entities of the sentence.^208 In more standard approaches, the two kinds of organization are conflated into a single
formal expression along the lines of (25).
(25) A fox ate a grape.
∃xFOXx∃yGRAPEy(EATX,Y)The descriptive material consists of the predicates FOX, GRAPE, and EAT, plus their functional organization, which
consistsofthefirst twobeing arguments ofthethird.Thereferentialmaterialconsistsofthetwoexistentialquantifiers,
which make claims that two individuals exist.
Here we will break these two kinds of organization apart. (26) offers the barest start on the referential tier; we will
elaborate it gradually.
Thesyntax/phonology in(26) is a traditionallabeledbracketing; bynowI hopethereadercan easilyunderstandthisas
an abbreviation for the two independent syntactic and phonological structures. In the interests of space I have
collapsed the descriptive tier back into labeled bracket notation. (See section 11.8 for the equivalencebetween labeled
brackets and the tree notation.)
The new part here is of course the referential tier, which at the moment consists just of three indices, correlated with
the two Object-constituents and the Event-constituent of the descriptive tier. The presence of these indices on the
referential tierencodes theexistential claims that go withthesentence.Theindices1 and 2 correspond to∃x and∃yin
(25). Index 3 is the clai mthat the event of the fox eating the grape took place; this corresponds to existential
quantification over a Davidsonian event variable. In the terms of the theory of reference in Chapter 10, these indices
are the“indexical features”invoked by the sentence: they pick out what constituents of the sentence are intended to
correspond to individuals in the world as conceptualized by the speaker.^209
If the referential tier simply copied the indices out of the descriptive tier, it
PHRASAL SEMANTICS 395
(^208) This idea was primordially present in Jackendoff (1972). which proposed separate semantic components for“functional structure”(here the descriptive tier),“table of
coreference”and“modal structure”(here combined intothereferential tier),and“focus-presuppositionstructure”(here informationstructure—see section12.5). In 1972
this idea was decidedly kooky. For one thing, the formalization was unusable. More important, at that time we didn't know anything about tiers; it tookthe phonological
precedent to legitimate the idea.
(^209) Signedlanguages typicallyindividuate thecharacters ina discoursebyassigningeachoneitsown positionin signingspace(Lillo-Martinand Klima1990). Itis interestingto
speculate whether this phenomenon might be a direct encoding of the referential tier.