Kijeb/sandhi

From FrathWiki

Contents 1 Kijeb sandhi 1.1 Kijeb sandhi table 1.2 Notes

Kijeb sandhi

The very restricted range of consonant clusters in Kijeb arose through sweeping sandhi changes. Most of these changes were assimilatory, like the pervasive regressive voicing assimilation, but some were dissimilatory, like the change of geminate nasals into nasal + homorganic voiced stop. In the table below the first ('leftmost') consonants of input clusters are shown in the leftmost column, while the second ('rightmost') consonants of input clusters are shown in the top row. Outputs are shown in the intersecting cells. Where convenient mergers have been indicated with merged cells, and unchanged outputs are indicated with bold type. Character values are as IPA, except that kw, gw, ñw are probably true labiovelars, y is /j/ and consonant + y are realized as palatalized consonants. For ky, gy, ŋy, xy this probably means they were realized as [c, ɟ, ɲ, ç].

Labiovelars as the first element of an underlying cluster behave like velars, and palatalized sounds as the first element of an underlying cluster behave like their non-palatalized counterparts. Second elements retain their palatalization, and non-palatalized second elements don't get palatalized by palatalized first elements, i.e. palatalization is subject to regressive assimilation, just as voice.

Maybe I could have a distinction so that labiovelars behave like labials before non-palatalized sounds but like velars before palatalized sounds, but I doubt there is any sound phonetic basis for that. Maybe a dialectal difference between labiovelar first element > velar and labiovelar first element > labial?

I'm not sure what happens to underlying combinations of palatalized velar + labial (*ky+p etc.), velar + palatalized labial (*k+py etc.) and palatalized velar + palatalized labial (*ky+py etc.), but my hunch is that at least the two latter become palatalized velars (ky etc.)

Kijeb sandhi table

	p	b	m	f	w	t	d	n	s	r	k	g	ŋ	x	y	#
p	p	b	mb	f	p	pr	br	br	p	pr	kw	gw	ŋgw	f	py	Ø
b					b					br					by	u[1]
m	mp	mb		mf	m	nt	nd		ns	mr	ŋk	ŋg		ŋx	my	n
f	p	b	m	f	f	fr	wr		f	fr	kw	gw	ŋw	f	fy	Ø
w					gw/b[2]	tw	dw	nw		wr					wy	u[1]
t	tp	db	dm	tf	tw	st	zd	dn	t	tr	tk	dg	dŋ	tx	ty	t
d					dw					dr					dy	r
n	mp	mb		mf	nw	nt	nd		ns	nr	ŋk	ŋg		ŋx	ny	n
s	sp	zb	sm	sf	sw	st	zd	sn	s	sr	sk	zg	sŋ	sx	sy	s
r	rp	rb	rm	rf	rw	rt	rd	rn	rs	dr	rk	rg	rŋ	rx	ry	r
k	kw	gw	gw	kw	kw	kr	gr	gr	k	kr	k	g	ŋg	x	ky	t
g					gw					gr					gy	Ø[1]
ŋ	mp	mb		mf	ŋw	nt	nd		ns	ŋr	ŋk	ŋg		ŋx	ny	n
x	f	w	ŋw	f		xr	r		x	xr	k	g	ŋ	x	xy	Ø
y	py	by	my	fy	wy	ty	dy	ny	sy	ry	ky	gy	ny	xy	gy	i[1]

Notes

• ↑  Whenever final **y or **w arose it colored a preceding **a or **i to u before disappearing itself.
• ↑  Pre-Kijeb *ww possibly gives b in Linjeb. Alternatively or simultaneously it is possible that underlying *uwwV/*wwuV give ugV/Vgu while *uwV/*Vwu give ubV/Vbu.