What Speakers of Australian Aboriginal Languages Do with Their Velums and Why: the Phonetics of the Nasal/Oral Contrast
Total Page：16
File Type：pdf, Size：1020Kb
WHAT SPEAKERS OF AUSTRALIAN ABORIGINAL LANGUAGES DO WITH THEIR VELUMS AND WHY: THE PHONETICS OF THE NASAL/ORAL CONTRAST Andrew Butcher Flinders University, Adelaide, Australia ABSTRACT (2) Nukunu [2]: Pre-stoppednasals, either as phonemes,or as major allophonesof /‘pana/ [p&ng ] ‘he,she, it ’ /‘pulana./ [pulnnn] ‘two ’ nasal consonants, are a well-documented feature of certain /‘kaJ;a/ [kti’n] ‘rock ’ /‘ka.rap/ [keJ=pn] ‘arrogant ’ Australian languages- mostly concentratedin the southern and Pre-stopping in these languages is often optional, however. central areas of the continent. A study of nasality in connected Cognatesin related languagesoutside the pre-stopping area - and speechin a wide variety of Australian languages,with and without often in place names and archaic song language within it - phonologisedpre-stopping, showsthe sametendency is even more typically retain the plain long nasal. widespread at a phonetic level. This kind of perseverativevelic closure, apparently well established in Australia, is rare in the world context. This paper considersthe putative phonetic origins of this phenomenonin terms of unusual parameter settings for velopharyngeal control: a higher degree of stiffness for the opening gesture and a closer resting position. A possible perceptualexplanation for thesesettings may be found in the need to preserve clear spectral cues to the place of articulation of postvocalic consonantsin languageswith up to seven places of articulation for nasals. 1. PIW-STOPPING OF NASALS Pre-stopped nasals have been found to occur in a number of Australian languages. The geographical distribution of these languagesleads to the conclusionthat this is an area1phenomenon. With a few exceptions,such as Kunjen in the Cape York Peninsula and Gidabal and Yugarabul around the Queensland/NewSouth Wales border, they are concentrated in central and southern Australia, amongst the Arandic languages and the ‘Lake Eyre languages’, extending south into the Yura group. In the case of most of the Arandic dialects, these sequencescan be analysedas Figure 1: Spectrogramand waveform of the Gupapuyr3t.tword unitary phonemes,and are clearly in contrast with both oral stops and plain nasals, with which they are in parallel distribution /‘cinaka/ (‘inside; underneath’). The epentheticoral ‘pre-stop’ precedingstressed vowels: begins at 150 ms and lasts about 25 ms. (1) Eastern/CentralArremte [ 11: There is thus very strong evidencethat these sounds arise /ap’ax/ ‘gum tree ’ /aImal/ ‘nest’ /aiPmal/ ‘camp’ from an original lengthened or geminated nasal 131,through a processof rightward ‘oral spreading’,whereby the velic closure of /a’kaM ‘cut ’ /a’ gap/ ‘crow 9 /aikg&/ ‘carried ’ the precedingvowel is carried into the articulatory closure of the /a’@n/ ‘bursting ’ /a’nam/ ‘staying ’ /a’*nam/ ‘yumstick ’ original postvocalic nasal consonant. The conditions governing /a’taW ‘built ’ /a’r&/ tfor a stick’ /alhoW ‘cried’ both the diachronic developmentof pre-stoppednasal phonemes, /a’taM ‘ground up ‘/a’naM ‘watered’ /u’lnak/ ‘bit ’ (v) and the synchronic occurrence of pre-stopped allophones are /a&no@ ‘pno‘s. fa. ’ /alfiana@n Tather ’ /a’“Jlanaqa/n ‘fell ’ strikingly similar across all of these languages.The first of these conditions is that the nasal segmentmust be phonetically long. In the southernlanguages, the pre-stoppednasals are either in This, in turn normally means it should be adjacent to a stressed complementary distribution or in free variation with the plain short vowel. In Alyawarr, for instance,they are restricted to the nasals.In this casethe pre-stoppedallophones occur after the first fast consonantalposition of a word - i.e. the onset of the syllable (stressed)vowel; the plain nasalsoccur elsewhere. bearing the primary stress. Thus a reduplicated form such as /a’pm%mwam/ - [r+‘mwtz:mwtzm]( ‘rubbishy ‘), from /alpmw/ - page 479 ICPhS99 San Francisco [$mwz] ( ‘bad’) has pre-stopping only in the fast element[4]. In from a following homorganic oral stop. This appearsto be just as the Southern languages there are also cases of variable pre- common in Australian indigenous languagesas it is in Australian stopping of nasals even in stressed syllables in long (i.e. four- and American English: syllable plus) words - e.g. Arabana-Wan&anguru: /‘kini-taka/ - (4) Burarra: [gIdnrtt?kn]or [gn-ntsk~] ( ‘scorpion ‘) [5]. This too might well be lrJuwu’mar&a./‘ we get it’ due k thereduction in length of segments in this context. Eg~m=?Kw-+ hv=w.Jal Similarly, pre-stopping has not normally occurred in nasals Pitiantiatiara: following a long vowel, presumably since these too would be /‘ganampa’mai/ ‘our fooa [gt?nt”nbmgr] -+ [~~nmtn~rnm] phonetically of insufficient length - e.g. Nukunu: &a./ - &rdno] * [3pImnm~I] ( ‘fbot ‘), but /‘wi:qa/ - [WING] ( ‘white chalk’) [2]. The second condition applying to pre-stopping (both synchronic and Thus as regards perseverative assimilation of nasality, diachronic) in all of these languagesis that the processis blocked Australian languagesappear to behave much as languagesfrom by the presenceof a precedingnasal in the sameword: elsewhere in the world. The more common form of nasal assimilation found in the world’s languages,however, is a right-to- (3) Adnyamathanha: left or anticipatory assimilation [ 81, whereby the velum lowers during a preceding oral segment(usually a vowel) in preparation /‘bu!na./ [~-C&Z] ‘scent’ ‘who’ bw hwl for an upcoming nasal consonant.This is generally regardedas a /‘acti [tips] ‘rock ’ /‘IJUJB/ [rppz] ‘bindweed’ universal process,but it seemsthat Australian languageshave a PI particular aversion to anticipatory coarticulation of nasality. Both The (phonetic) pre-stopping of (phonologically) plain nasals auditorily and through the visual inspection of spectrogramsit is has also been noted in Eastern Arremte [7], where, as we have quite apparent that in vowel + nasal sequencesspeakers avoid seen, the plain nasals are already in contrast with a full set of lowering the velum until the latest possible instant. As we have (phonologically) pre-stopped nasals. In fact the pre-stopping of seenin the word-internal case,the lowering of the velum is often nasal consonantsis a widespreadphenomenon at the synchronic ‘left too late’, with the result that oral@ perseveratesinto the phonetic level in many Australian languages,and is certainly not nasal consonant. The ultimate extension of this phonetic oral restricted to central Australia. Figure 1 shows an example of spreading,is reachedin rapid casualspeech when the avoidanceof phonetic pre-stopping in Gupapumu, a language not closely prematurevelum lowering can lead to no lowering at all. In other related to any of the languagesso far discussed,spoken in north- words, there is clearly a synchronic connectedspeech process of easternArnhem Land, some 1,000 km further north than Arremte. perseverative denasalisation operating in some Australian The word is /‘cinaka/ (‘inside ‘, ‘underneath ‘), here pronounced languages: [lCrdnEkhv].As already surmised,the phenomenonappears to arise through an asynchronyin the formation of the articulatory closure (6) Warlpiri and the lowering of the velum. The latter appearsto be delayed /‘janira ‘rjajir@ ‘I amjmt going’ Ijmmqmq~] -3 @mragxxrp] relative to the former - the velo-pharyngealport is not sufficiently Kunbarlanq open for the production of a nasal consonant until after the /ki’ttaqipuni/ ‘they would name’ [g~tnr@m]-+[g~d~wum] formation of the articulatory closure - and an epenthetic(but often inaudible) oral stop closure results. The oral ‘pre-stop’ in the 3. THEORETICAL IMPLICATIONS Gupapuygu example illustrated above is barely audible to the I have elsewhere described the phonologies of Australian unalerted ear. Once again the process appearsto be particularly languagesas having ‘long thin’ systemsof contrasts, in terms of associatedwith phonetically long nasals, and nasal lengtheningin the traditional way of setting out thesesystems on the printed page turn appearsto be associatedwith stressed syllables. And once [9] - i.e. they are relatively restricted in terms of manner-of- again the processis inhibited by the presenceof a precedingnasal articulation distinctions (traditionally displayed in the vertical in the word. dimension) and rich in place-of-articulation distinctions (traditionally represented horizontally). This means that 2. DENASALISATION AS A CONNECTED SPEECH oppositions within Australian phonological systems are heavily PHENOMENON reliant on systematicdifferences in formant transition patterns at The velum is generally regarded as a relatively sluggish mover, vowel-consonant boundaries. Furthermore, the relative lack of comparedwith some other articulators. Thus in connectedspeech mannerdistinctions is entirely within the domain of the obstruents: in many languagesof the world nasality may not be commed to a the majority of Australian languageshave a single seriesof these, single segmentin the way that it is assumedto be in the citation with no [voiced] N [voiceless] contrast and no [stop] - [fricative] form, as the velum is lowered before the articulation of that contrast. On the other hand, theselanguages have as rich a system segment begins, or fails to close in time to coincide with its of sonorant contrasts as any language in the world - and richer completion. A common manifestation of this timing shift between than most. This means that these systems have precisely the oral and velic articulation is often referred to ‘left-to-right’ or opposite proportion of obstruentsto sonorantsto