ABSTRACT
The!isalwaysaproblemforthosedescribingYUmanlanguages.
Somany!'sappeartobepredictablethatitisfrustratingnottobe
abletoanalyzeallofthemaway.!'sarenotstressedandfrequently
deleteinfastspeech.Theyareoftenheardasshort!:•_!,or~(appar-
entlydependingupontheconsonantalenvironment);maeyofthe
unstressedMojavevowelsIwriteas_!,~~or!:areprobablyana.lyzable
as!'s.(Idonotwritesome!'swhichareinsertedatmorphemeboun-
dariesbyregularprocesses.SeeunderWordStructurebelow.)
Vowel-initialwordspronouncedinisolationhaveamarkedaspi-
ratedonsetwhichisindistinguishablefromtheconsonantalphoneme.!!·
SUchonset_!!'s,unliketheunderlyingones(usually),arelostwhen
wordsareruntogetherinrapidspeech.Thusthesentence
however, have nonfinal stress or apparently equal stress on more than
one syllable. As noted by Kroeber and Wares, Mojave words also have
and
morpheme boundary, as described under Word structure below), but usually
would be heard as [hisam] and [ka?a:k] respectively. Sometimes such de-
ku~ 'what' and
addedsubjectcaseending-!: ku~-~[ku~]'wbat(subj.)'
ka:du:Cf-~[ka:du:Cf]'wbatkind(subj.)' Similar~,thelocativeending-17isneverheardafteranounwhich endsin-l7 :
mat'bhwel'Y'Parker',mat'bhwel'Y-l'Y[mat'bhwel'Y]'inParker'
Bysimilarprocesses,!isusua~lostbeforetheotherlabial
sounds!!andRland!iso:f'tenlostbefore!· Thesounds!!and!alternatetree~inagreatmanyMojave
wordsandinseveralgrammaticalmorphemes.Thisaltemationisdis-
cussedinsection1.224ofChapterr.
