ABKHAZIAN

The phonemic inventory used in this study appears in Ruhlen (153):

ABKHAZIAN (Abzhui) [Lomtatidze 1967a] [Caucasian: North: Northwest]
[SW USSR (Abxaz SSR); 90,000]

Remarkable here is the extremely elaborate consonantal inventory with only two vowels. The contoid inventory contains dental contoids [t^h], etc. which receive the following specification:

The distinctive features in use here allow us to distinguish dental contoids from alveolars without invoking a feature such as SPE [distributed]; [raised] indicates an articulatory gesture in which the tongue front goes higher than the teeth in contoidal articulation. The uvular stops, /q^?/ etc. are considered [+raised] since the back is raised but [-high] since the uvula descends to make the articulation. Using [raised] as defined here we can distinguish between both dental and alveolar contoids as well as uvular and velar articulation.

Abkhazian contoids also stand in contrast according to their palatal, and labial timbres which are indicated by the addition of these features to the specifications:

In the case of palatalized palatal contoids, /š^j, ž^j č^hj/ etc, they are marked twice for [palatal] to distinguish them from their closest congeners (cf. section on Lapp).

The most complex segment of this complex inventory is /č^?j/, a palatalized palatal glottalic affricate which needs 11 positive specifications to be identified and to contrast with /č^hj/.

AFRIKAANS

The phonemic inventory of Afrikaans possesses a vowel system that is more elaborate than most examined here, yet neither the vowel system nor the four distinctive points of articulation of the contoids pose a problem for the features proposed here. The system:

AFRIKAANS [Pienaar and Hooper 1948] [Indo-European: Germanic: West]
[South Africa; 2.5 million]

approaches the mean in the number of phonemes, in the average number of positive specifications per phoneme, and in the number of distinctive features needed to specify all the contrasting segments:

ALABAMAN

This inventory

ALABAMAN [Rand 1968] [Macro-Algonquian: Muskogean] [Texas; 400]

presents no problem for the features proposed here. It is included because it contains relatively few segments, because the feature [high] is not necessary.

Aleut

Beyond the sparseness of segments in this inventory,

ALEUT (Eastern) [Geoghegan 1944] [Eskimo-Aleut: Aleut] [Alaska
(Aleutian Islands); 600]

it is remarkable because there is no need for the feature [contoid] since there is no /r/ phoneme which would be negatively specified for all exclusively contoidal articulatory gestures and, thus, possibly confused with vocoid.

AMHARIC

This inventory is included

AMHARIC (Addis Ababa) [Leslau 1968] [Afro-Asiatic: Semitic: South: Ethiopic] [Ethiopia; 8 million] SOV/AN

because of its relative complexity and because of its utilization of the glottalic air stream in contoidal articulation. I find it remarkable that this language systematically distinguishes between long and non-long glottalic contoids.

ANDOA

This inventory

ANDOA (Iquito) [Eastman and Eastman 1963] [Andean-Equatorial:
Andean: Zaparoan] [N Peru; 500]

is included because of its segmental sparseness and because of the unusual segment, /h̃/, a surd, non-contoidal, nasalized fricative.

EGYPTIAN ARABIC

The inventory of Egyptian Arabic

(EGYPTIAN) ARABIC (Colloquial) [Mitchell 1962] [Afro-Asiatic:
Semitic: South: Southwest] [Egypt; 45 million]

was included because of its pharyngeal segments // which are specified as:

These segments contrast with pharyngealized contoids /t^D, d^D, s^D, š^D/ etc. which have a pharyngeal timbre and are specified as follows:

A phonological grammar of Arabic will state that raised contoids which have the feature [labial] will produce the extremity contraction at the pharynx.

BINI

This inventory

BINI [Wescott 1965] [Niger-Kordofanian: Niger-Congo: Kwa: Edo]
[W Nigeria; 300,000]

was included because of the coarticulated stops . These segments, if specified as

would not contrast with /k^w/ and /g^w/, labialized velars. Likewise they pose the problem of how one would specify theoretically possible labial contoids with velar timbres, i.e. p^x, b^g. I propose the following:

This solution for such contrasts resembles those for palatalization and pharyngealization which distinguishes timbres from articulations. I consider it debatable whether or not to add a delayed transition feature to the specification of coarticulated stops. For the time being and the purposes of this exposition, their special status is recognized by two positive specifications for [occlusion].

BUANG

This inventory

BUANG (Central) [Hooley 1964] [Austro-Tai: Austronesian: Oceanic:
Northeast New Guinea] [Territory of New Guinea (Morobe);
7,000]

was included because of its uvular contoidal series and its prenasalized segments.

CHEROKEE

This inventory

CHEROKEE [Bender and Harris 1946] [Macro-Siouan: Iroquoian: Southern]
[W North Carolina; 10,000]

was included because of its sparseness and because of the fact that, owing to its peculiar nature, [occlusion] is not a needed feature.

CHIPEWYAN

This inventory

CHIPEWYAN (Fort Chipewyan) [Fang-Kuei 1946] [Na-Dene: Athapaskan-
Eyak: Athapaskan] [Canada (Alberta); 5,000]

was included because of its complexity and because of its unusual segments: //, /c^c/, /t¹/ etc., that is [+slit] affricates which contrast with // which are [-slit] and [-raised] i.e. dental, thus [ distributed] in the SPE features. This contrast, . cannot be made in the SPE features whereas it is easily made using the revised inventory. The lateral affricates are also unusual, but easily specified here.

DAJU

This inventory

DAJU (Shatt) [Tucker and Bryan 1966] [Nilo-Saharan: Chari-Nile: Sudanic:
Eastern] [W Sudan; 100,000]

was included because of its relative complexity and because of its unusal segments: prenasalized stops and fricatives, implosives and labialized contoids.

DYIRBAL

This inventory

DYRIBAL [Dixon 1972] [Australian: Pama-Nyungan: Pama-Maric:
Pama: Western] [NE Australia (N Queensland); 40]

was included because of its simplicity and because of its lack of surds proving that languages may exist with no voiceless segments whereas no language exists which has no voiced segments. This supports my assertion that [surd] is the marked quality.

GADSUP

This inventory

GADSUP (Agarabi) [Goddard 1967] [Indo-Pacific: Central New Guinea:
East New Guinea Highlands: Gauwa] [Territory of New
Guinea (Eastern Highlands); 7,000]

was included because of its simplicity and because, again, the feature [surd] is not a necessary ingredient, nor is [palatal] nor [high].

GBEYA

This inventory

GBEYA (Bossangoa) [Samarin 1966] [Niger-Kordofanian: Niger-Congo:
Adamawa-Eastern: Eastern] [NW Central African Republic;
600,000]

was included because of its complexity and because of its unusual segments: coarticulated velarolabials, /kp, gb, qm/ (the nasal contoid is doubly marked for [nasal] as the occlusives are for [occlusion]); implosives, prenasalized stops, and preglottalized nasals, /?m, ?n/. Glottalized segments are specified as contoids or vocoids plus [glottal constriction]. Preglottalized would be specified as e.g. ?C [+contoid,+glottal constriction, +D.T.]. These are different from glottalic segments, C? [±contoid] which use the glottalic air stream.

!KÕ

This inventory

!KO [Maingard 1958] [Khoisan: Southern] [NW Botswana]

was included, along with that of Kung, because of the great number of segments and the especially elaborate set of clicks.

KOALIB

This inventory

KOALIB [Tucker and Bryan 1966] [Niger-Iordofanian: Kordofanian:
Koalib] [C Sudan (Nuba Hills); 24,000]

was included because of its five point occlusive contrast, its utilization of egressive and ingressive air streams and its rather elaborate vowel system.

KUNG

This inventory

KUNG [Snyman 1970] [Koisan: Northern] [NE South West Africa: 10,000]

was included because it was the most elaborate of all inventories found in Ruhlen.

LAPP

The analysis of the Lapp phonological system (Kert, 1966) appears to present several difficulties for the distinctive feature set postulated here. Kert has established the following inventory:

LAPP [Kert 1966] [Uralic: Finno-Ugric: Lappic] [NW USSR (Kola
Peninsula); 30,000]

Remarkable in this inventory is the number of segments that contrast in length and palatalization. The feature [length] of SPE has been discarded in favor of the utilization of the [delayed transition] features which are still sufficient to distinguish palatalized palatal affricates which are long or not long: This inventory contains which are considered ‘half palatalized’. I have chosen to contrast them as follows:

The ‘half-palatalized: sounds are specified the same way retroflex contoids in Indian languages are, [+high] but [ palatal] and [-dorsal]. This system claims that no language contrasts retroflex and ‘half-palatalized’ contoids.

In order to distinguish palatalized contoids from non-palatalized ones, I have employed the feature [palatal] without its (usually) concommitant feature [high] which is used to indicate palatal contoids. This, then, establishes a distinction between contoids with palatal timber, m^j, t^j, k^j, etc. and contoids articulated in the palatal area such as š, ž, č, etc. Palatized contoids are not complex in the same way affricates are, i.e., they are not positively specified for [delayed transition], because, contrary to affricates, they cannot be considered two contoids which pattern as a unit. The following matrix illustrates distinctions among, palatal, palatalized, and non-palatal segments:

Palatalization, I have claimed, is a timbre rather than a palatal contoidal articulation. The specification [+palatal] for non-palatal, palatalized segments suffices to distinguish them from ordinary nonpalatal contoids just as a specification such as [+palatal, +high, +contoid, +labial] would identify segments such as ñ^w which would contrast with m^j. This system can contrast x^j with š^x (a velarized palatal) by specifying the first as [+dorsal, +slit, +palatal], the second as [+palatal, +dorsal, -slit], thus they can be contrasted in this system if someday someone finds such segments in a language.

The contrast is not really problematic: is specified twice for palatal and once for [delayed transition]; is specified for both [D.T. 1, 2] and twice for [palatal]. In the Lapp inventory the palatalized quality of these palatal affricates is redundant, since they do not contrast with non-palatalized affricates—but, if they did, they could be distinguished from such segments using the strategem of double marking for [palatal].

MURA

This inventory

MURA (Piraha) [Sheldon 1974] [Macro-Chibchan:-] [W Brazil
(Amazonas); 100]

was included because it was the smallest of all inventories and contains no nasals, laterals or vibrants.

NORWEGIAN

This inventory

NORWEGIAN (Standard Eastern) [Vanvik 1972] [Indo-European:
Germanic: North] [Norway; 4.3 million]

is a rather elaborate Indo-European language with the interesting contrast of /ç/ and /š/ where [slit] and [groove] or [±dorsal] fricatives contrast in the palatal region.

NUNGGUBUYU

This rather sparse inventory

NUNGGUBUYU [Hughes and Leeding 1971] [Australian: Nunggubuyuan]
[NC Australia (NE Northern Territory); 400]

was included for its simplicity, but also because of the six point contrasts in stops. SPE DF’s cannot distinguish /p t̯ t ṭ c k/:

unless one uses the feature [distributed] for /c/ and /t/ and /high/ for /t/ contrary to Harris (1969).

URDU

This inventory

URDU [Bender 1967] [Indo-European: Indic] [Pakistan; 40 million]

was included partially because of its elaboration and partially because of the occlusive contrasts such as /p~p^h~b~b^h/ which are specified using my system through the presence or absence of a wide glottal opening. Using the Chomsky-Halle D.F.’s, the voiced aspirates are handled by an additional feature, [heightened subglottalic pressure].

Footnotes

¹Professor Ernie Scatton of the Slavic Language Department of SUNY Albany has informed me (personal communication) of a tendency among Soviet researchers to overelaborate the segmental inventories of the languages they investigate. Indeed, the long segments of this inventory could be analyzed as sequences, the palatalized as C+j. This would reduce the inventory to 33 segments. Hasselbrink (1965:6,20) postulates 28 phonemes for this language. He examines the virtues of analyzing palatalized consonants as units, sequences of contoid plus vocoid or as contoids modified by suprasegmentals. He decides upon a suprasegmental analysis, but such an analysis is not consonant with the phonetic theory involved here—it has no articulatory motivation. The solution that I use, considering palatalization a timbre rather than a contoidal articulation is implicitly suprasegmental inasmuch as [palatal] articulation may be added to any other sound. Obviously, the exact configuration of the tongue will vary according to the segment that is being palatalized.