foci per clause. But the foci in each clause are not independent of each other; they for mordered pairs. We can
paraphrase the reply rather closely by using topic-comment form:
(62) Well, as for FRAN, SHE danced with DAVID; as for ELI, HE danced with FANIA; as for ELIZABETH,
SHE danced with BARBARA.So these paired foci behave a bitlikea list of topic-focus pairs. Should we treat thefirst stressed element in each clause
of (61B)(Fran, Eli, Elizabeth)as a Focus or as a Topic? It is somewhere in between.
In the proper setting, it is just barely possible to embed paired foci under a larger topic:^222
(63) As for bad reLAtionships [topic], JOHN hateshis MOTHER, FRAN hates her SISTER, and my BROTHER
can't STAND[contrastive focus]ME.This suggests that thefirst element of a paired focus construction has a different status from an ordinary topic. We
might call it a“first focus,”preserving the generalization that wh-phrases and the phrases answering the mare always
foci. Or we might call it a“subtopic,”emphasizing its topic-likerealization in syntax and phonology. The terminology
doesn't really matter.
What matters is this: There are three roles in information structure that can be coindexed with constituents in the
descriptive tier, each of which displays characteristic prosodic and syntactic effects: Focus (obligatory), Topic
(optional),and First Focus(optional).Theremainingmaterial,whichdoes notnecessarilyform a unified constituent,is
Common Ground. So the formation rule for the information structure tier can be simply (64).
(64) (Topic) (Common Ground) (First Focus) FocusCompared toalltheother structures wehaveinvestigated, thisistriviallysimple. Onemightwonderwhether itpermits
any sort of embedding, say whether relativeclauses or indirect quotations can have theirown subordinate information
structure. I a mnot fa miliar with any proposals to this effect in the literature. Though the question see ms worth
exploring, I will not do so here.
Let us see how (64) can be coindexed with the descriptive tier. As already mentioned, Topic and Focus map to
constituents, but Common Ground does not appear to. For instance, consider (65).
414 SEMANTIC AND CONCEPTUAL FOUNDATIONS
(^222) I am grateful to Jong Sup Jun for bringing such examples to my attention, in the context of a discussion of topic-marking in Korean.