Advances in Role and Reference Grammar

CLAUSE COMBINING IN NOOTKA^263

(a much weaker example, however, in that the medial verb forms may
freely occur in isolation).
All the instances mentioned by Kiparsky (1968) involve progressive
operator projection, and Haiman (1983:128, fn. 2) understandably criticizes
him for insisting on this directionality. Haiman suggests that "deletion goes
rightwards only when the identical element is on the left-hand side of each
conjunct", mentioning the augment (better: preverb) in Classical Greek,
the adverb of time in Tongan, and the tense prefix in Maasai. When a lan­
guage has switch-reference applying to coordinate clauses, its directionality
seems always to correlate with that of clause chaining more generally:
anticipatory switch-reference with regressive clause chaining, reminiscent
with progressive (Lenakel). Haiman & Munro (1983:xii) similarly point out
that the directionality of switch-reference in such clauses correlates with its
marker being either a suffix or a prefix on the verb: suffix with anticipatory,
prefix with reminiscent.
Our Nootka and Central Yana cases suggest that for progressive clause
chaining the strictly verb-initial order is a stronger factor, since it overrides
the embodiment of the relevant categories in suffixes. The Vedic injunc­
tive, with progressive operator projection in a verb-final language, but with
a deleted prefix, constitutes a discrepancy in the opposite direction. And
the Jacaltec regressive case correlates neither with its verb-initial order nor
with the manifestation of the categories in prefixes.
All the SVO languages mentioned have progressive directionality of
clause chaining, and many of them, such as Swahili and Lenakel, have the
tense categories localized in prefixes as expected. Others are isolating
(Kpelle) or have the expression of the categories diffused within the word
(Hebrew), so that the affixal correlations are unavailable. And areal norms
of clause chaining may doubtless arise that override language-internal
characteristics of category marking.

13.5 Generalized clause chaining

But is this really clause chaining? Longacre (1983:285, fn. 6) has foreseen

the possibility of progressive clause chaining, as well as a correlation with

verb position, in a note that deserves to be quoted in full:

One might speculate that chaining structures could occur in which: (a) lan­

guages are VSO (verb-subject-object); (b) there is a special sentence-ini­

tial verb; and (c) all other verbs in the sentence are some sort of (mor-