34
Warluwarra Grammar | Gavan Breen
3. PRONUNCIATION
3.1 The lax stops.
A lax stop is most commonly voiceless or weakly voiced and unaspirated. Rarely heard
realisations are glides or zero, e.g. [wuʈijalu] or [uʈijalu] for burtiyalu ‘for a long time’.
These may be attributable to the advanced age of the informants, but are perhaps of some
interest in view of the tendency in Warluwarra for stops to undergo lenition. This was
discussed in the overview.
/b/ is a bilabial stop. Realisations other than [p], [b̥] or [b] are rare.
/g/ is a velar stop, and is released without noticeable friction. Labialised varieties occur
in the environment /u-a/, e.g. [ɳukʷaɾa, t ̪ukʷa] for nukarra ‘wind’, dhuka ‘throw (it)!’. In
fact, it may be stated as a general rule that any velar may be labialised in the environment
/u(C)-a/, where C is any consonant and () encloses an optional segment.
/dh/ is an interdental (lamino-dental) or dental (apico-dental) stop, whose release is
usually accompanied by some friction. The friction is much less noticeable with the dental
than with the interdental allophones. Tongue position for the interdentals is with the tip
protruding slightly between the upper and lower teeth, while in the dental articulation the
tongue tip touches the edge of the upper teeth. A weak voiced fricative [ð] sometimes
replaces a more common [d̪], for example in [d̪i:n̪ða] for dhindha ‘fat’ and in many
realisations of the vowel suffix -yidha. /dh/ has less tendency to be voiced than do the other
voiceless stops, especially in the word-initial position. The dental allophones, in particular
the voiceless dental stop, are more common word-initially than in other positions, possibly as
a result of the absence of /d/ from this position.
/dj/ is lamino-alveolar; the occlusion is made by pressing the blade of the tongue against
the gum ridge, the tip of the tongue being behind and in contact with the lower teeth. Like
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
35
/dh/, this phoneme has a somewhat fricative release. The common allophones are [tʲ, dʲ,̥ dʲ].
In the environment /#-i/ a voiceless palatal allophone [ç] has been heard. Occasional
realisations, possibly influenced by English, are [tʃ] and [dʒ].
/d/ is one of the less common phonemes and occurs only intervocalically or in the
intervocalic cluster /nd/. It is an apico-alveolar stop, and no noticeable friction accompanies
its release. Allophones are [t, d̥, d].
/rd/, one of the rarer phonemes, is a retroflex, or apico-postalveolar stop. The occlusion
is made by pressing the tongue tip against the palate, and the release is not noticeably
fricative. In word-initial position /rd/ may at times be fronted towards [d], but clearly
retroflex realisations are common also. As the opposition between /rd/ and /d/ is neutralised
in this position, this does not lead to confusion. The other common allophones are [ʈ, ɖ̥, ɖ].
3.2 The tense stops.
Tense stops are typically voiceless and, when intervocalic or in a homorganic cluster,
longer than simple stops, often being heard as geminated, one stop closing a (phonetic)
syllable and unreleased and the second, after a short period of silence, forming the onset of
the next syllable. This is often associated with shift of stress from the vowel before to the
vowel after the stop (see 3.9.2). However, it applies also to the rare fortis stops in later
positions in a word, usually in the final syllable of a trisyllabic word, which are occasionally
phonetically geminate. Examples are bumathamatha ‘every day’ (from bumatha ‘sun, day’),
yungutha ‘east’ and bapurtu ‘a few’. In a few cases the fortis nature of a consonant in a third
syllable, in a word in which the consonant of the second syllable is a glide, may be attributed
to the first two syllables coalescing to give a stressed diphthong as the initial phonetic vowel;
examples are wayitjaga ‘in the net’ and wiyaparriyidha ‘is calling’.
In a heterorganic cluster there is no contrast between lenis and fortis stops, and the stops
(which occur in second position) are voiceless and somewhat fortis; for example, in
dhurrkurnu ‘back’, garnkalwiya ‘top’, marrkana ‘hand’, ngurlpa ‘many’, garlilpi ‘gidgea’.
(Note that the stop written /k/ in clusters /nk/ and /rnk/ is simple.) There is a single instance
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
36
of a stop-stop cluster, in ngartpangu ‘gum from bloodwood’, which is probably a borrowing
from Bularnu, and is also the only known instance in Warluwarra of a consonant cluster with
a stop as the first member.
3.3 The nasals.
The nasals are voiced and have the same points of articulation (as far as can be
observed) as the corresponding stops. Apart from the standard forms — [m, ŋ, n̪, nʲ, n, ɳ] -the nasal phonemes have few allophones. /nh/ and /ny/ have in most environments, but
especially intervocalically, a noticeably longer duration than the other nasals. The transition
from /nh/ to the following vowel is accompanied by an interdental glide or weak fricative
while /ny/ has a palatal off-glide. These are not evident in nasal-stop sequences and it is
difficult to distinguish between such sequences as [n̪d]̪ and [nd̪], [nʲdʲ] and [ndʲ] (and
similarly for the corresponding clusters with geminate stops). As far as the laminal-dentals
are concerned there is no evidence of any distinction. In the case of the laminal-alveolars
there is some evidence: Hale transcribed ngantji ‘thin’ as [ŋantʲi] and bundjurra ‘emu’ as
[puntʲuɾa] although I have heard them as [ŋanʲtʲi] and [punʲtʲuɾa]; I have heard the word
dintja ‘snake sp.’ clearly with the heterorganic cluster, but on another occasion with the
homorganic cluster. I conclude that the contrast is not phonemic and there is free variation.
The clusters are assumed to be at least underlyingly homorganic, although there is a
simplification of the orthography and they are written as /ndh/, /nth/, /ndj/ and /ntj/.
Rarely, a nasal is realised simply as nasality in the preceding vowel; this has been noted
only in nasal-stop sequences. There is no evidence of any significant difference in this
respect between nasals before simple or geminate stops. Examples are [uŋaɾa̋ˑba] for
Wungarramba (a place name), [wa̋:pa] for wampa ‘dry’ and [maɳŋa̋ˑdʲi] for marnngandji
‘tree sp.’. In barnnga-barnnga ‘flat’ the /rn/ may be realised, in both occurrences, as
retroflexion as well as nasality in the vowel.
An occasionally heard allophone of the two apical nasals is the corresponding nasally
released stop, initially voiceless, occurring after a short stressed vowel, for example in
[utnaˌmaid̪a] (wunamayidha ‘spilling’), [buʈɳa] (burna ‘white’) and [gatnama] (ganama
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
37
‘push (it)!’). See 2.2.4 for a consideration of whether these should be analysed as containing
geminate nasals. Also, on a few occasions, a atop has replaced an apical nasal as the first
member of a nasal-nasal cluster, as [watmani] for wanmani ‘black’ or [ʈaʈmaɳa] for
darnmarna ‘bit’.
A dental allophone of the interdental nasal /nh/ is often heard, especially word-initial.
/rn/ is sometimes fronted towards [n] in word-initial position. As /n/, like the other alveolar
phonemes, does not occur contrastively in word-initial position, no confusion with /nh/ or /rn/
can arise here.
There are some indications that the opposition between /n/ and /rn/ may be neutralised
in clusters in which /m/ is the second member, and possibly also in some other clusters, e.g.
with /k/. For example, there is fluctuation in transcriptions between djanma and djarnma
‘speak’, danma and darnma ‘bite’. A suspected sub-minimal pair was gunma / ngurnma, but
when this was checked with Queen the realisation [ŋunma] was heard. The situation is,
therefore, not clear, but I am assuming that this confusion is caused by the age of the speakers
and the unclearness of their speech, and that these words, apart from gunma, have /rn/.
3.4 The laterals.
/lh, l, ly, rl/ are voiced laterals with the same points of articulation as the corresponding
stops. /lh/ and /ly/ are very common. The opposition between /l/ and /rl/ is neutralised wordinitially, and the apical lateral appearing there is frequently retroflex but is written /l/.
As mentioned above (2.2.4), laminal-dental or retroflex lateral phonemes may be
realised as a long consonant, e.g. [kaɭ:a, ŋul ̪:u] (garla ‘meat’, ngulhu ‘grass’). In the case of
a geminate lateral, e.g. dhawullu ‘jaw’ pre-stopping or flapping of the lateral is a common
alternative to plain length: thus [t ̪awul:u] or [t ̪awudlu]; [t ̪awulu] was heard only once and
immediately repeated as [t ̪awuľu].
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
38
An occasional realisation of /l/ and /rl/ in careless speech is as the corresponding glide,
[ɹ] (which is also a very common allophone of /rr/) and [ɻ] (which is, of course, the normal
realisation of /r/). For example, [maniɻig̊a] for manirlika ‘while travelling’.
3.5 The alveolar tap.
The normal realisation in stressed position (i.e. following a stressed vowel) is the
alveolar tap [ɾ]. In unstressed positions it is most commonly an alveolar glide [ɹ]. A
common variant in the idiolect of the main informant, Queen, is a voiceless tap [ɾ̊] often
accompanied by a distinct aspiration [ɾ̊ʰ]. Following a long stressed vowel the contact
between the tongue and the alveolae is often protracted, so that the sound approaches [d],
Rarely, there is a brief trill [r].
3.6 The glides.
/w/ is most commonly realised as a labio-velar semi-vowel [w], but also frequently as
zero or as lengthening of a following /u/ (see 2.2.4). Word-initial /w/ before /a/ or /i/ is
almost always [w]. Where /w/ forms the second member of a consonant cluster it is
sometimes realised as zero, or, more frequently, as length in the preceding consonant. See
the discussion in 2.3.4.
/y/ is often a palatal semi-vowel [j], but is also often realised as zero or as lengthening
of a following homorganic vowel, /i/ (see 2.2.4). In word-initial position, /y/ before/u/ is
always [j], but before /a/ it may occasionally be zero, or, very rarely, [ij] or [j^]. /yy/ is
realised as a voiced palatal fricative [j^] (by Fred Age) or as a voiceless palatal fricative [ç]
(by Queen, in both morphemes in which she used it). [j^] is distinguished auditorily from [j]
chiefly by its longer duration.
/r/ is a retroflex frictionless continuant [ɹ] in most realisations, but may, in intervocalic
position, be realised as zero. In effect, this may result in a [j] in such environments as /i-i/
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
39
and /i-a/. Examples are [ijat ̪a:ˌɾi:dʲa] for yirathaarriyidja ‘is afraid’ and [t ̪awiji] for dhawiri
‘small’. (The reduplicated form of the latter, dhawiyawi, probably involves lenition of the
initial consonant rather than a change to the /r/,) /rs/ is a voiced retroflex fricative, but with
considerably less friction than [ʐ] and distinguished auditorily from the glide [ɻ] mainly by its
greater length; or (in most realisations by Queen) a voiceless fricative [ʂ], sometimes with
some voice initially and thus approaching [ʐ].
/gh/ is typically a velar frictionless continuant. Like other velar phonemes, it is
labialised in the environment /u(C)-a/ and may then be indistinguishable from /w/; not
always, because the labialisation may not extend over the entire duration of the phone. An
example of labialisation is [jaɾwulu] for yarrghulu ‘one’. It has been heard as a labial also in
the word yughu ‘tree’; i.e. [juwu], but this is rare. Another common allophone, especially
but not exclusively used by Queen, is a voiceless variety, a weakly fricative [x] or even zero,
used intervocalically after a stressed vowel, e.g. [juxu], [ŋaxad̪a] for yughu, ngaghatha (the
latter ‘grub sp.’). In consonant clusters where /gh/ is the second member, it is often realised
(or at least heard) as zero, as length in the preceding consonant, or rarely as [x].
3.7 The vowels
Since there are only three vowel phonemes, one would expect each to have a wide
range of allophones. This is so. In the headings to the following descriptions I have used the
conventional designations: low vowel, high front vowel, high back vowel. It would be more
faithful to the nature of Warluwarra, and of many other Australian languages, to use, say,
open vowel, palatal vowel and labial vowel respectively (as implied by the phoneme chart in
Table 2.1).
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
40
3.7.1 The low vowel.
Figure 2.1 illustrates diagrammatically the range of normal realisations of the phoneme
/a/ when short and stressed. In this diagram, B represents any bilabial consonant, D (for
dental) any interdental, A any alveolar, P (for palatal) any laminal-alveolar, R any retroflex
and K any velar.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
41
Figure 2.1
following consonant
/y/
/y/
P
ɛ
ɛæ
A
R
D
æʳ
æ
/w/
æ
æʳ
P
B
K
æɐ
æ ɐ
ɐʳ
A
preceding
R
consonant
D
ɛæ
æ ɐ
æ ɛ ɐ
ɐ
ɐʳ
B
ɐ
/w/
K
æ
ɐ ɒ
ɐ ɒ
For example, when in the environment /dj-t/ (i.e. P-A) and stressed, /a/ is realised
approximately [æ]. In the environment /y-t/ (i.e. j-A) and carrying some stress, /a/ will be
realised in the vicinity of [ɛ] or [æ], or somewhere between them.
Figure 2.1 is simply a summary of average pronunciations and is necessarily simplified
to some extent. For example, we are told that the first /a/ in yandawu ‘in/for a while’ and in
yarra ‘river’ will be [ɛ] or [æ] (if it is a short vowel; the rules regarding vowel length will be
given in 2.3.8.4). but we are not told that in yandawu it is normally [ɛ] while in yarra it is
usually [æ]. The reason for this difference is not clear; it may be connected with the fact that
in yandawu there is some stress on the second syllable, and the first vowel is realised with
less stress than in yarra. (Stress is discussed in 3.9.)
The figure does, however, illustrate some important points:
/a/ is realised as a low vowel, but normally somewhat higher than the low cardinal
vowels [a] and [ɑ].
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
42
/y/, whether preceding or following, tends to raise and advance /a/, to [æ] or even as
high as [ɛ]. This is stronger than any of the other environmental influences on this vowel.
Lamino-alveolars and, to a lesser extent, alveolars tend to raise and advance /a/ to some
extent, especially when preceding it.
In spite of the frontness of their articulation, interdentals and bilabials show no
tendency to advance /a/, except for the anomalous case of the environment R-D. In fact,
bilabials favour a rounded back articulation, close to that of [ɔ], in some environments, e.g.
wapa ‘mother’s mother’ may be [wɔpa], gamarndili ‘liver’ is usually [kɔmaˌɳɖili].
The anomalous appearance of [ɛ] and [æ] in the environment R-D is exemplified by
[iɻæt ̪aˑˌɹi:dʲa] for yirathaarriyidja (contrast the pronunciation given above, in 3.6) and
[ɭɛ:ɭɛd̪a] for liyarlatha ‘heart’. Velars and /u/ also favour retracted and occasionally rounded
articulation; cf. the two examples in the preceding paragraph.
/a/ is raised and retroflexed before retroflex consonants, occasionally even becoming a
syllabic [ɻ], e.g. [bɻʈawɻʈa] bartawarta. 'big'.
In unstressed syllables /a/ is generally [ɐ] (but normally written [a] in examples) and is
not greatly affected by environment, with the exception that [j] tends to raise and advance it
in the same way as described above.
When realised as a long vowel (the circumstances under which this happens are
discussed in 2.2.6) /a/ is lowered to an approximately central low vowel [a] or [ɑ̘]. A
following lamino-alveolar may be preceded by a palatal on-glide, e.g. [maʲtʲu].
The double vowel /aa/ is realised almost always as a long vowel, similar in quality to
the long realisations of /a/ but often of longer duration. Unlike /a/, /aa/ retains the quality of
the long vowel even when circumstances, such as rapid speech, cause it to be shortened.
Nasality has been noted only rarely, and only as the realisation of a nasal phoneme; see
2.3.3.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
43
A rising intonation at the conclusion of an utterance (e.g. an exclamation) may be
accompanied by lengthening of a final /a/ and its distortion, usually in the direction of [u],
e.g. [n̪andau], ‘go on’ (said with urgency), but sometimes in the direction of [i], as in
[d̪aɻaˌjiwai] ‘Where is he?’.
Word-final /a/ (and other vowels) may occasionally be reduced to a voiceless form, e.g.
[ɒ̥], or even to zero.
3.7.2 The high front vowel.
/i/, when short, is most commonly realised as approximately [ɪ]. Following /y/ or a
lamino-alveolar, it is usually raised towards [i]. Preceding /w/, it is retracted to [ɨ], or may, in
addition be rounded to [ʉ]; yiwa ‘he/she/it’ may be [i:wa], [jɨwa] or [jʉwa]. A following
labial consonant may have a similar, but less marked, effect. In certain environments /i/ may
be lowered towards [e]; this has been noted word-finally (the suffix -mi as [me]), before an
alveolar (e.g. [jeta] for yita ‘tie [it]!’ and [jenda] for yinda ‘for you’) and in ngawiya
‘whatsaname’ (rarely [ŋaweja]). Before a retroflex, /i/ is centralised and retroflexed, and
may be close to [ə], e.g. [itʲəɻa] for Yitjira ‘Urandangie’
When /i/ or /iyi/ is realised as a long vowel, a phone in the neighbourhood of [i:] is
generally heard, but there are occasional lower realisations, e.g. [me:] for miyi (relative clause
marker)
3.7.3 The high back vowel.
/u/ is typically [ʊ] when short, and slightly higher, [ʋ̝:], in its long realisation. Before
/w/ it may be realised as zero, as in [ŋwɔkaˌmɛid̪a] (nguwagamayidha ‘hiding (something)’)
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
44
or [kʋŋwa] (gunguwa ‘coolibah’). Before /y/, [ɥ] has been heard, as in [ŋɥi:nda] for
nguyinda ‘east’. In word-final position it is rarely lowered towards [o], e.g. [kunatʲaɾo]
gunadjarru ‘black bream’. In the environment P-B an advanced form is heard, as in djuwala
‘night’ ([tʲʉwala]) and wundjumurru ‘tail’ ([wunʲdʲʉˌmuɾu]). This is also heard,
unpredictably, in other environments, e.g. [kaɭʉˌn̪aˑnda] for garlu nhanda, ‘go over there’,
[u:laˌŋʉɭu] for wulangurlu ‘from those two’ and [wupaˌmaɭʉ] for wupamarlu ‘because of
him’. Unstressed /u/ may be lowered and centralised to [ə], e.g. [wuɾən̪a] (wurrunha ‘you,
plural, accusative’).
3.7.4 Vowel length.
As stated above (2.4), the sequences /aa/, /iyi/ and /uwu/ are most commonly realised as
long vowels, approximately [a:], [i:] and [u:], while other sequences of the form VWV (e.g.
/awa/ and /ara/) may also be realised as the corresponding long vowel, as also may /anga/ on
occasion. The sequence /iya/ is heard as [ɛ:] in a couple of words: liyarladha ‘heart’ and (at
least some inflected forms of) wiyapa ‘to call out’. This does not apply to the (few) other
words beginning with Ciya, such as giyalparri ‘type of tree’. The remainder of this
discussion applies to the phonemically simple vowels /a, i, u/.
Phonemically simple, or short, vowels are normally short. A vowel that would
normally be long in slow speech is usually shortened in fast speech. A vowel that is long in a
disyllabic morpheme when that morpheme constitutes a word is shortened when it is part of a
compound or inflected (see 2.2.6). Also, unstressed vowels are normally short except
occasionally utterance-finally.
Simple vowels are long (or half-long) in the following circumstances:
1. When stressed and preceding a homorganic cluster of nasal and voiced stop in a
disyllable. Examples include winda ‘pigweed’, nhanda ‘go’, mindji ‘clever’, wamba ‘big
girl’.
2. When stressed and preceded by a homorganic consonant, i.e. /i/ preceded by a
palatal consonant or /u/ preceded by /w/ or a velar, for example, yika ‘cold’, Yitjira
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
45
‘Urandangie’ (which may have primary stress on the second syllable), nyimi ‘fish’, nyina
‘sit’, wuma ‘his, her, its’, guma ‘sandhill’, nguna ‘lie down’. Perhaps /a/ preceded by a velar
consonant belongs in this category too; examples are nganga ‘eye’, garlu ‘over there’, ngatha
‘for me’, ngama ‘breast’. The lengthening effect is stronger when the initial consonant is a
glide, and is inhibited by a following consonant cluster (other than homorganic nasal-voiced
stop), tense stop or retroflex. Thus, for example it is fairly common for yinya ‘him, her, it’ to
have a half-long initial vowel (given that it is a pronoun and so is often not strongly stressed),
it is less common for yipa ‘you’, and djirta ‘seed’ has not been heard with vowel length.
Lengthening of initial /yi/ is less affected by length of the word than is lengthening in other
circumstances; thus a long initial vowel has even been heard in the six syllable compound
yindangu-ngamurnu ‘like yours’.
Words in categories 1 and 2 are more likely to have long vowels, as opposed to halflong or short vowels, and to retain length when inflected, than are words in the remaining
categories.
3. When followed by a voiced alveolar consonant, as in bala ‘over there’, yarra
‘creek’, bili ‘coolamon’, dala ‘spit’, barru ‘yellow ochre’, bina ‘chop (it)!’, dhurra
‘corroboree’. (The last contrasts with the word for ‘armpit’, pronounced [gu:ɾu] or [guuɾu],
which is written guwurru; it is longer and more consistently long.) Some of the most
common words to which this rule would be expected to apply are rarely or never heard with
length in the vowel because they are pronouns or function words which do not normally take
primary stress; examples are gana ‘not’, garra ‘too, again’, gurra ‘come on’, ngali ‘we (dual
inclusive)’ and nganu ‘we (plural exclusive)’. The vowel is not long when the following
consonant is not voiced, for example in wata ‘stone’, ngutu ‘nose’.
4. A few other words have been heard on occasions with long vowels, but it is not clear
whether this is idiosyncratic or there is a rule applying. Yanga ‘look at (it)!’ was heard
several times with a long vowel and several times without; phonologically similar words
nganga ‘eye’ and nhanga ‘get (it)!’ have been heard with long vowels but more often
without, and ganga ‘carry (it)!’ has not been heard with a long vowel at all. High vowels
before /ng/ are not long: nyingi ‘march fly’, wunga ‘for him/her/it’. A long /a/ has been heard
preceding /w/ on occasions, as in dhawa ‘man’ (heard far more often, however, with a short
vowel), mawu ‘witchetty grub’ and mawi ‘cave’. Of words with /iw/, yiwa ‘he, she, it’
(probably the most common word in the corpus) sometimes has a long vowel (under the
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
46
influence of its initial /y/) but liwa ‘clean’ never has. A word in which a long /u/ might have
been expected, wuwa ‘call (to him)!’ was heard as [uwwa].
5. A vowel is long (and often distorted) when it occurs utterance-final and is realised
with a high rising tone, implying, for example, urgency or exasperation. Thus
[n̪u:ɳumaɻa:u], nhuwurnu mara, ‘The honey is here (not over there where you are going)’.
6. A vowel may be lengthened in story-telling, presumably to indicate duration or
frequency of an event. There are only two examples (18i, 18xii).
7. A vowel may be long in utterance-final position.
The length of vowels before homorganic nasal-voiced stop clusters as opposed to the
absence of length before other clusters and geminates suggests that the former do not close
syllables and that these clusters occur in the syllable onset, and so they are regarded as unit
phonemes.
3.8 Inter-phonemic fluctuation.
3.8.1 Vowel harmony.
Vowel harmony is not a prominent feature of the Warluwarra language, and progressive
vowel harmony occurs regularly in only one morpheme, the purposive suffix /-yi ~-yu/. The
/yi/ form is found after /a/ or /i/, as in ngamayi ‘for mother’, yapiyi ‘to fall’, yangathiyi ‘to
see’, ngankanayi ‘for kangaroo’, while /yu/ follows /u/, e.g. wukuyu ‘for water’, nhuwurnuyu
‘for sugarbag’.
However, optional retrogressive vowel harmony by which /i/ becomes /u/ before a
suffix whose vowel is /u/ is not uncommon. For example, the allative (suffix -lu) of miki
‘ground’ was heard as mikulu and mikilu (actually [mikˌkilu]) in successive elicited
sentences. Other examples include malugu ‘these (mali) (ergative)’, burrghandulu ‘into the
middle (burrghandi)’, wirlwirlungurlu ‘from the leaves (wirlwirli)’ and yawirnungurlu ‘from
the sky (yawirni)’
3.8.2 Interdental-lamino-alveolar fluctuation.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
47
This has already been discussed in 2.2.3. Examples showing the distribution of the
absolutive suffix on proper nouns, -nha ~ -nya, are bapanha ‘girl’, ngamanha ‘mother’,
miminya ‘mother’s father’, makarrinya ‘spouse’. This behaviour does not apply to the
accusative suffix on pronouns, however: note ngalinha ‘we (dual incl)-acc’ as well as
wurrunha ‘you (plural)-acc’, yanunha ‘they (plural)-acc’, etc. There is, clearly, no
prohibition against the sequence /inha/, and, as another example, the suffix -nharrangu
regularly follows /i/, e.g. yangathi-nharrangu ‘without seeing’.
There is an alternation between /dh/ and /dj/ in certain verb suffixes; thus we have
yangadhiyi ‘to see’ but garridjiyi ‘to stand’, yangayidha ‘sees’ but garriyidja ‘stands’. but
also matharriyidha ‘hits’. The choice of /dh/ or /dj/ depends on the conjugation to which the
verb belongs, and the grouping is only partly phonological; in fact, certain verbs of different
conjugations share the same imperative form. Thus djirra is the imperative of the verb ‘to
spear’ (present tense djirrayidha) and of the verb ‘to be sick’ (present tense djirriyidja).
3.8.3 Lamino-dental / apico-alveolar alternation
/d/ alternates with /dh/ in certain complex verbal suffixes, such as -dhadha ~ -dhada, dhadhiyi ~ -dhadiyi, -djadha ~ -djada. This is unlikely to be due to a bias against laminal
stops in successive syllables, since the suffix -djidha and words such as dhathali and dhithidhithi are not known to undergo such a change.
3.8.4 Other examples of alternation or interchange
A very few examples of interchange between a nasal and the corresponding stop have
been noted, e.g. [juŋun̪a] for yungutha, [gatʲiɳa] for ganyirna. In the second case, the
speaker experienced no difficulty in recognising the word when, two years later, it was
played back to her.
Note also the alternation between -dha and -nha in the associated motion verb
formative, see 8.7; yangadhada but bakarrinhada. (Another allomorph of the same suffix is dja, as in garridjada.)
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
48
The word for ‘dead’ is marradhu in the speech of the two oldest of the four main
informants, and maradhu in the speech of the other two. The difference between /rr/ and /r/
is not always easy to detect, but the difference here is well attested.
3.8.5 Fluctuations in rapid speech.
Assimilation of a vowel or consonant to a following vowel or consonant is not
uncommon. See 3.8.1 for some examples. Others are:
[n̪andaˑmuˌkaɻali] for nhandami garali ‘come here kid’
[n̪aŋaɭipalidʲaɳmariˑdʲa] for nhangarli yipala djarnmarriyidja ‘What are you two
saying?’
[t ̪akujan̪andaɳa] for dhakuyu nhandarna ‘went where?’
[ŋaɳanʲijaŋaɳa] for ngarna yinya yangarna ‘I saw him’
There is perhaps some tendency to loss of one of two contiguous identical syllables
(haplology), as in the sentence: Garla ngarna nakarriyidha. ‘I’m cutting meat.’
but this is
probably better thought of as an instance of dropping of the final vowel of a word (ngarna in
this case) rather than loss of the whole syllable /rna/.
Loss of a syllable will be noticed in the fourth and sixth examples above; in both cases
/yi/ has been realised as zero. Loss of one of two identical syllables is exemplified by
[kaɭaˌŋaɳakaɹi:d̪a] for garla ngarna nakarriyidha ’I’m cutting meat’.
The word yiwa ‘he, she, it, that, there’, is very often affected by these processes —
assimilation and elision — and may be realised as [ji, ju, ja, wa] or even [wi]. In such cases
it is auditorily similar to a suffix on the preceding word, being completely unstressed.
(Compare the realisation of this word when stressed, 2.3.8.2)
3.8.6 Alteration of stem-final vowels.
It will be noticed in the morphology chapters that the addition of certain suffixes may
result in a change in or deletion of the final vowel of the stem of a word. For example, when
certain allophones of the possessive suffix are added to a noun whose stem-final vowel is /i/
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
49
or /u/, this vowel is lost, e.g. gunhugunhu becomes gunhugunhaa; dhuwarli may become
dhuwarliya, but may also become dhuwarla or dhuwarlaa. There is, therefore, an optional
rule that, if a suffix beginning with a vowel is added to a stem, the final vowel of the stem is
elided. Elision is in fact the normal situation, and is probably obligatory for certain
morphemes, for example the irrealis mood suffix on verbs, -imarla.
For the purpose of determination of vowel length and of stress, the morpheme boundary
should be assumed to follow the initial vowel of such a suffix. We assume that the suffix is a
disyllabic foot with an unsyllabified initial vowel, which replaces the final vowel of the stem.
Secondary stress falls on the first syllable of the affixed foot, and thus on the second vowel of
the suffix.
In the case of a suffix like the accusative giving rise to, for example, ngankanaanha
from ngankana we can assume that the suffix is -anha and there is no replacement of the
stem-final vowel, or that the suffix is -aanha and there is replacement. The latter course
seems to be preferable, since replacement does occur in some cases, such as that described in
the previous paragraph. The suffix can be assumed to be a disyllabic foot with the first vowel
unsyllabified; however, in this case the two /a/s would merge into a single surface syllable
and stress associated with the second of these /a/s would fall on this syllable.
3.9 Stress and intonation.
Suprasegmentals, or prosodic units, are features which, extending over a sequence of
phones, perform three functions:
(a) They give the hearer clues to how the sequence is to be segmented,
(b) They add to the semantic content of the sequence by helping to signal that, for
example, a certain utterance is a question, not a statement,
(c) They add to the “emotional” content of the sequence by conveying such feelings as
surprise, anger, disappointment and so on.
Suprasegmentals are realised in various combinations of four basic properties of
perceived sound: loudness, pitch, quality and length. All four of the variables are to some
extent interdependent, but the first, loudness, appears to be the main variable involved in the
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
50
fulfilment of function (a), while functions (b) and (c) are fulfilled mainly by means of a
pattern of varying pitch. The stress pattern on each word, or word prosodeme, fulfils
function (a), while the intonation pattern fulfils functions (b) and (c). Given sufficient data,
one could further subdivide intonation patterns into those performing function (b) — phrase,
clause, sentence and utterance prosodies — and those performing function (c) — contours.
However, in view of the limited data available for Warluwarra, this has not been attempted.
3.9.1 Representation of stress and intonation.
In this section, four levels of stress — sentence stress (representing the highest degree
of loudness — see below), primary and secondary stress, and tertiary or unstressed — must
be considered. The primary stress mark is duplicated to represent sentence stress, thus .
Unstressed syllables can remain unmarked. Tones will be specified as normal, high, very
high and low, abbreviated to n, h, vh and l.
Stress in Warluwarra is realised chiefly as loudness. However, there is no direct
correlation between degree of stress and degree of loudness. The word stress levels are
intended to represent the relative degrees of stress within a word, and have no reference to
relationships between words; e.g. a secondary stress on one word may correspond to a higher
degree of loudness than a primary stress on another word in the same utterance. This is
especially likely if the former word is one whose primary stressed syllable also carries a
sentence stress.
If, as is normal, a word has one syllable which is stressed more than the other syllables
in the word, this syllable is defined as having primary stress. A word may have only one
primary stress unless it is followed by a word in which all syllables are unstressed, in which
case there may be a second primary stress on the last syllable (see the examples to rule 4
below). Stressed syllables other than primary stressed syllables carry secondary stress.
Superimposed on the word stress patterns are sentence prosodies, which may be
realised mainly in loudness, but usually mainly in pitch. One effect of this superposition is
that one, or occasionally two, words in a sentence may carry a stress (normally on the
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
51
primary stressed syllable) which is noticeably higher than normal primary word stress. This
is referred to as sentence stress.
Because of the lack of natural speech in the recorded corpus, data on sentence prosodies
and contours are meagre; as a rule, only Queen produced utterances with sufficient fluency
that a sentence stress could be noted. Only a few remarks will therefore be made on these
aspects of the language.
3.9.2 Word stress.
A very common pattern of stresses in Australian languages is that in which there is a
primary stress on the first syllable of a word and secondary stress on the other odd-numbered
syllables except the last. Thus, considering, for illustration, only words containing CV-type
syllables, we have CVCV, CVCVCV, CVCVˌCVCV, CVCVˌCVCVCV and so on. A
common modification of this is for there to be a secondary stress on the first syllable of a
disyllabic or longer bound morpheme, thus negating any stress that may otherwise have
fallen on the syllable immediately preceding and forming a new beginning for a series of
alternating secondary stresses. A second common modification is for secondary stress to fall
on a (phonetically) long vowel or a diphthong (which may comprise two underlying
syllables). A third is for secondary stress to be lost, so that there may be a sequence of three
or even more unstressed syllables.
Examples of words in Warluwarra with stress patterns fitting these guidelines include:
[n̪anda] ‘go!’, [ŋat ̪a] ‘for me’, [d̪ajili] ‘spear’, [d̪uwana] ‘snake’, [n̪andaɳa] ‘went’,
[ŋankana] ‘kangaroo’, [waɾaˌwuɭa] ‘dog’, [dʲiwaˌlaga] ‘in the dark’, [mun̪aˌmun̪a] ‘soft’,
[laɾiˌdʲid̪a] ‘is going along listening’, [jaɾaˌmanaga] ‘on the horse’, [d̪uwanaˌmaɭu] ‘for fear
of the snake’, [gaɭaˌn̪aɾaŋu] ‘without meat’, [ŋutʲanaga:gu] ‘wife-ergative’ (contrast
[ŋutʲanagu], same meaning; the -gaa is a prosodic suffix used with the ergative suffix -gu;
see 6.4), [man̪alaˌman̪ala] ‘hot’, [ganad̪aˌɾa:] ‘hungry’ (‘hunger-having’),
[bilinʲtʲiˌn̪aɾaŋuɹiji] ‘will get tired’ (lit. ‘energy-without-become-will’), [buɭɭugudˌijidʲa] ‘is
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
52
getting hard’ (contrast [buɭɭuˌgudiɳa] ‘got hard’), [jaŋaid̪a] or [jaˌŋaid̪a] ‘sees, is looking’,
[jaŋaid̪ami] ‘is looking this way’.
This pattern, with or without any of its modifications, is probably an option for any
word in Warluwarra, but there is a number of factors which may condition variations, often
with more than one permissible stress pattern. The strongest conditioning factor is a tense
stop, especially if it has a geminate realization, following the first vowel of a word of three or
more syllables. A geminate stop is made up of an occlusion, followed by a period of silence,
followed by a release. It seems very natural that, being preceded by a period of silence, the
release and the following vowel should behave as the first, and thus the primary stressed,
syllable of a word, as long as at least one syllable follows to give the shape of at least a
minimal word (a foot) following the silence. Thus there is competition between the first and
second syllable for the primary stress. The result is that, where 1 refers to primary stress, 2 to
secondary and 3 to unstressed, we can have the patterns 133, 123, 113 and 213 (and/or
perhaps 313; these are difficult to distinguish). There are tokens of ngathangu ‘my’
transcribed with all of these possible patterns (although 113 is rare).
This tendency is counteracted by the tendency, noted above, for a phonetically long
vowel or diphthong to have some stress, and to a lesser extent for /a/ to be stressed in
preference to /i/ or /u/. Thus we have sets like:
muka
[mukˑɐ]
‘good’
mukamarna
[mukkamaɳɐ]
‘made, repaired’
mukamayidha
[mukˑɐˌmɛiðɐ]
‘making, repairing’
Here the first word has stress on the initial syllable, since primary stress cannot
normally fall on the final syllable of a word. The second word has primary stress on the
second syllable; it could have had some stress on both of the first two syllables, although
stresses on successive syllables are not one of the more favoured patterns. I have transcribed
it on occasion with primary stress on the first syllable, but on the other hand when I have
pronounced it that way to Queen she has repeated it with second syllable stress. The third
word has five syllables underlyingly, but the third and fourth have coalesced into a single
syllable containing a diphthong (from the sequence /ayi/) and this attracts stress; in
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
53
preference to having stress on the second and third syllables, the primary stress has remained
on the initial syllable and secondary stress is on the third.
Some other examples:
bakarna
[bakkaɳɐ]
‘jumped’
bakarriyidha
[bakɐˌɾi:ðɐ]
‘jumping’
makarrinya
[makkaɾinʲɐ]
‘sister-in-law (absolutive)’
makarriba
[makkaɾibɐ]
‘sister-in-law’s’
gapirrimayidha
[gappiɾiˌmɛiðɐ]
‘taking (with you)’
gukuwaba
[gukkᵘˌwaˑbaˑ]
‘poor fellow’s’
natayidha
[ɳatɛiðå]
‘going’
yaparna
[jæppaɳɐ]
‘fell’
yapimarla
[jæpiˌmaɭɐ]
‘might fall’
yapalaparriyidha
[jæpɐˌlapaˌɾi:ða]
‘stumbling’
Other situations favouring movement of stress to the second syllable are:
(a) a glide in the second consonant position;
(b) /rr/ in the third consonant position;
(c) a consonant cluster with a stop in second position (other than a cluster of nasal plus
homorganic voiced stop) in the second consonant position; and
(d) a secondary stress in the fourth syllable.
Type (a) would be related to the widespread tendency for a sequence CVG, where G is
a glide, to become a consonant cluster CG. (Similarly, CVL, where L is a liquid, is subject to
the same tendency in many languages.) In Warluwarra this occurs only with the semi-vowels
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
54
/w/ and /y/ and not with the other glides /r/ and /gh/, and is strongest in the case of the
sequence /Cuw/ where C is peripheral. Extreme examples, in which the first vowel is not
heard at all, are [ŋwɒˑgəməɳa] for nguwagamarna ‘hid (transitive, past tense)’, [ŋwaŋwa]
for the place name Nguwanguwa ‘Walgra’ and [wwaˑt ̪aɳa] for wuwadharna ‘called while
going along’. A less extreme example is [guwaiða] for guwayidha ‘cooking’. Another
widespread tendency is for a sequence /Cuy/, where C again is peripheral, to be pronounced
as /Cwi/; a Warluwarra example is [ŋwi:nda] for nguyinda ‘east’. The glide /y/ does not
have such a great effect on stress placement, although the pronunciation [jˑæɭami] for
yiyarlami (more commonly [ijaɭami]) ‘can send this way’ is comparable with the
pronunciation of wuwadharna given above. Note also [biˌjæða] for biyadha ‘cloud’ and
[wɛ:paɳa] wiyabarna ‘called’ (and for the imperative form of this verb [wjæppa] and
[wijæpa]). Examples with non-peripheral initial consonants include [d̪uwana] and
[d̪uˌwana] dhuwana ‘snake’ and [dʲiwala] djuwala ~ djiwala ‘night’.
Type (b). The trill is a sound which, for some speakers at least (including this linguist
and the main informant in the study of this language), requires a lot of energy in its
production and it is natural that this energy might be heard as stress in the preceding syllable
or the syllable which the /rr/ is closing. However, some of the examples are from other
speakers. Examples include: [gulaɾiga] ‘heron’, [wuɾɰaɾaɳa] ‘ran’, [ngulaɾiɳami] ‘came
back here’ and [ŋaŋaɾidʲadami] ‘is coming up’. The last example could also have secondary
stress on the fourth syllable, but it was not heard on the particular token that was chosen for
the example.
Type (c). This is perhaps analogous to the geminate stop situation, although there is no
similar period of silence. It is a much weaker tendency. Some examples of this type, such as
ngunparrilaga [ŋunpaɾilaga] ‘day after tomorrow’ and warlkarra-dhukiyi [waɭˌkaɾaˌd̪ukiji]
‘to sharpen’, conform also to type (b) and may be influenced more by the /rr/ than any other
factor. Other examples, such as [maɾˌkana] ‘hand’ or [ŋankanaji] ‘for kangaroo’, are
unusual pronunciations.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
55
Type (d). This may have arisen by analogy with other situations in which stress is
moved to the second syllable coupled with a desire to have the primary stress just two
syllables away from the secondary stress, which in turn is placed on the first syllable of a
disyllabic bound morpheme or a series of morphemes. Examples are [gaɻaliˌwali] garaliwali
‘children’ (and its accusative form [gaɻaliˌwalin̪a]), [muŋalluˌmaɳa] mungallumarna
‘missed’ and [jindaŋuˌɭagu] yindangurlagu ‘your-ergative’.
There is another type of stress movement that has been noted at times in fluent speech,
and this is movement of stress forward from a clitic or a function word that does not normally
carry a strong stress to the final syllable of the preceding word. Sometimes such a stress is
quite strong. Examples are:
[gunabəɖa: n̪unda ŋaɳa] gunaburtaa nhunda ngarna ‘I’m full now’ (lit. ‘stomachhaving now I’)
[bitʲiɳidʲi ŋaɳa] bitjirnidji ngarna ‘I was swimming’
[wan̪undajaka] ‘right!’ (wanhunda an emphatic particle and yaka an emphatic clitic)
[dʲuwalaˌga jalu] djuwalaga yalu ‘early in the morning’ (lit. ‘still in the dark’, ‘dark-loc
still’).
A secondary stress has been noted at times on a penultimate syllable even though it is
the third syllable after the preceding stressed syllable, perhaps only when a morpheme
boundary follows. An example is an alternative pronunciation of yarramanaga ‘on a horse’.
Finally, a perplexing word, to me and earlier to Hale (pers.comm.1969), was yirlwanga
‘to hit with a missile’, which usually has primary stress on the second syllable. This does not
seem to be associated with either a /yi/ initial syllable or a following lateral-glide cluster;
neither of these conditions vowel shift in other words except rarely in the very common word
yindangu ‘your’.
Characterising these tendencies as rules, I can make a few remarks about the
prosodemes generated. In these rules V is any vowel phoneme, ´V a primary stressed vowel
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
56
and `V a secondary stressed vowel, C any consonant and T any tense stop, parentheses ()
enclose optional segments or boundaries, and {} a set of alternatives, a hyphen - represents a
morpheme boundary and # a word boundary, @ means ‘at a particular point of articulation’
and ~@ means ‘not at the point of articulation denoted by @’, and a comma means ‘or’. —>
means ‘is pronounced as’ and / means ‘in the environment’; — shows where in that
environment the vowel in question is. Lower numbered rules, if they take effect, take
precedence over higher numbered rules. The “rules” are:
1(optional)
V —> ´V /{#, -}CV{T, y, w, r, rr, C@C~@}—(-)CV
2(optional)
V —> `V/VC(C)VC(C)—-C(C)V#
3(optional)
V —> ´V /#CVC(C)—C(C)V(-)C(C)`V
4(optional)
V —> V / #C—C(C)V#
5(optional)
V —> {´V,`V} /—#CVC(C)V#
6
V —> ´V /#C—C(C){`V,V}
7(optional)
V —> `V /#C—C(C)´VC(C)V#
8
V —> `V /-C—C(C)V
9(optional)
a —> à /—a
10(optional)
V —> `V /´VC(C)—yi
11(optional)
a —> a /{´V, `V}(-)C(C)a—(-)CV
12(optional)
i —> i /{´V, `V}(-)C(C)ay—(-)CV
13
V —> `V /V(-)C(C)—(-)C(C)V
(can apply to only certain words)
A word can have only one primary stress, unless the following word is unstressed (rules
3, 4 and 5). Rule 3 can generate the pattern in which the word has no primary stress only in a
limited number of words: certain adverbs such as nhunda and watji, emphatic particles yaka
and gunhu, some pronouns and, occasionally, the vocative form of a proper noun, such as
bapa. In all other cases, two-syllable words have the stress on the first syllable. Threesyllable words may, however, have any of the prosodemes of the types C´VCVCV, e.g.
dhathama, galangu, yarrarra), C´VC`VCV, (e.g. gapulla, dhathali, ngathala) or CVC´VCV,
(e.g. bakarru, kapirri, wumarlu). This applies equally to words containing syllable types
other than CV. Using 1 to represent a primary stressed syllable, 2 a secondary stressed
syllable, and 3 an unstressed syllable, we can abbreviate these three prosodemes to 133, 123
and 313. This system will be used below. Longer words build on these patterns; the degree
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
57
of variability is decreased by the restrictions which prevent more than two unstressed
syllables from occurring in sequence (except that there can be three if two of them are
realised together as a single ‘phonetic syllable’ — to use Pike’s term — with a long vowel),
stressed syllables from occurring together (except that the first two syllables of certain three
syllable words may be stressed) and stress from falling on the last syllable of a word (unless
an unstressed word follows). Thus a four syllable word can only have the patterns 1323, or
3133, a five syllable word 13233, 13323 or 31323, and a six syllable word 133233, 132323,
313233, or 313323. (Patterns permissible only when one of the conditions mentioned in
parentheses in the previous sentence applies are not included.) Not all of these patterns are
available for any one word of the appropriate length; only the rules can show which are
applicable in any case. Thus dhayili-nharrangu 'without a spear' can be only [t ̪ajiliˌn̪araŋu],
[t ̪ajilin̪aˌraŋu] or [t ̪ajiliˌn̪araŋu]; the first pattern, 133233, is permitted by rules 5 and 7, the
second, 313323, by rule 1 and the third, 313233, by rules 1 and 7. No other rules apply.
Some other examples of the operation of the rules follow:
yarrarra
133 by rule 6.
warrawurla
1323 by rules 6 and 13.
bakarruba
3133 by rule 1
yarramanaga
13233 by rules 6 and 13 or 13323 by rules 2 and 6.
ngutjanagaagu
133233 by rules 6 and 8.
larridjidha
1323 by rules 6 and 13, or 3133 by rule 1.
bilintji-nharrangarriyi
13323323 by rules 2, 6 and 8.
ngangarridjarna
13323 by rules 2 and 6, 13233 by rules 6 and 13, or 31323 by rules
2 and 3. (All three of these realizations are attested.)
The rules appear to have been overruled in a few cases in which the stresses on an
utterance appear to follow a rhythmical pattern with no reference to the normal word
prosodemes, and therefore providing no clues to the segmentation of the utterance. Examples
are:
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
58
[kala:kajaluŋaɳan̪andaɳamaˌɻalu]
Galaaga
yalu ngarna nhandarna maralu.
long.ago-EMPHyet 1sg
go-PAST here-ALLA
I came here a long time ago.
[nʲinaˌipalan̪aŋaɭiˌpalakuliˌrinʲa]
Nyina yipala nhangarli yipala gulirrinya?
sit:IMP 2du
what
2du
fight-COMP
Sit down, you two; what are you fighting for?
[puɭ:uˌkutiɳan̪əˌndɛja]
Burlugudirna
nhunda yiwa.
hard-INCH-PAST
nowthat
It’s hard now.
In the first example, the only reason for the stress on the final syllable of nhandarna
seems to be a desire to have a stress on every second syllable of the utterance. The stress
distribution and elision in the second example probably are for the same reason; the normal
pattern on yipala is 133 or 313, and on nhangarli 133. The stress pattern in the third example
can possibly be explained by two applications of rule 4 followed by elision of a syllable from
the last word; the final effect is definitely rhythmical.
3.9.3 Sentence stress.
Sentence stress is quite unpredictable and depends on what item or items in an utterance
the speaker wishes to emphasise. Some examples are given. Notation, as in the previous
examples, is broad phonetic; thus a non-phonemic item such as vowel length, which is of
significance in the realisation of stress, is shown. Sentence stress is shown with two primary
stress markers as in the first syllable of the first example below.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
59
[ma:tʲu jiwa waɾaˌwuɭajiˌndaŋu]
’Your dog is no good.’
Matju yiwa warrawurla yindangu.
bad
that dog
2sg:GEN
[mukˑa jaka jiwa]
’He’s good, all right.’
Muka yaka
yiwa.
good emphatic he
[kana ipa n̪andaɳa ŋat ̪aŋuˌɭalu ŋuriɳiˌla:lu]
Gana yipa nhandarna ngathangurlalu ngurrirnilaalu.
not 2sg go-PAST 1sg:GEN-ALLA camp-ALLA
’You didn’t come to my camp.’ (In the actual context the English equivalent was ‘If you had
come to my camp ---’.)
[n̪aŋaˌɭi:wa pirɰi]
’Well, fancy that!’ (or ‘Is that so?’)
Nhangarli yiwa birrghi?
what
that true
In some cases a sentence stress is confined to the normal primary stressed syllable of a
word (e.g. in ngathangurlalu), while in other cases it is spread over most or all of the word;
thus in ngurrirnilaalu the second and third syllables are both stressed similarly to syllables
carrying secondary stress in other parts of the utterance and the fourth syllable carries the
equivalent of a primary stress. So, given a different set of conventions for the representation
of stress, I could have written [ŋuˌriˌɳiɭa:lu]; similarly I could have written [pirˌɰi] instead
of [pirɰi].
Sentence stress, unlike word stress, is often realised in a significant rise in pitch instead
of, or as well as, in loudness. Further reference will be made to this in the next sub-section.
3.9.4 Intonation.
Intonation patterns observed in Warluwarra can be grouped into four categories:
(i) declarative,
(ii) interrogative,
(iii) imperative,
(iv) exclamatory contours.
The declarative prosodes are realisations of the declarative prosodeme, and consist of a
steady normal pitch throughout the utterance except in the last part, normally the last syllable,
when there is a fall in pitch. Superimposed on this are the small pitch variations due to the
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
60
pitch component of the word prosodes and, often, larger variations due to the sentence stress.
The latter may lift the pitch into the high range.
A declarative prosode may be spread over a whole utterance, which may include a short
pause between two sentences, i.e. there is no fall in pitch (or loudness) at the end of the first
sentence. Alternatively, each sentence of an utterance may have a separate realisation of the
prosodeme.
Some examples of declarative prosodes follow. The notation is broad phonetic and [/]
represents a pause. Words will be bounded by spaces in the phonetic notation, except when
two words are particularly closely bound together, as [n̪aŋaˌɭi:pa], nhangarli yipa ‘what you’.
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
61
n
l
ˈkaɻali ŋat ̪aŋu jiwa
Garali ngathangu yiwa.
child 1sg:GEN
3sg
‘He’s my boy.’
vh
h
n
n
n
l
l
kaʈu ŋaɳa inʲa ŋutaɳa / jiwa kimaɳa ˌjaka
Kartu ngarna yinya
ngutarna, yiwa gimarna
yaka.
fruit 1sg
3sg:ACC give-PAST 3sg let go-PAST emphatic
’I gave him some fruit but he wouldn’t have it.’
h
n
n
l
iwa ɭaɾitʲa: ˌjiˑna
Yiwa larridjaa yina.
3sg hear-POT 2sg:ACC
‘He might hear you.’
h
n
l
kana ŋaɳa ˌinʲa ŋuti: kaɭa
Gana ngarna yinya
ngutiyi
garla.
not 1sg
3sg:ACC give-PURP meat
’I’m not going to give him any meat.’
h
n
h
n
l
t ̪aŋgumat ̪akaɾi janu ŋankaˌna:n̪a
Dhanggumadhagarri yanu ngankanaanha.
roast-HAB-PAST
3pl kangaroo-ACC
’It was kangaroos they roasted.’
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
62
n
l
wula kuliɾitʲi maɻaka waɾawuɭa / indaŋu ŋat ̪aŋwaɻa kaɾa
Wula gulirridji maraga warrawurla, yindangu ngathanguwara garra.
3du fight-GER here-LOC dog
2sg:GEN 1sg:GEN-PROP
too
’Your dog and my dog were fighting.’
The interrogative prosodeme is realised as a steady normal pitch with a rise over the last
few syllables and a drop towards normal pitch in the last syllable. A high pitch, realisation of
a sentence stress, is often superimposed, and is frequently on a question word, such as
nhangarli? ‘what?’ or ngaragu? ‘who?’. The interrogative prosodeme may not be realised
on the interrogative clause or sentence, but on another portion of the utterance; see the third
example below.
h
n
n
waji ŋaɳa n̪andimaɭa
Wayi
ngarna nhandimarla?
question 1sg
‘Can I come?’
go-IRR
h
n
n
ɳatiɳaj
Natirna
yiwa?
‘What happened?’
what.happen-PAST there
(Interpretation doubtful; it may be just natirna with the final vowel emphasised. It was
immediately followed by Natirna wurru? ‘What did you(plu) do?’)
h
n
l
n̪aŋaˌɭi:pa jaŋait ̪a ŋan̪a / papa
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
63
Nhangarli yipa yangayidha nganha, bapa?
what
2sg look-PRES
1sg:ACC girl
’What are you looking at me for, girl?’
h
n
n
n̪anŋuka ipa inʲa pumaˌt ̪aka mukamaɳa
Nhannguga
yipa yinya
bumathaga mukamarna?
how many-LOC 2sg 3sg:ACC day-LOC
make-PAST
’How many days did you take to make it?’
The imperative prosodeme is realised as an initial high pitch falling to normal. It may
be repeated several times in a long imperative utterance, and may be preceded by a short
period of normal pitch.
h
n
n
ŋuwakama
Nguwagama!
h
‘Hide it!’
h
n
n
kana ŋan̪a katnama
Gana nganha ganama!
‘Don’t push me!’
not 1sg:ACC push:IMP
vh
vh
h
h
kuɾa n̪anda
Gurra nhanda!
‘Go on, go!’
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/
64
The exclamatory contour is characterised by a rising pitch, usually to a very high level,
on the last syllables. This is used to convey urgency, surprise, sympathy, agitation, etc.
Distortion of the final vowel is common.
vh
h
n
kukuwan̪aˑu
Gukuwanha!
‘Poor fellow!’
vh
h
n
kul:ika ŋamau
Gulika ngama!
‘Quickly, mummy!’
h
n
wan̪unda
Wanhunda!
‘That’s enough!’
An alternative realisation of the exclamatory contour is as a uniformly high pitch.
h
kaɻalija japa:
Garali yiwa yapaa!
child
that fall-POT
’Be careful, the baby might fall!’
Warluwarra Grammar | Gavan Breen | 3. Pronunciation
http://central.batchelor.edu.au/warluwarra-grammar-gavan-breen/