East Tor Coast

Timothy Usher, Santa Fe Institute

The East Tor Coast family consists of three closely related languages, Kwinsu, Tena and Fitou (Vitou,) spoken in three villlages, Ansudu, Betaf and Takar, along the north coast of the Pantai Timur district of Sarmi county in Indonesia's Papua province (Lee and Wambaliau 2005: 2-3.)

The internal classification of East Tor Coast is as follows:

East Tor Coast

Kwinsu

Fitou-Tena

Tena

Fitou

Galis (1955) 28 comparative terms for Kwesten (this is most likely Kwesten of Takar village i.e. Fitou)

Voorhoeve (1975: 107) 39 comparative terms for Kwesten after Anceaux (n.d.) (this is not a single language but a conflation of all varieties designated “Kwesten” in Anceaux n.d., q.v. Smits and Voorhoeve eds. 1994 below)

Smits and Voorhoeve eds. (1994: 18-266) three comparative vocabularies of Kwesten of Takar village (Fitou) after an unnamed teacher and two unnamed constables respectively and one comparative vocabulary for Kwesten of Betaf (Tena) village after an unnamed constable

Lee and Wambaliau (2005: 40-48, 49-56) 239 comparative terms and (pp. 57-62) 21 sentences for Vitou of Takar, Tena of Betaf and Kwinsu of Ansudu village

[under construction]

Proto-East Tor Coast had 12 consonants and 5 vowels as follows:

In addition to the simple vowels given above, at least five diphthongs are found as follows:

Any consonant, vowel or diphthong can occur intially or medially.

Only a restricted set of consonants occurs word-finally:

A final /k/ has been clearly found in only a few Kwinsu nominals which have no Tena and Fitou and may well be loans from other Tor River langugaes. Because Tor River /*k/ has been dropped in East Tor River, in which present-day /*k/ reflects /*g/, final /*k/ seems unlikely to be a valid proto-East Tor River segment.

An unusual feature of Tor River languages from a New Guinean perspective is the frequent occurence of root-initial clusters /*CC/. These result from the loss of destressed schwa vowel [*ə] in initial syllables. The following combinations have been found:

It is possible that there were clusters /*kw/, but these would be interpreted as unit phoneme /*kʷ/. The reason that /*kʷ/ is interpreted as a unit phoneme, besides its origin as Tor River /*gʷ/ (ultimately an occlusion of /*w/,) is that initial clusters of three consonants are disallowed, yet both /*kʷɾ/ and /*fkʷ/ occur.

Bilabial voiced stop /*b/ occurs as the second member of only one combination, /*db/ in /*dbaɾ/ “head.” This might just as well have been reconstructed as /*df/, as [p] is an allophone of /*f/ in certain environments, e.g. following root-final /*n/ in a final cluster (below,) and the behavior of /*dk/ is parallel, yielding [dg] where medial /*k/ would usually be lost (below.) Historically, this reflects /*nfaɾ/ (cf. Apauwar River /*nVbaɾ/ where /*b/ regularly corresponds to Tor River /*f/.) If interpreted as /*df/ here, cluster rules would be simplified as otherwise /*b/ cannot be the second member of an initial cluster. It is shown here as /*db/ because medial /*b/ has been established in East Tor Coast, if only marginally (below.)

Clusters consisting of nasals followed by homorganic stops are found medially, but not initially, perhaps because they have been denasalized in this position as in /*dbaɾ/.

The most conspicuous gap is the lack of any clusters with apical stops /*t *d/ as their second members. The reason for this is likely their rarity in medial position (below,) which in turn is due to the fact that, while original /*d/ is reflected as Tor River /*t/ initially, it has been merged with non-stop /*ɾ/ medially, hence the high frequency of clusters /*Cɾ/.

Apical non-stop /*ɾ/ cannot be the first member of a cluster because it does not occur word-initially except as a reflex of Tor River /*kɾ/ where /*k/ has been lost, East Tor Coast /*kɾ/ reflecting /*gɾ/. Similarly, the cluster /*wɾ/ ultimately reflects /*kʷɾ/ with East Tor Coast /*kʷɾ/ reflecting Tor River /*gʷɾ/.

Palatal non-stop /*j/ has not been found as either the first or second member of an initial cluster, although at least /*mj/ occurs medially (below.)

In principle, allowable medial clusters include any consonant allowed root-finally /*m *n *f *t *s *k *ɾ/ followed by any consonant or consonant cluster allowed root-initially, though not all such combinations have been attested, geminates do not occur and a few combinations such as /*mf/ may be forbidden as /*f/ might be realized as [*b] in this position.

Initial and medial simple consonants correspond as follows:

Initial consonant clusters correspond as follows:

Final consonants correspond as follows:

Vowels and diphthongs correspond as follows:

These correspondences are exemplified as follows, with Kwinsu, Tena and Vitou (Fitou) attestations drawn from Lee and Wambaliau (2005: 40-48, 49-56, 57-62) and Betaf (Tena) and Takar (Fitou) attestations from Anceaux (n.d.) as excerpted in Smits and Voorhoeve (1994: 18-266) after unattributed elicitors (above.) For clarity, our presentation is divided into five sections, beginning with initial and medial simple consonants, followed by initial consonant clusters, medial consonant clusters, final consonants, then vowels and diphthongs.

Initial and medial bilabial nasal /*m/ is retained as such in all descendants:

Initial and medial apical nasal /*n/ is retained as such in all descendants:

Initial and medial labiodental fricative /*f/ is retained as such in all descendants. It is sometimes realized as voiced [v] in Tena and in Fitou:

Initial apical voiceless stop /*t/ is retained as such in all descendants:

Medial /*t/ is uncommon:

Initial and medial laminal fricative /*s/ is retained as such in all descendants:

Initial velar voiceless stop /*k/, reflecting Tor River voiced /*g/ as original /*k/ has been dropped, is uncommon. It is retained as such in all descendants:

Medial /*k/ is dropped in Tena and in Fitou:

Initial rounded velar voiceless stop /*kʷ/, reflecting Tor River voiced /*gʷ/ as original /*kʷ/ has been deoccluded to /*w/, is retained as such in all descendants:

Medial /*kʷ/ is deoccluded to /w/ in Tena and in Fitou:

Bilabial voiced stop /*b/ is retained as such in all descendants:

Several of the comparisons above are known to be loans from the Sarmi Coast subfamily of Oceanic: /*…/ “sweet potato” …

Medial /*b/ is uncommon:

Initial apical voiced stop /*d/ is retained as such in all descendants:

Several of the comparisons above are known to be loans from the Sarmi Coast subfamily of Oceanic: /*…/ “thorn” (Oceanic Ross: CITE,) /*…/ “fence”, and probably also /*…/ “blood.”

Medial /*d/ is uncommon, and likely occurs only in compounds and loans. It is merged with apical non-stop /ɾ/ in Kwinsu to form a single phoneme, as rare initial /*ɾ/ is likewise merged with /d/ (below):

There is no laminal voiced affricate /*dz/ because Tor River /*dz/ has been merged with voiceless /*s/ (above.)

There is no velar voiced stop /*g/ or rounded velar voiced stop /*gʷ/ because Tor River /*g *gʷ/ have been devoiced to /*k *kʷ/ respectively (above) following the loss of original [*k].

Bilabial non-stop /*w/ is retained as such in all descendants:

Initial apical non-stop /*ɾ/ is very uncommon. In the only known nominal example, “bone” below, it reflects Tor River cluster [*kɾ] with regular loss of /*k/ everywhere except Jofotek (Mander.) It is also found in several verb roots in which it is likely allowed because it is typically prefixed. It is merged with initial /d/ in Kwinsu, in parallel with the lenition of medial /*d/ to /r/ (above):

Medial /*ɾ/ is common and is retained as such in all descendants:

Palatal non-stop /*j/ is retained as such in all descendants:

In addition to the simple consonants given above, consonant clusters occur both initially and medially. Known examples of segment [kw] reflect Tor River unit phoneme /*gʷ/ and hence are not analyzed as clusters here. Excepting words suffixed with locative /*-f/ (below,) clusters do not occur finally.

Cluster /*fn/ is retained as such in all descendants:

In clusters /*tm *dm *tf *db/ consisting of an apical followed by a bilabial or labiodental nasal or stop, the apical stop is dropped in Kwinsu and apparently also in modern Tena and Fitou. Its retention in colonial-era vocabularies suggests this to be a very recent change:

Cluster /*db/ is likely underlyingly /*df/, with /*f/ being voiced in parallel with the behavior of /*k/ in cluster /*dk/ (below,) but is shown here as /*db/ because medial /*b/ has been established in proto-East Tor Coast, if only marginally.

In clusters /*tk *sk/, velar voiceless stop /*k/ is lost in Tena and Fitou as it is medially (above):

In cluster /*dk/, /*k/ is not dropped but retained as voiced [g] in both Tena and Fitou, a sound which does not otherwise occur in East Tor Coast languages as formerly voiced /*g *gʷ/ have been devoiced to /*k *kʷ/. This behavior parallels that of cluster [*db] which is most likely underlyingly /*df/ (above):

…:

…:

Initial clusters /*mw *nw *tw *sw *dw/ are retained as such in all descendants:

Intial clusters /*mɾ *fɾ *bɾ *tɾ *kɾ *wɾ/ are retained as such in all descendants:

Intiial clusters /*kʷɾ/ are realized as [kuɾ]-:

… medial clusters …:

Cluster /*nt/ …:

Cluster /*ns/ …:

Cluster /*nk/ is generally realized as [ŋk] in all descendants; however Kwinsu preserves [nk] as often as not, showing /*nk/ to be original:

Cluster /*nkʷ/ is generally realized as [ŋkʷ] in all descendants:

Cluster /*mb/ …:

Cluster /*nd/ …:

Only a restricted set of consonants /*m *n *f *t *s *k *ɾ/ occurs word-finally.

Final bilabial nasal /*m/ is retained as such in all descendants:

The majority of East Tor Coast nominals have a final apical /n/ in their citation forms. These can be divided into two classes, those in which /*n/ is part of the underlying root, and those which take a suffix /*-(n)/ marking nominality. This suffix appears on every nominal which ends in a vowel other than a diphthong. There are several reasons for this analysis. Only nominals are affected. and there are otherwise no vowel-final nominals. This final /*-(n)/ is dropped when the root is the first member of a compound, even when the second member begins with a vowel (below.) These two finals correspond to /*n/ and zero respectively in Orya-Tor languages which do not add this suffix.

Final apical nasal /*n/ is retained as such in all descendants:

… /*-(n)/ …:

In many instances this final /*n/ is dropped when the root occurs as the first member of a compound, even when the next root has an initial vowel or when the result would otherwise be an allowable cluster:

In a smaller number of examples, final /*n/ is retained as such, even in complex clusters:

The reason for this difference in behavior is found in comparison to other Tor River languages. Here there are two final correspondences, one in which /*n/ is retained as such everywhere and a more common one in which /*n/ is found also in Keijar, Dabe and Kwesten but unexpectedly absent from Edwas, East Tor River, Berik, Bonerif and Orya forms. These correlate with those East Tor Coast roots which drop final /*n/ in compounds and those which don't (above.)

A priori, this discrepancy might be viewed as phonological in origin, for example by positing a velar nasal /*ŋ/ which appears only finally, and is either dropped or merged with /*n/ in various descendants. There are several reasons however to believe it to reflect a suffix which is mandatory in only some Tor River languages. Nearly every East Tor Coast nominal, but not adjectives or verbs, ends with a final consonant, the majority of which are /*n/, or a diphthong (below.) Moreover most but not all of these final /*n/ are dropped in compounds, as shown above. The rule would seem to be that any vowel-final nominal root is suffixed with /*-n/ when appearing as a freestanding word, unless the final vowel is a diphthong.

Further evidence for the morphological origin of /*-(n)/ is found in words shared with the Sarmi Coast subfamily of Oceanic (q.v. Lee and Sawi 2005: 30-43,) which, like other Tor River languages, retain /*n/ but show no sign of /*-(n)/:

…

Final labiodental fricative /*f/ is generally retained as such in all descendants:

In several examples where root-final /*f/ is followed by locative particle /-de/ in Fitou, it is realized as [p]:

A special case of final /*f/ is found in locative terms which take a suffix [-Vp] in all descendants following roots with final /*n/.:

That [*-Vp] is a suffix is evident from the fact that the terms [*oɾ-mwan-ap] “inside” and [*masan-ap] “in front” are derived from /*oɾ-mwan/ “mouth” and /*masan/ “nose” respectively. The second vowel of [*mun-up] “behind” must be interpreted as a copy of the vowel in the preceding syllable as this is clearly the same suffix as those realized as [-ap]. This syllable-final [a] is not centralized in Kwinsu as is /*a/ following other /*a/ in a previous syllable of the same root (below.) This unexpected discrepencies suggest that the original form was merely /*-f/, broken up here because phonetic consonant clusters are not allowed word-finally. Otherwise, no final [p] occurs.

…:

Final apical stop /*t/ is retained as such in all descendants:

Final laminal /*s/ is retained as such in all descendants:

Final velar voiceless stop /*k/ …:

Final apical non-stop /*ɾ/ is retained as such in all descendants:

Vowels /*i *u *e *o *a/ and diphthongs /*uⁱ *eⁱ *aⁱ *oᵘ *aᵘ/ are generally retained as such in all descendants, with simplification of diphthongs in Kwinsu.

High front vowel /*i/ is retained as such in all descendants:

High back vowel /*u/ is retained as such in all descendants:

Mid front vowel /*e/ is generally retained as such in all descendants:

Where /*e/ follows low central /*a/ in a previous syllable, it is usually but not invariably realized as [a] in Tena and in Fitou:

… /*e/ …:

Mid back vowel /*o/ is retained as such in all descendants:

Two roots with the vowel pattern /aCo/ have become /oCa/ in Tena and Fitou for unknown reasons:

…:

Low central vowel /*a/ is generally retained as such in all descendants:

Where /*a/ in a final syllable follows another /*a/ in a previous syllable, it is generally realized as mid central [ə] in Kwinsu:

Two examples of /*a/ following diphthong /*aⁱ/ followed by /*k/, which has been neutralized to [a] in Kwinsu (below,) are likewise affected:

One example in which Lee and Wambaliau give Kwinsu [o] following bilabial non-stop /*w/ might be viewed as a variant of this pattern:

There is at least one unexplained exception:

In several examples where /*a/ follows mid back vowel /*o/ in a previous syllable, it is levelled to /o/ in Kwinsu:

In reduplicated adjectives /*CaC-CaC/, both /*a/ are realized as [ə]:

In addition to the simple vowels given above are found at least five diphthongs /*uⁱ *eⁱ *oᵘ *aⁱ *aᵘ/.

Diphthong /*uⁱ/ is merged with high back vowel /u/ in Kwinsu:

There are several reasons to consider the term for “egg” to contain a diphthong, as opposed to possible forms /*suwi/ or /*swi/. First, the behavior of /*uⁱ/ in Kwinsu is not consistent with either of these alternatives, but follows the same general trend of simplification found with diphthongs /*aⁱ *oᵘ *aᵘ/ (below.) Second, the presence of the offglide prevents the appearance of final /*-(n)/ which otherwise appears on every nominal besides those with underlying final consonants (above.) In this respect, final diphthongs behave like final clusters /*Vj *Vw/, but are not reconstructed as such because they can themselves be followed by final consonants (below) and there are otherwise no final clusters.

Diphthong /*eⁱ/ is generally retained as such in all descendant, though sometimes merged with high front vowel /i/ in Kwinsu and Tena:

Diphthong /*aⁱ/ …:

When /*aⁱ/ is followed by velar voiceless stop /*k/, it is merged with low central vowel /a/ in Kwinsu:

Diphthong /*oᵘ/ is realized as [u uᵘ] in Kwinsu (cf. /*aᵘ/ below) and sometimes in Tena and Fitou:

Diphthong /*aᵘ/ is realized as /oᵘ/ [o oᵘ u] in Kwinsu:

For unknown morphological reasons, adjectives are often suffixed with a copy of the vowel preceding the final consonant. This tendency is not phonologically regular. Kwinsu is usually not affected: