competence, so be it. For my own taste, such interaction between the two theories is the proper way for linguistic
research to proceed in the future.
6.11 Appendix: Remarks on HPSG and Construction Grammar
HPSG and Construction Grammar (CG) lead to a consolidation of grammatical phenomena similar to that developed
here, though in a different formalism. Pollard and Sag's (1994) exposition of HPSG is in fact far more rigorous than I
have been here, dealing with many of the tricky details of how syntactic combination is accomplished. Much of that
machinery appears to be readily adaptable to the present notation.
Beyond the notation, though, there are some interesting conceptual differences between the present approach and
HPSG/CG.
- HPSG and CG both insist on the primacy of thesignin Saussure's sense: they require lexical items to be
complexes of sound (phonology), meaning (semantics), and grammatical properties (syntax). In particular, I
have mentioned several times the view of Construction Grammar and Cognitive Grammar that all syntactic
structure is inherently meaningful. Thus neither HPSG nor CG admit the possibility of“defective”lexical
items that lack one or twoof the threecomponents; such items have come up repeatedlyas important to the
present analysis. (HPSG, however, allows components to be null rather than absent.)
I suspect this insistence on the sign has arisen fro ma rejection of syntactocentris m. It is a way of giving syntax a
cooperative rather than dominantrole intheformation ofsentences, a stancewithwhichI completelyagree.However,
insisting on the integrity of the sign is not the only way to achieve this stance. The prototypical word is indeed a
Saussurean sign; but no argument is ever given that the entire grammar must be so uniform. That is, the integrity of
thesignisbasicallya stipulation. Ifweabandonit,lexical structuresdo becomemoreheterogeneous. But thisopens up
interesting opportunities for analyses of the sort explored here. As we have seen, abandoning the rigid integrity of the
sign hardly forces us to fall back into the clutches of syntactocentrism; in fact it draws us away from them.
- Because unification in HPSG/CG is always complete sign by complete sign, combinatoriality must proceed
more or less in lockstep among the threecomponents (though HPSG allows constituentsto be rearranged in
thecourse of combination). Such correspondence is fairlynatural between syntax and semantics, where it has
been worked out intensively. However, the syntax–phonology connection (section 5.4), which has been
studied hardly at all within HPSG and CG, is far more problematic. And when we enter the world of the
phonological tiers (which to my knowledge has never been addressed in these theories), constituent
organization bears no resemblance to syntax and semantics. (Some of the phenomena to be