C'ali update: tonal phonology
|From:||Thomas R. Wier <trwier@...>|
|Date:||Sunday, May 23, 2004, 9:12|
Given the relatively low traffic on the list, and that I haven't done
any work on C'ali in a long time, I thought I might post some musings
about prosodic phonology in C'ali and Phaleran.
I have already discussed before the issue of stress assignment
in Phaleran (see archives), but I don't think I had decided at
that time what actual acoustic correlate most strongly defines
stress, but now I believe it is a basically unremarkable kind of
pitch-accent system: the pitch remains "high" up to and including
the stressed syllable (which syllable gets stress is determined
primarily by the moraic weight of the syllable, though it is also
sometimes underlyingly specified), after which the pitch markedly
decreases providing the effect of "stress".
C'ali, however, is somewhat more complicated system. Stress, as
such, does not exist. In its place is a tonal system,
which is partly underlyingly specified and partly predictable
according to verbal and nominal category. There is an underlying
contrast between H(igh) and L(ow) tone, though a large number of
morphs have no specification, receiving their surface tone by
phonological or morphological processes. Tone, when present, seems
always to be associated with the final vocalic mora of the root.
(1) H L H H H L L L
Underlyingly H: t|-anat-essa=luN => t|anatessaluN
'She ran off long ago'
(2) L L H L L L L H L
Underlyingly L: t|e-qwöl-essa-xela => t|eqwölessaxela
'She shot at them(III)'
L H H L L H H L
Cf. t|e-qwöl-ku-xela => t|eqwölkuxela
'It (e.g. some god) shot at them(III)'
(3) H L L H H H L L
No underlying tone: ?i-itsh'a-n-an-ni-tse => ?itsh'anannitse
'He may give it to us'
L L L L L L L L
cf. ?i-itsh'a-n-t|on-ni-tse => ?itsh'ant|onitse
'They(masc) may give it to us'
It's clear that the tonal specification for agreement suffixes is
important to understanding the surface tone structure, so below I
will repeat my chart of a previous post (almost a year ago) with
tonal values underneath:
AGENT PATIENT GOAL
Sg Pl Sg Pl Sg Pl
1st -ta -?V  -nu -kas ?V- -tse
H H L L L
2nd -(i)m -qwo -thæ -thæ twe- -(l)la
H H L L L
3rd I -(a)n -t|on -këi -xela me- -nar
H L L H L
II -(e)ssa -t|on -këi -xela me- -nar
L L L H L
III -ku -möra -xela -parsu kwe- -nar
H H H L L
IV -pha(ma) -ku  -ni -yö tis- -kwe 
H H L L L
V -ku -phai -ni -yö swi- -mi
H H L L L
(where roman numerals refer to noun classes)
A few generalization are to be pointed out:
(1) Agentive suffixes generally bear H tone. Exceptions are
feminine third person suffixes, and masculine and feminine
third plural suffixes (which are homophonous in all other
ways as well, an obvious reason for this feature of the paradigm).
(2) Patient suffixes generally bear L tone. Exceptions follow
from the same facts in the agentive: segmental homophony leads
to reanalysis as tonal homophony.
(3) Dative prefixes bear no underlying tone, while dative
plural suffixes bear underlying L tone. This situation follows
from two other facts: (a) prefixes never bear any tone, and tonally
always acquire the tone of the root (whether or not that has
underlying tone); (b) historically, the dative plural markers
had no tone either, but they are typically word-final (except
for tense clitics), and all word final syllables bear L tone --
they were thus reanalyzed as having L tone underlyingly at some
Some of the following processes are apparent from even this
small data set:
(1) Leftward spread: prefixes acquire tone of root.
(2) Word-final L: all final syllables must be L.
(3) No short contour tones. Underlyingly, there is a contrast between
long and short /u/~/u:/ and long and short /o/~/o:/. When a tone spreads
onto an underlyingly long /u:/ or /o:/, it may spread onto the adjacent
mora, but not the further mora. Now, the contrast of underlying short
/u/ and /o/ with /y/ and /ö/ is always neutralized to [y] and [ö]
respectively, while their long counterparts shorten. In this circumstance,
only underlyingly long /u:/ and /o:/ receive surface contour tones;
their fronted, underlyingly short counterparts may not do so.
(4) Morphophonologically, the sonorants /m n l r/ and the stop /?/ may
act as depressor consonants when in coda position, making any immediately
preceding tone a L tone:
H H H HH
saxmë-thei 'man-AGT' => saxmëthei
H H H H
saxmë-tsi 'man-PAT' => saxmëtsi
H L L L
saxmë-n 'man-OBL' => saxmën
Some morphemes seem resistant to this effect, however, as is clear from
the nasal consonants in (1) 'she ran off long ago'. The most straightforward
answer to this is that these depressor consonants lexically specify a
L tone in the UR, while the non-depressor 'depressor' consonants as in
(1) simply lack this specification. (When not in coda position, these
consonants are by definition not moraic, and thus cannot host any tone.)
One last fact about this system is that Phaleran tends to borrow words
from C'ali with underlying H tone into Phaleran with pitch accent permanently
docked to that segment. (On which, see archives)
Anyways, that's about it for now. Any questions or comments?
Thomas Wier "I find it useful to meet my subjects personally,
Dept. of Linguistics because our secret police don't get it right
University of Chicago half the time." -- octogenarian Sheikh Zayed of
1010 E. 59th Street Abu Dhabi, to a French reporter.
Chicago, IL 60637