The Phonemic Structure of Lower Grand Valley Dani

Part of the Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde book series (VKIV)


Skewness implies internal configuration as well as comparison. In Chapter II emphasis was placed upon comparison, and section 2.7 focused attention on the phonemic pattern of lower Grand Valley Dani as askew with the patterns of proto-Dani and other extant dialects. This chapter is a detailed description of that one atypical structure with the perspective afforded by the preceding survey. An examination of the distribution of allophones as members of phonemes precedes treatment of the distribution of the phonemes themselves within larger units.


Phonological Word Final Syllable Front Vowel Plural Subject Glottal Stop 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Pike’s treatment of these levels of phenomena has been germinal for this analysis, but the term macrophonemic seems preferable to his hyperpho-nemic. A macrophonemic unit is part of a contrastive system of units larger than the segmental phoneme, although terminals of such units may be no longer phonetically than a single segment. See Kenneth L. Pike, Language in Relation to a Unified Theory of the Structure of Human Behavior (Glendale, 1955), Part II, pp. 41 ff.Google Scholar
  2. 2.
    For one problem area relative to syllable terminal, see the note on section 3.1215 relative to consonantal vocoids.Google Scholar
  3. 3.
    Symbolization of the macrophonemic units is omitted until those units are described, unless they are clearly relevant to the discussion. Symbols used for phonetic transcription are those commonly used by American linguists, except that [r] is gingival backward flap; [t] is a lamino-gingival stop; [b] and [g] are voiced bilabial and back velar fricatives, respectively; [k] is a back velar stop; [k̃]; and [g̃] are one segment lamino-palatal voiceless stop and voiced fricative, respectively; a stop symbol raised above the line, e.g. [P], represents an unreleased stop; [n] is a voiced velar nasal; [j] is a high close front unrounded pre-syllabic vocoid; [‘] is glottal stop; [3] is lowered mid open front unrounded vocoid; [c] is mid open back rounded vocoid. Vocoid symbols written above the line represent vocoid off glide quality when following another vocoid symbol, but vocoid timbre or articulatory position when following non-vocoid symbols. Thus [a] marks a low fronted central vocoid with offglide toward [i]; [1°] indicates a lateral with [o] timbre. Following velar consonants, symbols for round vocoids written above the line indicate lip shape for production of one sort of labialized velars; [k°] indicates a voiceless velar stop with simultaneous labialization with lip shape as for [o]. Sequences of periods [] indicate any phonation in the same contour. A raised dot following a symbol indicates a half mora of added phonetic length: [p°]; colon following a symbol indicates a full mora of added phonetic length: [p:]• Grave accent marks phonetic stress: [à]. Subscript apostrophe indicates syllabicity: [n]. [V] stands for any syllabic vocoid phone; [C] stands for any consonant phone. Comma between bracketed forms indicates free variation of those forms. Underlined material between brackets is comment on the forms with which it occurs.Google Scholar
  4. 4.
    See the note on /kw/ on page 30 for discussion of the contrast between /k/ and /Kw/.Google Scholar
  5. 5.
    This interpretation is critical to the analysis of LGV stops. See section 3.12113 below for a discussion of the point.Google Scholar
  6. 6.
    The form /akwkwe/ ‘its opening,’ might also be interpreted as /akkwe/, since only /k/, not /kw/, occurs unit finally at most borders, and initial members of clusters are elsewhere phonemes which may occur unit finally. The sequence / kw-s / has been recorded at nucleus terminal, and this is considered evidence for the interpretation in the text. For practical orthography, as discussed in Chapter IV, other spellings are better.Google Scholar
  7. 7.
    One could interpret the data as including a cluster of glottal stop and labial stop, but this interpretation is rejected to avoid positing a unique three member cluster of glottal stop, stop and liquid; three member Clusters do not occur otherwise. The unit cannot be interpreted as an allophone of glottal stop, for that phoneme occurs without labial closure in this environment!! /a’la/[a’la] ‘inside it, ‘The complex unit is therefore interpreted as an allophone of /p/ even though it includes the distinctive feature of glottal stop. Parallel units do not occur for /t/ or /k/. Another case of overlap is described for /u/, for which a complex allophone including bilabial closure occurs; see section 3.12211 below.Google Scholar
  8. 8.
    These short vocoids contrast with full vowels. Compare the following with the examples given above: /nappot-ekken/ [napèr3k3n] ‘my vertebrae1’; /nappot-opa/ [napèrcba] ‘on my back.’Google Scholar
  9. 9.
    Contrast note 7 about a complex unit interpreted not to include /’/.Google Scholar
  10. 10.
    For discussion of the contrast between this subphonemic labialization and /kw/, see the last paragraph of this numbered section, describing /kw/.Google Scholar
  11. 11.
    See section 3.12211 below for discussion of the allophone of /u/.Google Scholar
  12. 12.
    The occurrence of /k/ unit finally after /Vi/ and /Vu/, when those two part syllabics are not followed by another consonant, is automatic in LGV. Some of these forms have correspondences with final /k/, some with final vowel, elsewhere. Sequence /aik/ is most widespread. Compare BV /u/, LGV /ouk/ ‘sickness’; BV /pi/, LGV /heik/ ‘oak’; BV, LGV /aik/ ‘his tooth.’Google Scholar
  13. 13.
    The phoneme /kw/ occurs only word initially and medially, only before vowels /e/ and /i/, and only following /a/ word medially.Google Scholar
  14. 14.
    The labialization of /kw/ contrasts in quality, and the labialized unit in distribution, with labialization of /k/ after rounded rowels. The first fact to be noted is that /k/ and /kw/ contrast. Although all items in which /kw/ is heard also occur with /k/, many items have only /k/, never /kw/, in the same or analogous environments. Compare with the illustrations in the text the following: /ket/ [k3t] ‘new’; /kem/ [k3m] ‘variety of grass’; /ake/ [ag3] ‘its tail.’ Further, this contrast cannot be assigned to contrast of /k/ with a cluster of /k/ with /w/ or a sequence of /k/ and a rounded vowel. The interpretation of [kw] as /k/ plus /w/ is rejected because it posits the occurrence of a consonant cluster following silence or plus contour terminal, where no consonant clusters are otherwise observed to occur. Interpretation of [kw] as /k/ plus /u/ or /v/ or /o/ is rejected as not squaring with the phonetic data, for these vowels are always fully syllabic preceding another vowel: /luppuet/ [lupu3t] ‘ingressive whistle’; /sue/ [st3] ‘bird’; /noe/ [no3] ‘my older sibling.’ Informants can produce without hesitation the hypothetical parallel forms: */kue/ [ku3]; */kue/ [ku3]; /koe/ [ko3]. The phonetic item [kw] is therefore interpreted as /kw/, a unit phoneme in contrast with /k/. It is further observed that allophones of /kw/ are in complementary distribution with the labialized velar allophones of /k/ following rounded vowels. Assignment of those latter allophones to /kw/ is rejected for the following reasons: (l) The allophones in question are also in complementary distribution with non-labialized allophones of /k/, which are in contrast, as discussed above, with /kw/ (2) The labialization of those allophones of /k/ following rounded vowels is easily interpreted as environmentally conditioned, but the phonetically different labialization of allophones here assigned to /kw/ cannot be so interpreted; note that /kw/ may follow silence and precedes unrounded vowels, but the allophones in question only follow round vowels. (3) Assignment of the allophones in question to /kw/ would leave that phoneme still limited in range of distribution, not occurring word or contour finally and not occurring following front vowels, and would give /k/ a uniquely limited range of distribution, occurring following front and central vowels but not back vowels before front and central vowels, and not after /u/ before /o/. Assignment of labialized velar allophones following rounded vowels to /k/ gives that phoneme distribution before and after all vowels, parallel with the distributions of /p/ and /t/.Google Scholar
  15. 15.
    These allophones of /h/ are not symbolized but are to be assumed in other places where /h/ occurs and another allophone is not indicated.Google Scholar
  16. 16.
    The intervocalic stops all occur with alternate added length. This is to be assumed, as noted in section 3.12 above.Google Scholar
  17. 17.
    In AY there are five intervocalic series of phones, e.g. the gingivals [t], [tsh],[th], [r] and [s]. The first two and the last occur in contrast word initially; all apparently contrast intervocalically. They art interpreted as intervocalic occurrences of AV /d/, /tt/, /tk/, /t/ and /s/, respectively. In LGV initial [s] and medial [t] replace AV [tsh] so that the intervocalic correspondences of AV [tsh] and [t] coincide phonetically.Google Scholar
  18. 18.
    voiceless liquids are allophones of /h/; see section 3.12112 above.Google Scholar
  19. 19.
    For various phonetic sequences including [m] and representing /m/ and a vowel phoneme, see sections 3.12211 and 3.12214 below.Google Scholar
  20. 20.
    See section 3.12214 below.Google Scholar
  21. 21.
    reasoning here is not circular. The range of allophone [I] of /l/ is described in terms not including occurrence after [ñ] in this case.Google Scholar
  22. 22.
    phoneme /n/ has not been recorded following /i/ or /u/ in local words. Note that many forms like /nan/[nan] ‘eat it’ occur only with /n/, never with /n/; these contrast with forms occurring with /n/, all of which also occur with /n/ in LGV. Compare the /k/ — /kw/ contrast. These facts and the limited distribution of /n/, which does not occur word initially, as second member of clusters nor after /i/ or /u/, mark the /n/ — /n/ contrast as one of minimal functional load. Perhaps it is being lost, as seems much more likely, or currently being borrowed from dialects like GD, where /n/ has distributional range parallel to that of /n/.Google Scholar
  23. 23.
    Glottal stop is treated throughout this study separately from other stops for two reasons. First, on the synchronic level, it is distributed differently from other stops. Second, comparative data indicate that it has an origin very different from that of the other stops. This phoneme is in some dialects of Dani lexically contrastive in only a very few forms, principally the word /a’it/ ‘his dislike,’ and some proper names like the LGV name /jano’e/ [jane’3] ‘Janoe (name of unknown meaning).’ In LGV the substitution of glottal stop for /k/ and many occurrences of /t/ as first member of a cluster with liquids produced a few additional contrasts. It appears that in proto-Dani glottal stop was a junctural or terminal phenomenon which perhaps became morpheme medial in /a’it/ when an old sequence of two items with plus terminal border between them became fused. With this contrast introduced, even though it has a very light functional load, a synchronic analysis is forced to recognize glottal stop as phonemic wherever it occurs, even in association with contour onsets and terminals. A practical orthography needs to symbolize only contour medial occurrences, and a cross-dialect practical orthography needs to mark only intervocalic occurrences.Google Scholar
  24. 24.
    The interpretation of some sequences as including or not including /j/ and /w/ is still problematical. Two vowels occur in sequence with no. consonant between them: /kain/ [kain] ‘species of pandanus’; /laet/ [la3t] ‘shrub with spicy leaves’; /la-ok/ [làck] ‘women’s greeting.’ These sequences would appear to be evidence for interpreting [uV], [uV] and [oV] as not including intervocalic /w/, and sequences [iV], [iV] and [eV] with no intervocalic /j/. Informant reaction favors this interpretation in the cases of /vV/, /oV/, /iV/, and /eV/, for local speakers will not accept a pronunciation of these sequences with a presyllabic [j] or [w] before the second vowels, even though the first vowels of sequences /ea/ and /oa/ often are heard with postsyllabic offglide: /lia/ [lia, but not *lija] ‘light’; /luok/ [luck, but not *luwck] ‘let’s go’; /weak/ [weiak but not *wejak] ‘bad’; /oak/ [ovak but not *owak] ‘bone.’ These sequences may be interpreted then as simple two-vowel clusters wherever they occur. However, informants react differently with regard to [iV] and [uV] sequences; when asked to syllabify words including these sequences, they produce a second syllable with initial /j/ after /i/ and /w/ after /u/, and they accept these pronunciations: /uwe/ [u3, uw3] ‘Uwe river.’ Tentatively, the sequences [iV] and [uV] are interpreted as representing sequences /ijV/ and /uwV/. Another problem involving /j/ occurs in the sequence heard in /ajuk/[ajuk] ‘his fear.’ Earlier this form was interpreted as */aivk/, but it now appears probable that no two part syllablcs occur in LGV without a following consonant other than /j/ or /w/, and that this consonant is recognizable by friction absent in the form under discussion. Informants forced to syllabify produce the form as [a.jvk].Google Scholar
  25. 25.
    Note that /i/ is closer to /u/ than /i/ is to /v/.Google Scholar
  26. 26.
    The possibility of recognizable or contrastive timbre for a nasal has been questioned, but it is reported also by Trager and Smith, Outline of English Structure (Studies in Linguistics Occasional Papers No. 3. Norman, 195l) p. 41. In reference to this report—albeit the reference is skeptical-Pike proposes an analysis identical to that proposed here for LGV Dani. Kenneth L. Pike, op. cit part II, p. 48, fh.Google Scholar
  27. 27.
    For other similar allophones of /i/, see section 3.12214 below.Google Scholar
  28. 28.
    This analysis is preferred because: (l) The bilabial stricture does not occur invariably; the velar stricture does. (2) The bilabial stricture is phonetically similar to features of /u/, which occurs in all these cases; it can be interpreted as conditioned by the environment; the velar stricture cannot. (3) Some of these forms are cognate with forms in other dialects with final /k/; none is represented elsewhere with final /p/: LGV /touk/, BV /dvk/ ‘species of bananas.’ (4) The analysis parallels the analysis of /k/ after two part syllablcs with final /i/, where no other type of stricture occurs. Note that an analysis of the complex unit as representing a consonant cluster would posit a cluster word finally, where none occur otherwise.Google Scholar
  29. 29.
    Analysis of stylistic modifications is incomplete, but is referred to on a tentative basis; in the illustrations. Note that the environment specified includes syllables where /u/ is not adjacent to a bilabial but where a preceding or following syllable not contour-initial includes /u/ adjacent to a bilabial; it also includes contour initial syllables preceding syllables including /u/ adjacent to a bilabial consonant.Google Scholar
  30. 30.
    This last form appears to be differently modified from the others.Google Scholar
  31. 31.
    For other allophones see section 3.12214 below.Google Scholar
  32. 32.
    discussion of the phonemic interpretation of tnese sequences, see footnote 24 in this chapter.Google Scholar
  33. 33.
    This vowel ranges around a norm which is near the lower boundary of the writer’s English vowel in ‘but,’ so that early recordings repeatedly symbolized that phone and a lower one as allophones of /a/. Hearing and recording were not consistent, however, and it began to appear that the variation was around a single norm and not around two different norms as for positional allophones. The coincidence of the LGV norm with the border between two of the writer’s English vowels and two phonetic norms he had drilled prejudiced his hearing. Some utterances of /a/ are definitely higher than others, but they range around the single described norm.Google Scholar
  34. 34.
    This analysis is adopted in preference to an analysis of these items as vowel plus /j/ or /w/ because: (l) The laminai allophones occurring after /i/ elsewhere occur after /Vi/ also; if the two part syllabic is described as including /i/, the common element in the two environments is the conditioning factor for the allophones in question. (2) More importantly, the analysis adopted avoids positing unique final clusters.Google Scholar
  35. 35.
    See section 3.123 for extra-systematic nasalization of vowels.Google Scholar
  36. 36.
    See section 3.123 for extra-systematic voicelessness of vowels.Google Scholar
  37. 37.
    This kind of data was called contrastive for /kw/ and /n/, but is here considered extra-systematic since nasalization cannot be demonstrated as contrastive in the general vocabulary but only in this special reply form. Compare extra-systematic nasalization in English ‘uh-huh.’Google Scholar
  38. 38.
    voicelessness is structurally different from stylistic voiceless-ness as described in section 3.12223. Here certain forms are always voiceless; there forms with normally voiced vowels occur with voiceless vowels in certain positions.Google Scholar
  39. 39.
    Compare extra-systematic airstreams in English gasp, and ‘tsk.’Google Scholar
  40. 40.
    Compare Kenneth Pike’s hyperphonemes, especially the discussion in section 9.6, op cit., Part II, pp. 63–66.Google Scholar
  41. 41.
    The term interlude is from Charles Hockett, A Manual of Phonology. (Baltimore, 1955), p. 52.Google Scholar
  42. 42.
    There is a phonetic difference between the /kh/ sequences in /wakeikha/ and /lek-han/; in the first the /h/ is fortis and closely joined to the /k/ like strong aspiration, but in the second the /h/ as part of a stressless clitic syllable, is lenis and more loosely joined to the /k/.Google Scholar
  43. 43.
    The presence of the second is recognized by intonational features to be discussed in sections 3.133 and 3.136 below.Google Scholar
  44. 44.
    In these illustrations a plus mark after a superscript number indicates raised pitch above the level marked by that number, and a minus sign indicates lowered pitch. Raised dots next to symbols mark slight length.Google Scholar
  45. 45.
    An alternative analysis would be to consider this secondary stress as a shifted primary stress in these forms. Since the pre-terminal syllable still carries short pitch upglide, it is however convenient to consider the primary stress as still occurring on that syllable, although it is not markedly longer or louder than the preceding syllable. The main contrast in forms including and not including secondary independent stress is in rhythm pattern and relative stress of the syllables. This has been marked in this section of the discussion only to avoid cluttering illustrations with unexplained detail in earlier sections. Earlier during the analysis this secondary stress was considered a predictable phenomenon on the basis of the phonemic shape of words, and some data indicate that this is very nearly true. For example, this stress only occurs on central and front vowels and only on syllablcs followed by voiced segments. Further checking of the data indicates, however, that no statement of predictability thus far proposed is valid.Google Scholar
  46. 46.
    The determination of clitics is not yet completely clear in every bit of recorded text, but it is clear in most cases and probably could be cleared in the remainder by rechecking the questionable material. This has not been possible during the actual preparation of this thesis. Writing of borders on the basis of stress and the other phonetic data described above conforms very closely to grammatical fact» In a small residue where clitics are phonologically demonstrable but morpheme boundary is not yet recognized, it is probable that further analysis will disclose such a boundary. Note that presence of a grammatical boundary does not indicate presence of a clitic, however; many morphemes are suffixed to form part of single phonological words.Google Scholar
  47. 47.
    Contour stress has not been marked in preceding illustrations.Google Scholar
  48. 48.
    Thus plus terminal is considered to occur wherever pause occurs, but not all plus terminals occur with pause. This analysis does not distinguish different types of pause (except for the mention of hesitating or interrupted forms), as Pike did for English and still does in modified fashion: Kenneth Pike, op. cit., Part II, section 9.4. The difference he attributes to two sorts of pause is here attributed to the presence or absence of sentence terminal. Pause is considered a secondary feature of the contour group with which it occurs, as Pike now treats it, and not as a segmental phoneme, as Bernard Bloch treats pause in Japanese: “Studies in Colloquial Japanese IV,” language 26:115 (Baltimore, 1950).Google Scholar
  49. 49.
    The term zero is chosen for parallelism with the term plus and because of the absence of most special terminal features with this terminal.Google Scholar
  50. 50.
    that although four contrastive levels are postulated and some contours are similar to English contours, the system is quite different. First, stress is usually marked not by high but by a short upward gliding pitch, and lengthened syllables, when mechanically slowed to half speed, are heard as a series of quavers or short glides. Second, in most cases the ‘intonation contour’ is actually only a single contrastive pitch occurring contour group finally, and preceding pitches are predictable.Google Scholar
  51. 51.
    All illustrations in this paragraph are from text material.Google Scholar
  52. 52.
    For discussion of predictable pitches ss section 3.136 below.Google Scholar
  53. 53.
    Sentence stress has not been written in earlier illustrations.Google Scholar
  54. 54.
    Thus pause is made a feature of only one contrastive unit, plus contour terminal. Also the sentence appears clearly as made up entirely of contour groups each with its own terminal. Sentence terminal is thus made to account only for distinctive features not occurring elsewhere.Google Scholar
  55. 55.
    The parentheses around the intonation marking numbers indicate these pitches occur on these syllables when they are voiced.Google Scholar
  56. 56.
    This example, including reduplication, and all others of four syllable morphemes are doubtful, but until morpheme boundary is clearly indicated, it is perhaps easier to describe morphemes of four syllables.Google Scholar
  57. 57.
    Initial glottal stop is part of contour onset; see section 3.133.Google Scholar
  58. 58.
    These forms are not single morphemes, but what is word initial must also be morpheme initial.Google Scholar
  59. 59.
    The two part syllabic /oi/ has not yet been recorded initially.Google Scholar
  60. 60.
    This is one of very few examples with /kw/ here; there are none with word final /kw/.Google Scholar
  61. 61.
    There may be morpheme boundary in this form.Google Scholar
  62. 62.
    There may be morpheme boundary here, although the cluster is probably morpheme medial in any case.Google Scholar
  63. 63.
    The forms with initial /a/ may include morpheme boundary, but the parallel forms with no recognizable boundary between /’n/ and /’l/ indicate that this is a valid cluster type, at least.Google Scholar
  64. 64.
    This form is probably morphemically complex, consisting of a very uncommon verb form now used as a name for western saws.Google Scholar
  65. 65.
    Apparent clusters with initial /i/ are interpreted as including /j/, and those with initial /u/ or /ou/ as including /w/; see footnote 24, page 40, above. Note the following: /sijep/ ‘clan name’; /sijat/ ‘area south of the range’; /asijok/ ‘breeding boar’; /uwe/ ‘Uwe river’; /uwan/ ‘long’; /huwon/ ‘a type of bark container’; /ouwak/ ‘physiological term of uncertain meaning’; /houwok/ ‘small adze.’Google Scholar
  66. 66.
    Some speakers do not have /w/ I-word medially.Google Scholar
  67. 67.
    Some speakers have this pair as homophones, with alternate pronunciations as for /watti, wathi/ ‘I killed him.’ LGV exhibits a tendency to replace stqp-/h/ clusters with geminate stop clusters.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1961

Authors and Affiliations

There are no affiliations available

Personalised recommendations