Niels Smit & Miriam van
Staden,
Universiteit van Amsterdam, Amsterdam,
The Netherlands
The
paper addresses the workings of the representations used in
Functional Discourse Grammar (FDG; Hengeveld & Mackenzie
2006), that formalise the semantic structures underlying
linguistic expression. It proposes a number of modifications
that aim to enhance notational consistency on the one hand,
and solve a descriptive problem on the other.
We argue for a
reappraisal of Dik’s original treatment of restriction as in
(1), because it enables a fully unified treatment of all
entity types.
(1)
αi: [(αj) (αi)Ø]
: [(αk) (αi)Ø] :
…
The restrictors in (1) are endocentric,
in the sense that the argument of the restricting predication
is coreferential with the layer as a whole. It can be shown
that this endocentricity is what enables the construal of
complex denotations through a recursive process of subset
restriction. Closer examination of the behaviour of nuclear
predications, however, reveals that portraying them as
endocentric layers poses a descriptive problem (cf. also
Mackenzie 1987):
(2)
ei: [(fi: walk) (xi:
man)A (ei)Ø]
‘event
of the man walking’
The endocentricity of ei in
(2) results in a scope ambiguity, since the repeated (ei)Ø
is not an argument of the predicate walk. Similar
problems arise also in the case of lexical heads, in
particular (but not exclusively) in representing the
difference between referent
modification and reference
modification (Bolinger 1967). It seems, then, that in
addition to endocentric layers, the formalism should also
allow for exocentric layers:
(3) αi: [(αj) (αi)Ø]
endocentric
(4)
αi: [(αj) (αm)Ø]
exocentric
The paper argues that it is no accident
that lexical heads and nuclear predications should behave
similarly, given that they constitute primitives in the view
of FDG, that are inserted from the Fund (lexicon), rather than
being generated by a semantic formulator. The remainder of the
paper is dedicated to an exploration of the properties of
endocentric and exocentric layers, and the consequences these
have for the FDG formalism. Among others, these include the
formal means to differentiate between restriction and definition,
the adoption of a new kind of variable, and the removal of
some redundant brackets.
|
