Consonant Clusters at the Beginning of Words (I) Saul Epstein Wed, 5 Nov 1997 20:29:06 -0600 Just to narrow things down and start somewhere, let's look at stop+continuant clusters at the beginnings of words (or syllables). To simplify things, let's ignore aspirated and nasal stops for the time being. (And, by "continuant" I mean those sounds which aren't stops.) In the charts below, clusters marked with an asterisk (*) are currently unattested -- that is, they don't appear in the data so far. I've left out all the /w/ sounds for now because I think they're all fricatives, but I'm not sure. They may merit special consideration. ;-, STOP+TAP, STOP+APPROXIMANT pr pl *py br *bl *by tr tl ty dr *dl *dy kr *kl ky *gr *gl *gy *qr *ql *qy Apparently, any stop can cluster with /r/, regardless of voicing. The data for stop+l clusters are inconclusive. /pl/ and /tl/ each occur once, leaving room for the prohibition of clusters with voiced stops or stops articulated towards the back of the mouth. Likewise the data regarding stop+y clusters are inconclusive. I can only say that /qy/ is a physiological challenge for Terrans. STOP+FRICATIVE *pf *pv *ps *pz *pc *pj *px *ph *bf *bv *bs *bz *bc *bj *bx *bh *tf tv (ts) *tz (tc) *tj *tx th/(th) *df *dv *ds *dz *dc (dj) *dx dh/(dh) *kf *kv (ks) *kz *kc *kj *kx kh *gf *gv *gs *gz *gc *gj *gx *gh *qf *qv *qs *qz *qc *qj *qx *qh The data are even spottier here. Out of the eight attested clusters, two (possibly four) are phonemes, and two (possibly three) are actually aspirated stops. Just to get a potential rule set going, I'd like to propose that any word-initial cluster of a stop and a fricative is permitted provided the two consonants have the same voice value -- either both voiced or both voiceless. This is almost a required rule, because without a pause or a vowel between the two consonants, it just naturally happens that the voicedness or voicelessness of one voices or devoices the other -- especially at the beginnings of words. So the permitted stop+fricative clusters would be: pf ps pc px ph bv bz bj tf (ts) (tc) tx th dv dz (dj) kf (ks) kc kx kh gv gz gj qf qs qc qx qh I know: that eliminates /tv/. If people are attached to it, we can work to preserve it... Rob, this would modify your proposition that any stop can be followed by any continuant. That does seem true for the approximants (though we have no data for /rr/) and /r/. Except of course, that /qy/ is highly unlikely. It would nearly always come out as some form of /k/. Remember that I'm only looking at possibilities regarding clusters at the beginnings of words. Different rules will govern clusters at the ends of words, as well as what sequences are permitted along syllable boundaries. Also, the /'/ phoneme seems to act largely as a "discontinuant," negating whatever rules would normally apply to the sounds immediately before and after. -- from Saul Epstein liberty uit net www johnco cc ks us sepstein "Surak ow'phaaper thes'hi thes'tca'; thes'phaadjar thes'hi suraketca'." -- K'dvarin Urswhl'at