ (?) / [ 'wontutwapQ] 'from far across (the valley)'

The patterns of pitch and intensity function as if each locative is
one word. The time given to pronounce the word ['amsu] is the same as
other two-syllable words. To leave the analysis with this as the
conclusion would appear to require positing two additional syllable types,
viz. VC and CVC , both being found only as first syllables of locatives or
in borrowed Duna names. This does not seem justified. The cluster appears
to be sufficient clue to the Hewa speaker and hearer to mentally divide the
word between the two members of the cluster on the basis of morphemes.

If the word-final vowel loss is regarded as morpheme-final instead,
then the earlier solution would apply here.

An alternative would be to insert the minimal vowel /g/ behind the
nasal in the ' location ' -morpheme creating */'amcsu/. However, when later
written, this may be pronounced *['a.mc. su] after the manner of /ome'ya/
[a.ne.'ya] 'fruitbat*, as a three-syllable word. This would have to be
watched when a practical orthography is taught.

Some examples are set out here :

/'amcsu/ ['amsu] 'in the direction of upstream'

/' imcpi::)lc/ [LmpiolCj 'nearby on the downstream side'

/' woncpu I apa/ ; ' wjnpu I opQ] 'from across on the mountain'

/'DtuluapQ/ ['DtutwopQ] 'fror. far up the mountain'

/'iwiti/ ['jALtl] 'low on the mountainside'

The better solution, I feel , is to regard the 'direction ' -morphemes as
unstressed clitics which are not bound, but purely are unstressed words

es Workpapers in Papua New Guinea Languages

causing a locative phrase to phonetically sound like on; word. The
word-final vowel loss pattern would thus still be app: cable. The examples
Immediately above would be represented like this phonemically and
phonetically:

/ ' amc su/ [ ' amsu 3

/ ' i»c 3 i Die/ [ ' Lap iolc]

/'monc pulopa/ ['wonpuIapQ]

/'3 tutuapo/ ['otut^apQ]

/■ i witi/ [* iWLt!]

7. Distribution of Syllables and Phonemes

7.1 Syllable patterns

Hewa syllable structure may be summarized by the formula:

S = (C) V

Two types of syllables may be generated from this formula and bo^h are
found in use. These are CV and V. Initial, medial and final positions
within a word are open to either syllable type. Words may be from one to
eight syllables in length. Some examples are given below.

One-Syllable Words - 2 of the 2 possible patterns occur.

V. /'3/ 'skin, leaf

CV, /'wa/ 'handheld drum*

Two-Syllable Words - 4 of the 4 possible patterns occur.

V.V /'do/ 'string bag'

V.CV /'aki/ 'chin'

CV.V /nio/ 'eye area'

CV . CV / ' t3 I c / ' tongue , flame '

Three Syllable Words - 7 of the 8 possible patterns occur (there are no
V.V.V patterns) .

V.V.CV /' aita/ 'father'

V.CV.V /'axoi/ 'sweet potato'

V.CV.CV /'alclo/ 'hunger {n. ) '

CV.V.V /"koai/ 'the Friar Bird'

CV.V.CV /'nicpa/ 'hair, fur, feathers'

CV.CV.V /'lipio/ 'theirs ( dl . ) '

Hewa 69

CV.CV.CV /'kamcni/ 'mosquito'

Four-Syllable Words - 15 of the 16 possible patterns occur (CV.V.V.V does
not occur) . The incidence of this word type is much higher when there are 2
or more CV syllables in the pattern.

V.V.V.V. /'aiai/ 'fight, argument'

V.V.CV.CV /'aipcse/ 'trunk of tree'

V.CV.V.CV /'iniclc/ 'sickness'

V. CV.CV.CV /o'loDcnc/ 'upper arm'

CV.V.V.CV /'lieopa/ 'male name'

CV . V . CV . CV / ■ m i cs i mc / ' deaci '

CV.CV.V.V /'toneau/ ' H^nd it (to me)!'

CV. CV.CV.CV /yapo'sina/ 'lizards'

Five Syllable Words - 16 of the 32 possible patterns occur. Those patterns
which do occur must have a minimum of two CV syllables. If only two such
syllables are present, they cannot be adjacent to each other. Examples are
nearly all personal names or verb forms.

Six-or-More-Syllable Words - Very few of the vast number of possible
patterns occur. As in the five-syllable words above, there must be 2 CV
syllables in a six-syllable word, and more in seven and eight-syllable
words, but they cannot all be adjacent to each other. Many verb forms are
six syllables, but the few longer verb roots create seven- and
eight-syllable words. The longest word recorded is /mc * anco I opi owe/
'(people) already were gathered'.

7.2 Co-occurrence of consonants with vowels in syllables

All the consonants may fill the onset slot and all the vowels may fill
the peak slot of either syllable type. Of the eight possible V syllables
and the 88 possible CV patterns, only eight CV syllables do not occur.
These involve combinations with /x// and /e/ and /a/. Front vowels and
high vowels are not found following .'x/ (*/xi/# */xe/, */xc./, */xu/»
*/xa/). The other syllables not found are */sa/, */yA/ and */ye/.

7.3 Vowel Clusters

When three vowels come together they consist of arrangements of seven
of the vowels, but not /a/. At least one vowel in a string of three must be
a high vowel, and it may be in any of the three positions. Two vowels of
the three may be high ones. The combinations heard are as follows:

70 Workpapers in Papua New Guinea Languages

f

e

a

G i

i a

a

i

o

Q

i

i

a

1

c i

i u

u

e

i

3

U

i

i

a

u

G i

1 o

u

e

a

3

u

a

e

a

i

a i

i a

u

a

i

3

u

e

a

u

a J

t u

u

a

u

When four vowels come together they consist of arrangements of only
three of the vowels (/i u a/). At least two of the vowels must be high
vowels, but are never adjacent to each other. The combinations heard are

as follows:

a i a i
a u Q u
tola

As discussed above in Section 3.2.4 the high vowels always are
pronounced with the preceding phoneme, generally as the off-glide of the
preceding vowel with total timing of the entire vowel sequence approaching
the timing of two unambiguous vowels.

7.4 Distribution of phonemes in words

All the consonants occur in initial, medial and final syllables of
words. All the vowels occur in initial, medial and final syllables of
words In both V and CV configurations. There is no restriction on the
location of any phoneme within a word.

8. Morphophonemics

8.1 Review of areas already discussed

One area of morphophonemics has been discussed above in Section 6.2
regarding phonetic consonant clusters across morpheme boundaries. This
problem is the chief ongoing concern left in morphophonemic analysis.

Another area already mentioned {Section 3.1.1) discussed word-final
vowel loss when the following word began with a vowel which was exactly the
same or very similar phonetically to the final vowel of the preceding word.
This also occurred when the consonant beginning the next word was exactly
the same or very similar phonetically to the last consonant of the
preceding word.

In Section 3.2.4 sequences of three or more vocolds were discussed.
Some verb roots, however, end in the same vocoid as the attaching suffix
and should create geminates or y-insertlon. It appears that geminates of
the high vocolds collapse into one, but other geminates are separated with
a semivowel. The collapsed geminates create two definitions for one
phonetic utterance.

/*■!/ 'go/recently went* [*■!] < {«i-i}

go nr.past

/'■opu/ 'stand/will stand' ['■opu] < {nopu -u )

stand fut

Hewa

71

/■03o'Iaya/ 'is disposing of [Hopo'la^ya] < {«opo*la -a)

dispose. of cont

/mul u ' onoyo/ ' repeatedly asks' [bu lu ' anoyo] < (wulu ' ano-o)

ask habitual

8.2 Clitics

There is one more category of phonological change which occurs. The
pitch pattern changes when clitics are attached to either end of a word.
The normal pitch pattern (Section 4.3.1) was described as having highest
pitch on the first syllable and following syllables being at a lower pitch,
the word terminating with a fading, falling pitch before silence.

8.2.1 Contraction of the first person singular pronoun

Often the first person singular pronoun /'ana/ 'I' is contracted to
/a/, loses its stress, and becomes phonologically bound to the beginning of
the next word. Though the combination of contraction and word sounds like
one new word, the word to which the contraction attaches still retains its
normal stress pattern. For example.

/*ana 'mou/ ['ana 'aa"]
I eat . future

'I will eat' (Full form pattern)

/a * mau/

[a'aau]

(Frequent contraction pattern)

It sounds as if the stress has shifted to the second syllable, but in
reality, the word 'eat' has not changed. Representing the combined words
phonemically as one word will allow the word 'eat' {or other verb) to
retain its normal pronunciation, but it will still be distinguishable from
/'aiiau/ ['ona"] 'food' by the stressed syllable.

8.2.2 Vocative

The same type of thing occurs in the pitch pattern when the vocative
marker /o/ is added as a suffix to a word or message being spoken across
the room or called across a greater distance, and is intended to draw the
hearer's attention. Some examples follow.

/'ycnc/ 'son' /'ycnco/ 'Son, . . . '

/ ' i na i / ' daughter ' / ' i na i o/ ' Daughter , . . . '

/ ' ap i g/ 'male name ' / ' api ao/ 'Oh, Apia, ... '

The pitch on the added final syllable does not fall and fade, but carries
on at the unstressed syllable pitch level.

In the following examples the contrast is the pitch level only. In the
first sentence the /o/ is integral to the word meaning; in the second it is
at the utterance (sentence) level of meaning.

72 Workpapers in Papua New Guinea Languages

/'ana 'm\o/ 'I often go'

I go. habitual

/'ana 'Bio/ 'I'm letting you know that I'm going now.'
I go-voc .marker

I propose using the space to indicate the levelling of the pitch by
regarding the vocative marker as an unstressed word, the same as was
proposed In Section 6.2.

e.g. "Ana mio." vs. "Ana mi o."

I go. habitual I go voc. marker

8,2.3 Conjunctions

When items are together in a series, a suffix is added to each item
and affects the pitch pattern of the word to which it is attached, the same
as mentioned In Section 8.2.2 immediately above. Nouns or clauses may be
Joined. The space will again be used to indicate the phonological change Ln
pitch pattern. An example follows.

/'napa ki'awApA wcpc so'saiapa wcpc 'aliou wcpe
we . pi KiawApA and Sosalapa and Al iau and

*nopa 'ote ' uwe se *ycl«c/

we.pl river bank at go . upstream. direction .past

•We (KiawApA, Sosalapa, Allau and I) went upstream along
the river's bank.'

8 . 3 Verbs

All Hewa verb roots end in a vowel. When tense and aspect markers
beginning with a high vowel (or semi-vowel) are added as suffixes to verb
roots ending In /a/, they cause the /o/ to glide. The examples shown below
are represented phonetically to Illustrate this point.

Far Past Near Past Future Past
Root Present Realis Realls Irrealis Completed

/-0/ /-a-e/ /-I-c/ /-u-o/ /-po-a-c/

*'a" ' ao ' aoae 'sa^yc 'ma"w3
eat

aapoae

-le'a- aele'a acle'aac ecle'a'yc acIe'auwD acle'cpoac
return

-*na- MA'na sa' naac aA'noiyc «a 'nauwo ba 'napoac
hit, beat

-•an I no- 'aonlna 'aaninoac 'aanlnaiyc 'aanlnauwo 'aaninapoac

Hewa 73

All other vowels at the end of a verb root do not change because the
meaning is tied to the vowels.

-'uwa^- 'nuwo* 'auffa'ac 'Muwafyc 'auiraiHo 'muwa'poac

fill

- ' aneo"- '■anea^' 'iBanea"»c *«anea"yc * aaneauirD '■anea"po«c
hand over

- ' 3- 'bo ' momc ' moyc ' ■owo ' bdpobc

chop, carve

- ' DU- • Bou 'aDuac ' Mouyc ' bouko ' aaupoac
descend

-tu'ono- 'aulu'ano aulu'cnoKe mulu'anoyc Bulu'cnowo mu I u ' anopoac

Because no other vowels glide when morphemes are suffixed to the verb
roots, but the roots retain their shape, these morphophonemic glides on /a/
will not be written, but the basic root form will be kept. This is subject
to later testing when there are Hewa literates. (Cochran 1977: 91-92).

Three verbs have been found to follow another pattern when only the
future tense vowel suffix /u/ is added to the root. The last vowel is
dropped and replaced by the /u/. These verbs return to the pattern when
additional suffixes are added. The expected form is given last in the
following examples .

/' mcku/ 'will sleep' < (acki -u } ( */'mcki u/ )

sleep fut

/'mckiwD/ 'intend to sleep' < {mcUi -u -a}

sleep fut irrealis

/■«ctu/ 'will build' < {mcti -u) (*/'«ctiu/;

build fut

/'mcliwD/ 'intend to build' < {acti -u -o )

build fut irrealis

/ao'pu/ 'will put' < {ao'po-u) (*/ao'pou/)

put fut

/ao'powo/ 'intend to put' < {»o'po-u -3}

put fut irrealis

The phonemic shapes do not have the same form as any other known verb
root, so there does not seem to be any confusion with other verbs. Also,
later transfer of the phonemic shapes to an orthography wil' probably cause
fewer errors for new readers than might spellings of the morphemic shapes
on the right. Therefore, I prefer to use the phonemic shapes as the
exceptions to the rule, rather than the morphemic shapes . This , too, is
subject to later testing.

Workpapers in Papua Kew Guinea Languages
Volume 31

Five E>l:^ono3. osr J- <=SLi S-t:\aca±GS3

Summer Institute of Linguistics
Ukarumpa, Papua New Guinea