LINGUIST List 4.744

Thu 23 Sep 1993

Sum: Projections

Editor for this issue: <>

Directory

1. "I. Plag", summary of query on projections

Four weeks ago I posted the following query to Linguist:

(BEGINNING OF QUERY)

Prepositions, verbs, or nouns seem not to subcategorize for maximal
projections with the same kind of head. I.e. prepositions do not take PP
complements, verbs do not take VP complements, nouns do not take NP
complements. In formal terms this would amount to a constraint on phrase
structure rules:

(1) * XP -> X XP

I would be happy if you could help me with the following
questions:

1. Is the above observation correct cross-linguistically? Are
there languages where the above generalizations do not hold?

2. Are there any language-theoretical arguments why the phrase
structure rule ruled out by (1) should be disallowed on principle? Maybe
some principle of UG at work here? Where in the literature has this
problem been discussed?

Any comments, hints, references, etc. will be most welcome.
I will summarize the answers for the list.

Many thanks in advance,
Ingo Plag

(END OF QUERY)


I would like to thank the many people who responded:

Carl Alphonce, Gregory Ward, Avery Andrews, Gregory K. Iverson,
John Phillips, Laurie Bauer, Adam Meyers, Sascha Felix, Henning
Lobin, Jan Odijk, David Powers, Larry Trask, Daniel Buering,
Martin Barry, Gerrit M. Rentier, Jan-Wouter Zwart, Marint
Haspelmath, Reinier Post, F. Murphy, John Paolillo,
oneilhusc.harvard.e, Georgia Green, Dale Russell, Andrew Barss,
Peter Gebert, Yehuda N. Falk, Kimberly A. Weiss, Sherri Condon,
Steven Schaufele, mike.maxwellSIL.ORG, Wayles Browne


In the following I will summarize the points that I feel to be
most important or interesting. I am looking forward to more
discussion.

Apparently, there are two camps of scholars:

One camp favors a restrictive phrase structure theory, and
sympathizes - at least to a certain extent - with the above
proposed constraint on phrase structure rules (not without
mentioning the problems of such an approach, I should add).

The other camp (the majority of respondents) argued against the
constraint.

In my query, I did not make any theoretical assumptions explicit,
assuming that more or less uncontroversial assumptions apply (ok, ok,
in a sense, everything is controversial in generative grammar these days,
but I thought let's pretend), and that people with differing assumptions
would come up with interesting ideas.

Consequently, several contributors discussed the problem of
theoreticalassumptions that are involved. What is a 'complement'? What
countsas a 'VP'? If we, e.g., deny the existence of phonetically empty,
but syntactically represented subjects of infinitives (like PRO in
GB theory), there will be lots of verbs that take VP complements.
English modals are - according to some analyses, which in turn
rest on their own assumptions - verbs that take VP as their complement.
 Many answers contained examples of PPs that obviously violate
the constraint (e.g. from under the bed, off of the shelf, bis
nach Hamburg, tussen de bomen door).
 Some people pointed out that [N NP] does occur in languages such
as German, Russian, and others with a genitive case. Even English
surface strings like [the destruction of the city] have been
analyzed as base-generated [N NP] with _of_ as an inserted case
marker.
 Sascha Felix mentions Speas (1990), who argues that heads
cannot take phrases with the same kind of head. (I did not get
around to check her arguments yet). Felix
also notes that under a DP analysis the [N NP] structure is ruled
out anyway, because N takes DP complements.
 Jan Odijk writes that "Stowell [1981] proposes the `Case-resistance'
principle (if I recall well), which basically states (as one part)
that a Case-assigning category cannot bear case. For this reason P
and V (Case-assigners) cannot take PP or VP complements. The
impossibility of N NP (in English) is attributed to a different
cause. Since N is not a Case-assigner, and NP must bear Case, this
is excluded in English." My question then is, how does Stowell handle
the problematic data mentioned above? In other words, how can a preposition
be case-assigning in one structure (i.e., P NP) and non-case-assigning
in another (i.e., P PP)?
 Gerrit Rentier mentions an article by Bennis and Hoekstra in_
The Linguistic Review_ 1984, these authors "note that a category X
seldom selects a complement of category X, and formulate the
"Unlike Category Condition" as a (universal) constraint to capture
this insight." In another paper by van Riemsdijk arguments and
counterarguments are discussed (full references?).
 Andrew Barss writes that "it is often argued that small-clause
complements of verbs are of the same category type as the
predicate of the complement, so V can take a (thematically
saturated) VP complement, as in:

3) make [/vp John run]"

Kimberly Weiss mentions that "for VP -> V VP, Larson (1988) (a
Linguistic Inquiry article) proposes that verbs with two objects
do in fact have this type of configuration. Therefore, the
sentence "I gave the book to John" would have the structure:

[ip [vp I [v' [v e ] [vp the book [v' gave (to) John]]]]]

where the verb "give" is base-generated in the lower VP with the
NP "John" as its complement and the NP "book" as its specifier.
It then moves through the higher V before landing in I (or
whatever nodes IP is composed of)."

Some of the references mentioned:

Jackendoff (1973) `the base rules for prepositional phrases' in
Anderson & Kiparksy, _A Festschrift for Morris Halle_.
Speas, M. (1990) Phrase structure in natural language. Dordrecht:
Kluwer.
Stowell, T. (1981) Origins of phrase structure. Cambridge: MIT
Press.

Many thanks again to all contributors. Your answers were of great
help to me. More discussion welcome.

<End of summary>

Ingo Plag
Philipps-Universitaet Marburg
Institut fuer Anglistik und Amerikanistik
Wilhelm-Roepke-Str. 6 D
D-35032 Marburg
Germany

e-mail: plagmailer.uni-marburg.de
Tel 06421-285560
Fax 06421-287020