status of words; HPSG and CG

[Berthold Crysmann]

Luis Casillas wrote:

> On Thu, Aug 16, 2001 at 04:08:10PM -0700, dmellow at sfu.ca wrote:
>
> > As I understand it, HPSG assumes strict lexicalism: word structure and
> > phrase structure are governed by independent principles (a quote taken
> > right from the Stanford web page). CG makes a different assumption
> > (Goldberg, 1995, p.7):
> >
> > ?In Construction Grammar, no strict division is assumed between the
> > lexicon and syntax. Lexical constructions and syntactic constructions
> > differ in internal complexity, and also in the extent to which
> > phonological form is specified, but both lexical and syntactic
> > constructions are essentially the same type of declaratively
> > represented data structure: both pair form with meaning.
>
> I think there is a risk of confusing two ideas of lexicalism here:
>
>   (a) syntax and morphology are governed by independent principles;
>   (b) a language is best described by giving 2 disjoint sets:
>       - a set of fully general grammatical rules;
>       - a list of idiosyncratic lexical items.
>
> As I see it, your definition of lexicalism in the first paragraph above
> has to do with (a), while the Goldberg quote about CG is essentially
> a denial of (b). But I believe it's possible to deny (b) while still
> holding on to (a).
>
> > It is not the case, however, that in rejecting a strict division,
> > Construction Grammar denies the existence of any distinctly
> > morphological or syntactic constraints (or constructions). Rather,
> > it is claimed that there are basic commonalities between the two
> > types of constructions, and moreover, that there are cases, such as
> > verb-particle combinations, that blur the boundary.?
>
> What does it mean "to blur the boundary"? You can simply have listed
> units which constrain both levels of organization simultaneously, and
> thus are not properly syntactic units nor morphological units.

Essentially, this is the idea of coanalysis that I have been trying to
develop over the past few years. In a monotonic framework like HPSG such a
multimodular view differs quite drastically from the kind of simultaneous
licensing advocated by e.g. Autolexical Syntax, where one level of
description (typically morphology) may win out, as far as linearisation is
concerned (Sadock,1990).
Although the main evidence for integrating coanalysis in HPSG comes from
European Portuguese cliticisation, one of the studies to support the claim
that parts of a single lexical sign get represented on more than one
syntactic atom, was based on syntactically discontinuous  inflection in
Fox, another language of the Algonquian family. The paper also features
some discussion on lexical integrity which I believe is still preserved by
the brand of coanalysis I advocate. In essence, as morphological objects
and their composition differ from those in syntax, it is in general not
possible (and I don't think even desirable) to simply open up morphological
structure to syntactic licensing. One of the main obstacles is that
affixes, in realisational morphology, are  phonology-only, while syntactic
rules demand access to at least categorial information. The way I solve it,
is to provide the syntactic representation by means of a lexical
construction, and only align the phonology of affixal material with the
syntactic objects so introduced (technically, I use multiple
lexically-introduced domain objects, building on a proposal of Andreas
Kathol's).  As a result, most of the internal morphological representation
of the lexical sign is hidden from the syntactic component.

Cheers,

Berthold

Refs:

@InCollection{crysmann_b99csli,
  author =  {Berthold Crysmann},
  title =  {Morphosyntactic Paradoxa in {Fox}},
  booktitle =  {Constraints and Resources in Natural Language Syntax
                  and Semantics},
  year =  1999,
  editor =  {Gosse Bouma and Erhard Hinrichs and Geert-Jan Kruiff
                  and Richard Oehrle},
  series =  {Studies in Constraint-Based Lexicalism},
  address =  {Stanford},
  publisher =  {CSLI publications},
  pages =  {41--61}
}

@InCollection{crysmann_b00coord,
  author =  {Berthold Crysmann},
  title =  {Clitics and Coordination in Linear Structure},
  booktitle =  {Clitics in Phonology, Morphology and Syntax},
  publisher =  {John Benjamins},
  volume =  36,
  year =  {2000},
  editor =  {Birgit Gerlach and Janet Grijzenhout},
  series =  {Linguistik Aktuell/Linguistics Today},
  pages =  {121--159}
}

@InCollection{crysmann_b00csli,
  author =  {Berthold Crysmann},
  title =  {Syntactic Transparency of Pronominal Affixes},
  booktitle =  {Grammatical Interfaces in Head-driven Phrase
                  Structure Grammar},
  publisher =  {CSLI Publications},
  year =  2000,
  pages =  {77-96},
  editor =  {Ronnie Cann and Claire Grover and Philip Miller},
  series =  {Studies in Constraint-based Lexicalism},
  address =  {Stanford}
}



>
>
> I think the central point here is that the units of morphological
> and syntactic combination (morphemes, words, phrases) don't have to
> match up in any simple manner with the listed form-meaning pairings
> (constructions). The "independent principles of morphology and syntax"
> constrain the units of the first kind; the units of the second kind
> don't have to correspond in any simple way to the first. Nothing in
> the HPSG framework stops the listed units from mixing levels and/or
> being very schematic, and such a move does not compromise lexicalism in
> the (a) sense above. The crucial thing for (a) is that, after you've
> composed the constructions to create a well-formed sign, you end up with
> something which accords to the constraints set by a lexicalist grammar
> (in the (a) sense).
>
> Thus, I fail to see the incompatibility that you are asking about. On
> the one hand, you have the structural assembly of morphemes into words
> and words into phrases; on the other, the composition of constructions
> into signs. These are two different things and don't have to be strictly
> parallel. Constructions constrain some combination of phonology,
> morphology, syntax and semantics simultaneously, but still these levels
> can have their own independent organizational principles. You can
> assemble lexicalistically well-formed sentences out of listed units
> which are not all strictly lexical items.
>
> --
> Luis Casillas
> Department of Linguistics
> Stanford University

--
Berthold Crysmann
Deutsches Forschungszentrum Kuenstliche Intelligenz (DFKI) GmbH
Stuhlsatzenhausweg 3
D-66123 Saarbrücken