The person charging this material is re- sponsible for its return to the library from which It was withdrawn on or before the Latest Date stamped below.Theft, mutilation, and underlining of books are reasons for distiplinary action and may result in dismissal from the University. Here are listed, underlined, the surface forms of the verbal endings.Even if the treatment of this morpheme as a non- segmental low tone which triggers 'downstep' and is then deleted is correct, which it may be, there would still seem to be a logical jump to the assumption that all contour tones on short vowels should be treated in this manner.
•' (b) Assign a low tone to the remaining syllable of the second member." (198) The operation of this rule is illustrated by the following forms from Mende, given in Leben (^1971). Against Woo's hypothesis of the segmental nature of tone, Leben also raises the problem of short vowels with a contour tone.
These cases could conceivably be handled by allowing features like [ [ hi] [j J/)]], or [±[ L] [ H] [ L] ] (for mb'a), but Leben argues: "In this case the claim that contour tones are underlying sequences of level tones would be nearly empty, since the representation [ [ High] followed by [-High]] on a segment is empirically equivalent to the representation [ Falling]." (196-7) To handle this problem Fromkin (1972) suggests: "One can however utilize a [-segmental] but tone bearing unit which could accomplish the same thing (as was proposed by Schacter and Fromkin (1968)) or adopt the Abadan Conference proposal utilizing tone-bearing non-syllabic vowels." (19) The proposal of this conference, which was entitled "Tone in generative phonology" and which took place at the University of Ibadan in Nigeria in 1970, was to represent a contour tone on a short vowel, for example a rising tone ds [ syllabic, low] followed by a unit [-syllabic, high].
Fromkin argues for her [-segmental] approach on the basife of "...certain cases where tone is the only realization of grammatical morphemes." (19) She offers as an example the 'habitual' low tone morpheme of Akan.
This morpheme results in a 'downstep' of a following high tone.
LHL niklll nyaha mba" HL kenya mbu In other words the generalizations about the pitch patterns on morphemes are independent of segmental composition, i.e. In addition, a constraint which rules out the HLH sequence on all morphemes regardless of the number of syllables would- be difficult to state if tones were a feature of segments.
Leben' s evidence for the treatment of contour tonfes as sequences of level tones comes from processes of Tone-copying and Tone-deletion.
It should also be pointed out that, with the explainable exception discussed in footnote 8, the third-person-object transitive endings are also used as possessive endings on nouns, with subject corresponding to possessor and object to possessed. Notice that there is redundancy here (and elsewhere in the system) in that number is marked both by the number- marking affixes (55, -k-, -t-) and by the person- marking morpheme, the form of which depends on the number of the referent. We will mention those which we consider to be the most representative and influential.
The intransitive verb is also marked by (t)u- preceding the verbal endings, the transitive by a(q)-. This -q (where is expected) is an instance of a pervasive tj alternation whose details are not clear to me. Here the -q- is the reflex of the transitive marker posited by Koo. In the corresponding possessive ending we find the expected -ka . It may be that e-epenthesis should be formulated to apply in these forms (yielding te te from t te) , If so, then the problem arises of preventing epenthesis from applying in the same way tp the 2p-plural- svibject/3p-dual and plural-object forms (0 t te k -xci , 0 t te t -ci ) . These proposals differ with respect to whether or not tone should be represented suprasegraen tally and whether contour tones exist in the underlying repre- sentation.
This can result in the attempt to fit all languages into a framework which is appropriate only for some languages.
Leben also indulges in this practice when hie extends his suprasegmental repre- ' sentation of tone to Northern Tepehuan and Mandarin Chinese.
Both Tone-copying and Tone-deletion do not involve contour tones as a whole, but rather the last level tone.