268 WILLIAM H. JACOBSEN, JR.
- An exception pointed out already by Sapir (1911a: 19) is that imperatives may take
suffixes indicating first person object in addition to second person subject. Nichols
(1986:69, Table 3) appropriately catalogs Nootka for these head-marking charac
teristics. The tally 0/1 under Clause recognizes this single pronominal argument
(note that the explanation in the Legend should be reversed to read: Number of
dependent-marked patterns/number of head-marked patterns). Nootka gets a
"full total" of 0/4 and "full type" of -4, where negative numbers correspond to
head-marked patterns. Related Makah (80, Table 5) gets -5, doubtless because its
verbs can bear two pronominal suffixes when each of them is first or second per
son. Nichols also notes (81, 104) that head-marking morphology favors verb-initial
word order. She cites (64, 109) a Makah example (from Jacobsen 1979a: 113)
meaning 'T'm tired because I overworked", but I think the superscript markers
should be differently placed:^ Hon the first word because it is the auxiliary verb of
the main clause,^ M on the last word because it is the subordiate clause and is
marked as such by -h (corresponding to the Nootka Subordinate with third person
-qa* [section 10.1 below]). This is thus a dependent-marked construction. The first
word, centering on a suffixal verb meaning "to have as a reason for ...-ing", lexi
cally implies the subordination in that it is transitive, taking a factive complement
clause. - The verb-final characteristic is not quite universal with switch-reference. It per
tains to the widely spread type found in the Americas and New Guinea, in which
the index is a suffix on the verb, and where, if the clause is not subordinate, the
marking looks ahead to sameness or difference of subject in the following clause.
In agreement with Givón (1983:77) I call this anticipatory switch-reference. Com-
rie's (1983:23-25, 35) incisive discussion of directionality of switch-reference con
cluded that it is marked always on the dependent (= subordinate or cosubordi-
nate) clause. In Australian languages switch-reference is generally marked only on
subordinate clauses, which follow their main clauses, and hence the marking
refers backwards to the preceding clause. (Nevertheless these languages have suf
fixal markers and verb-final word order.) I call this directionality reminiscent
switch-reference (= Givón's non-anticipatory ). An exceptional pattern is found in
(Austronesian) Lenakel (Lynch 1983), which has verb-medial (SVO) word order,
and displays reminiscent switch-reference on coordinate clauses, with the marker
prefixed to the verb. Moreover, this language comes close to exemplifying the
possibility foreseen in Jacobsen (1967:241, 1983:168) of having a single marker for
retention opposed to its absence indicating switch. (But this is not quite the ideal
case, since the marker replaces the pronominal prefix, rather than accompanying
it.) The non-canonical version of logophoric switch-reference seen in African
(Benue-Congo) Gokana (Comrie 1983) also shares the characteristic of being
reminiscent (from subordinate clauses) in a verb-medial language and of marking
retention, but with a suffix on the verb. - For other structural synopses of Nootka see, on the one hand, Sapir's (1911a: 15-
- early report, contrasting the language to Kwakiutl, and on the other, Rose's
(1981:ii-iii, 31-34, 192-194, 343-347) recent typological surveys, including areal
and genetic considerations.