blob: f9b3ebc6c8fe8b84ed64a2dd85d4ca6d94aa3881 [file] [log] [blame]
# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html#License
#
# File: xh_xh_FONIPA.txt
# Generated from CLDR
#
# Pronunciation rules for isiXhosa.
#
# Author: mjansche@google.com (Martin Jansche)
#
# These rules transcribe isiXhosa into the phoneme inventory used within the
# NCHLT Speech Corpus (https://sites.google.com/site/nchltspeechcorpus/home).
#
# The rules were tested using the NCHLT-inlang isiXhosa pronunciation dictionary
# (http://rma.nwu.ac.za/index.php/resource-catalogue/nchlt-inlang-dictionaries.html).
# They correctly account for 14,999 out of 15,000 entries in the dictionary.
#
# The NCHLT 2013 phone set does not distinguish short and long vowels and does
# not indicate tone in any way. Transcription of tone is out of scope without a
# dictionary, since tone is generally not indicated in the orthography. Nasal
# clicks are not treated as separated phonemes in the NCHLT 2013 phone set and
# are transcribed as a sequence of nasal plus click instead.
#
# One minor notational deviation from the NCHLT 2013 phone set is that we use a
# tie bar within the complex (slack voiced) clicks, e.g. ɡ\u0361ǀ instead of ɡǀ, to
# avoid ambiguity and make the phoneme inventory uniquely decodable.
::Lower;
nyh ɲʰ;
n { tsh t\u0361ʃʼ;
tsh t\u0361ʃʰ;
tyh cʰ;
bh bʰ;
ch ǀʰ;
dl ɮ;
dy ɟ;
gc ɡ\u0361ǀ;
gq ɡ\u0361ǃ;
gr ɣ;
gx ɡ\u0361ǁ;
hl ɬ;
kh kʰ;
kr k\u0361x;
mh } [^l] mʰ; # <mhl> denotes /mɬ/ instead
nh nʰ;
ny ɲ;
ph pʰ;
qh ǃʰ;
sh ʃ;
th tʰ;
tl t\u0361ɬʼ;
ts t\u0361sʼ;
ty cʼ;
xh ǁʰ;
aa | a;
ee | e;
ii | i;
kc | c;
kq | q;
mm | m;
oo | o;
rh | r;
uu | u;
a a;
b ɓ;
c ǀ;
d d;
e ɛ;
f f;
g ɡ;
h h;
i i;
j d\u0361ʒ;
k kʼ;
l l;
m m;
n } g ŋ;
n n;
o ɔ;
p pʼ;
q ǃ;
r r;
s s;
t tʼ;
u u;
v v;
w w;
x ǁ;
y j;
z z;