Featured Linguist!

Jost Gippert: Our Featured Linguist!

"Buenos dias", "buenas noches" -- this was the first words in a foreign language I heard in my life, as a three-year old boy growing up in developing post-war Western Germany, where the first gastarbeiters had arrived from Spain. Fascinated by the strange sounds, I tried to get to know some more languages, the only opportunity being TV courses of English and French -- there was no foreign language education for pre-teen school children in Germany yet in those days. Read more



Donate Now | Visit the Fund Drive Homepage

Amount Raised:

$34674

Still Needed:

$40326

Can anyone overtake Syntax in the Subfield Challenge ?

Grad School Challenge Leader: University of Washington


Publishing Partner: Cambridge University Press CUP Extra Publisher Login
amazon logo
More Info


New from Oxford University Press!

ad

What is English? And Why Should We Care?

By: Tim William Machan

To find some answers Tim Machan explores the language's present and past, and looks ahead to its futures among the one and a half billion people who speak it. His search is fascinating and important, for definitions of English have influenced education and law in many countries and helped shape the identities of those who live in them.


New from Cambridge University Press!

ad

Medical Writing in Early Modern English

Edited by Irma Taavitsainen and Paivi Pahta

This volume provides a new perspective on the evolution of the special language of medicine, based on the electronic corpus of Early Modern English Medical Texts, containing over two million words of medical writing from 1500 to 1700.


Email this page
E-mail this page

Review of  Argument Structure


Reviewer: Andrew McIntyre
Book Title: Argument Structure
Book Author: Eric J. Reuland Tanmoy Bhattacharya Giorgos Spathas
Publisher: John Benjamins
Linguistic Field(s): Linguistic Theories
Syntax
Book Announcement: 19.3650

Discuss this Review
Help on Posting
Review:
EDITORS: Reuland, Eric; Bhattacharya, Tanmoy; Spathas, Giorgos
TITLE: Argument Structure
PUBLISHER: Benjamins
YEAR: 2007

Andrew McIntyre, English Department, University of Neuchatel

INTRODUCTION
This is a 240 page collection of ten articles on subjects (mostly) connected to
argument structure. Nine of the articles employ Minimalist assumptions, so the
book is not meant to be an overview of very different perspectives on argument
structure, unlike Butt & Geuder (1998). I will suggest that the essays vary in
quality from very worthwhile work to work which should clearly not have been
published in its present form. The next section of this review discusses the
articles individually. The final section briefly comments on the book as a whole.

I should apologize in advance for the possibly repetitious or disjointed
presentation of the discussion of certain chapters. I decided that this was a
necessary price to pay if I am to live up to the LINGUIST List's ideal of strict
separation of synopsis and evaluation.

SUMMARY AND DISCUSSION OF INDIVIDUAL CHAPTERS
1. ''Do verbs have argument structure?'' by Tor Afarli
The article argues for the constructionalist approach to argument structure:
verbs *get* argument structure from the syntactic context, rather than
*projecting* it. Contra the constructions-as-primitives view of e.g.
Construction Grammar, Afarli's view sees itself as ''neo-constructionalist'' in
that constructions are decomposed using functional heads (specifically Bowers'
(1993) Pr(edication) head is used to introduce agents and the arguments in small
clause configurations).

Afarli gives two arguments for adopting the constructionalist view: (i) The
ability of native speakers to use previously unfamiliar verbs in constructions;
(ii) the occurrence of arguments which are not associated with the inherent
conceptual content of the verb. Thus, Norwegian can say ''That evening I danced
myself a sweetheart.'' Clearly, the sweetheart-acquisition aspect of the meaning
is not part of the basic meaning of ''dance''. Afarli infers that this aspect of
the meaning comes not from the verb but from the construction in which it appears.

I must say that neither of the arguments given for the constructionalist
position is enlightening. The claim that only constructionalists can handle the
data associated with argument (ii) is wrong. There is a copious literature which
makes use of rules/operations operating at some lexical or semantic level which
enrich the basic meaning of a verb, which can result in the addition of new
arguments (say Jackendoff 1990, Pinker 1989, Rapapport Hovav & Levin 1998,
Wunderlich 1997 and scores of others). Applied to Afarli's example, a linguist
could posit a lexical/semantic rule changing the basic meaning of ''dance'' in
such a way that it means ''X causes Y to have Z by dancing'' (see e.g. Pinker's
1989 analysis of ditransitives). Had Afarli shown weaknesses in a lexicalist
treatment using such strategies, rather than simply ignoring the lexicalist
literature dealing with similar problems, this article would have been far more
successful.

Now consider argument (i), the claim that the use of previously unfamiliar verbs
in constructions supports a constructionalist approach. This argument, which
resembles one made by Goldberg (e.g. 1995:120), is discussed with reference to
the attested example ''He Russelled a Frege-Church''. Afarli claims that a
lexicalist approach would have trouble with this because: ''In a lexically driven
model, proper names that are used as verbs in this way must be listed in the
lexicon with an argument structure specification that stipulates that they have
the required Theta-grid. In addition, they must be listed as proper names. Since
any proper name may be used as a verb in a similar way, all proper names must
have a corresponding double specification. In other words, the argument
structure specification has no explanatory power and is reduced to a stipulative
description of whatever syntactic configuration the lexical element occurs in''
(p.7).

I am not a lexicalist, but would sympathize with lexicalist readers who find
this dismissal of their position unfair. A misunderstanding of lexicalism here
is the implication that the multiple lexical entries required for different uses
of a word are just listed in the permanent lexicon with no attempt to capture
their productivity. However, the whole point of lexicalist studies is that they
posit lexical operations which take an item from the permanent lexicon and alter
its lexical specifications (regarding e.g. syntactic category, meaning, argument
structure) to create a derived (and not necessarily permanent) lexical entry.
For Afarli's example, lexicalists could say that English has a productive
lexical rule forming denominal verbs expressing the affecting of an object in
the manner associated with the entity named by the incorporated entity. (The
exact characterization of the semantics is an equally difficult problem in all
theories.) This verb would encode an agent and a patient in its meaning.
Lexicalists could then use their preferred argument linking devices to predict
the realization of these arguments in syntax. Such approaches will doubtless be
called 'stipulative', but are lexical rules more stipulative than selectional
restrictions involving functional heads that would be needed in any
neo-constructionist account that wants to capture the observed data without
overgenerating?

A final criticism I have of Afarli's approach is that there is a discrepancy
between the radically constructionalist perspective that the article purports to
be adopting and the analyses it actually proposes. Afarli's trees look only
moderately neo-constructional, since lexical verbs are (at least notationally)
represented as heading VPs containing DP or PrP complements. For it to be true
that verbs obtain *all of* their argument structure from syntax, the
constituents labeled as VPs would presumably have to be understood as items with
lexically listed interpretations and empty slots into which verbs and other
objects are inserted, as in Construction Grammar. Afarli is not explicit on this
crucial point.

2. ''Projecting argument structure: The grammar of hitting and breaking'' by Nomi
Erteschik-Shir and Tova Rapoport
The approach to verbal argument structure espoused here partly resembles Hale &
Keyser (e.g. 1997), but introduces several original features. It assumes that
there are three possible components in verb meanings: Manner
(=means/manner/instrument), State and Location. (Simplex) verbs can have at most
two components. Here are some illustrations of the use of the components in
describing verb meanings:

(1) BREAK: Manner ('forceful means'), State ('broken')
(2) HIT: Manner ('forceful means'), Location ('point of contact')
(3) CUT: Manner ('sharp instrument'), State ('cut')

The components can project in the complement of V, and together with the verb,
form a predicate over the DP in spec,VP. Manners can be N-projections which
combine with V to form an activity VP with an agent as specifier.
States/locations project APs/PPs, forming with V a state/location change VP with
a theme in spec,VP. The principle of Full Interpretation says that every
component must be interpreted and that every V projection involves a distinct
component. Components need not project. If they do not, they are interpreted as
modifiers.

I will briefly discuss some of the authors' analyses.

BREAK: ''The vase broke'' involves the projection of the state component in (1).
The manner component is a modifier of the event, either of the state change or
of a causing event if we add an additional V-shell with a causer.

CUT: Obligatory transitivity with ''cut'' is due to a need to interpret the
components in (2): The theme is needed for the state to be interpretable. The
agent is obligatory since instruments require a ''referential wielder'' (p.23).

HIT: ''The car hit the wall'' involves the projection of Location (as a silent P),
with the manner as a modifier. (Extra evidence for the authors' unaccusative
structure might be gleaned from the German translation ''der Wagen ist gegen die
Mauer geprallt'', featuring a ''be''-perfect and an overt P.) In ''Jane hit the ball
to the other side of the field'', the Location component is said to be
interpretable because a contact relation subsists between the ball and Jane.
(Since the sentence could describe hitting the ball with a stick, the authors
would I guess have to allow Jane represent an instrument by metonymy.)

Overall the approach seems to have considerable elegance, but I have a couple of
criticisms of specific analyses. It is unclear to me whether they doom the
proposal as a whole. Firstly, in the analysis of ''break'' described above, the
manner component 'forceful means' seems unmotivated. Fine glass can break gently
if heated. The difference between ''smash'' and ''break'' seems to be precisely that
''smash'' has the forceful manner component while ''break'' lacks it.

A second criticism of the approach concerns the use of copies of DPs interpreted
in both the lower and the upper V-shell (i.e. as both agent and theme) in cases
like (4). Even if one allows DP copies in multiple theta positions in principle
(as do Manzini & Roussou (2000) and Ramchand (e.g. 1997) in different empirical
domains), I would think twice before using this device in (4). It seems unable
to capture the well-known contrast between unaccusative resultatives like (4b)
and reflexive resultatives like ''Jane ran herself to death''. (See McIntyre (2004
and refs.) on how the tighter unity between the activity and motion subevents
produces the more compact, unaccusative syntax in (4b).) As for (4a), we find
nothing like *''Jane hit herself the table''. One can debate about whether the
argument-structurally crucial fact about (4a) is that she hit something against
the table or that the table is a patient, but it appears unlikely that Jane is
grammatically represented as a theme in this structure if the sentence means
that she hit the table with a stick.

(4)a. Jane hit the table.
b. Jane ran to school.

3. ''The argument structure of the dative construction'' by K.A. Jayaseelan
Dravidian languages like Malayalam have dative constructions literally glossing
as follows:

(5) Me(dative) happiness/hunger is. [experiencer dative]
'I am happy/hungry.'
(6) Me(dative) money is. [possessor dative]
'I have money.'

The construction in (5) with a copula, a dative experiencer and a predicate
nominal is the Malayalam equivalent of English predicative adjective
constructions. (Jayaseelan argues that Malayalam lacks true adjectives.) (6)
illustrates Malayalam's use of a dative possessor with 'be' where English
normally uses 'have'-structures (modulo things like ''there is no lid/end/sequel
to this'' noted by Jayaseelan).

Jayaseelan proposes the structure in (7) (cf. e.g. Freeze 1992, Kayne 1993, den
Dikken 1997) as the basic structure from which the constructions in (5) and (6)
and their English equivalents are derived. (Since subscripts are unavailable
here, labels for constituents are marked with a postposed colon.)

(7) BE [KP: K [PP: DP(possessor/experiencer) [ P NP ]]

English 'have'-structures are created by incorporation of K into BE. Adjectives
are created by incorporating N into K via P. If K remains unincorporated, it can
attract the possessor/experiencer DP to its specifier; where it receives dative
case. Jayaseelan suggests (p.47) that a destabilized case system (as found in
English but not in Malayalam) is a precondition for the incorporation phenomena
involving K.

In my opinion, the attempt to connect the existence of 'have'-structures and of
adjectives to the absence of a stable case system in a language is questionable
given that 'have' and stable case marking coexist in e.g. German, and many more
case-rich languages have adjectives (Icelandic, Russian, Latin...). The proposal
about adjectives remains programmatic. It would take a lot of work to
reinterpret the tests distinguishing adjectives from prepositions in languages
with both (say English modification: ''very high'' vs. ''right up'') in accordance
with this theory.

4. ''Syntactic categories and lexical argument structure'' by R. Amritavalli
Amritavalli discusses problems partly similar to those treated in Jayaseelan's
chapter, but focuses on another Dravidian language, Kannada. This language
expresses possession as in (6) above (''this(dative) is a lid'' instead of ''this
has a lid''). Kannada is also argued to lack the categories P and A. Instead it
has N and Case. Amritavalli (p.52) suggests that, in language change, case
markers either become a new syntactic category (P) or get absorbed into existing
lexical categories (e.g. incorporation into 'be' yields 'have'). ''If so, we
deduce that languages with case-markers do not have P, and do not have a verb
'have'. [...] Neither should these languages have adjectives [...]''

Section 4 tries to explain Emonds' (1985:40, note 18b) generalization that
languages with serial verbs ''often apparently lack PP structures''. Amritavalli
notes that Kannada has a serial construction allowing perfect and progressive
participles, literally glossing as follows:

(8) a. I mango pluck(perf.part) wash(perf.part) cut(perf.part) ate
'I plucked, washed, cut and ate a mango.'
b. I mango pluck(present.part) wash(present.part) cut(present.part) danced.
'I danced, plucking, washing, cutting a mango.'
c. She dancing singing flowers strewing came
'She came dancing, singing and strewing flowers.'

From the gloss in (8c) we see that English allows (roughly) analogous structures
with progressive participles. Amritavalli sometimes calls these structures
''serial verbs'' and sometimes (more accurately) ''participial adjunct
constructions''. English lacks analogues of the perfective participle structures
in (8a). Amritavalli's explanation for the more liberal behavior of progressive
participles is as follows. The fact that progressive participles are selected by
'be' (''I am eating'') and that perfective participles by 'have' (''I have eaten'')
implies that the progressive participle incorporates P/Case, while the perfect
participle does not. Furthermore, ''adjuncts, which by definition are not
externally case-licensed, instantiate only those participles with incorporated
case, which do not require external case; hence English perfect participles are
proscribed as adjuncts...'' (p.56). The author claims that apparent exceptions to
the last statement like ''the washed clothes'' are in reality passive participles.

Section 5 offers some brief remarks on how a structure attributed to
Jayaseelan's article explains properties of constructions of the type ''there is
a lid to this''.

Overall the article's proposals are sketchy, making the reader wonder whether
they would hold up if worked out more fully. As it stands, the proposal has some
dubious aspects. Firstly, the claim (section 4) that adjunct uses of participles
must be passive rather than perfective is counterexemplified by the well-known
perfective non-passive participles with unaccusatives like ''fallen leaves, risen
saviours, departed guests''. This, coupled with the obscure notion of
case-licensing of participles, renders the whole approach questionable.
Secondly, the idea that the existence of 'have'-structures, adpositions and
adjectives in a language presupposes a destabilized case system is refuted by
German and several other languages, as I noted with reference to Jayaseelan's paper.

5. ''Adpositions, particles and the arguments they introduce'' by Peter Svenonius
This chapter discusses several aspects of the grammar of the category P
(=adpositions, verb particles) crosslinguistically.

Section 2 discusses the question as to whether P is a universal category,
favoring the 'universalist' view. Section 3 discusses the thematic roles
associated with prepositions. Relevant here are the notions of Figure and Ground
(e.g. Talmy 2000). The normal generalization is that complements of P are
Grounds and external arguments of P are Figures. Svenonius upholds this in the
face of apparent exceptions, such as ''a pot with a lid''. The claims about
Figure/Ground are extended to particles.

Section 4 discusses non-spatial P. Some non-spatial PPs exploit spatial
metaphors, so the generalizations about the Figure/Ground configuration are
tentatively extended to them. Grammatical and case-marking prepositions (e.g.
passive ''by'', case-marking ''of'') are excluded from the purview of the
generalizations about the thematic roles of P, e.g. since the items in question
are not instances of P and do not themselves theta-mark the DPs after them.
(Svenonius' arguments could be reinforced by the observation of Carrier/Randall
(1992) that resultative predication is impossible over complements of real
prepositions (''They shouted (*at) Egbert into submission'') but is possible with
case-marking 'prepositions' (''The frightening of Egbert into submission'').)

Section 5 argues that, analogously to the standard Minimalist use of little v to
introduce agents in VP, Figures are introduced by a little p head taking a PP
complement. (This reminds one of Bowers' 1993 claim that agents and small clause
subjects are introduced by the same Pr(edication) head.)

Apart from being of interest to linguists inside and outside the prepositional
community, this very informative essay would serve well as an introduction to
adpositions for undergraduate students.

6. ''Argument structure and aspect: The case of two imperfectives in Malayalam''
by M.T. Hany Babu and P. Madhavan
This article addresses two imperfective constructions in Malayalam, using them
as a way of finding out about the nature of the little v tier. [Note before
proceeding that Malayalam orthography can be reproduced only roughly here.]

The imperfective construction in (9a) features ''-ukayaane'' (= ''-uka'', an
infinitive marker, plus ''aane'', which the authors argue is a focus marker). The
imperfective construction in (9b) features ''-unnunte'' (= the imperfective affix
''unnu'' and the existential copula ''unte'').

(9) a. kutti urann-uka aane
Child sleep-infinitive BE
''The child is sleeping.''
b. kutti urann-unn unte
Child sleep-imperfective BE
''The child is sleeping.''

The ''-ukayaane'' construction in (9a) is analyzed as a reduced cleft
construction. The ''aane'' part heads a FocusPhrase and takes a little vP
complement. The VP (being focused) is attracted to spec,FocusP. The subject of
the sentence moves out of spec,vP (or out of VP with unaccusatives) to a
position above FocusP.

In the ''unnunte'' construction in (9b), ''unte'', being an existential verb, has
the effect of asserting the existence of the verbal event. It is claimed to
spell out Harley's (1995) Event head, which is distinct from (other?) variants
of little v and may or may not host an agent in its specifier.

In my opinion, the data and results are potentially interesting, though gaps in
the exposition detract from the argument somewhat. For instance, too little is
said concerning the idea that ''ukayaane'' construction in (9a) is a *reduced*
cleft construction. The reduced part of the construction is said to be either an
agentive little v (which is realizable overtly as ''cey'' ('do') and appears with
transitive and ergative constructions), or an eventive little v (which means
'happen' and occurs with unaccusative verbs). We are left to guess whether the
'reduction' involves gapping of particular light verbs, or ellipsis of an
evacuated vP.

The analyses of the two imperfective constructions exploit two rather different
conceptions of little v, and one wonders how ''unte'' is simultaneously an
existential operator and an agent-introducing light verb. Perhaps the dual role
of ''unte'' could be expressed by separate heads without harming the authors'
overall analysis.

7. ''Argument features, clausal structure and the computation'' by Halldor Armann
Sigurdsson
This article discusses a number of (non-)interactions of Case, the EPP and
clausal syntax, particularly as they relate to the architecture of the
Minimalist Program.

The author reminds us of empirical problems with viewing movement to the subject
position as being driven by the need for DPs to receive (nominative) Case.
Instead, nominative is assigned vP-internally, and the features connecting a
subject DP to the functional categories in the IP domain are Person and a
distinct EPP feature. (Person and Number features are separated because
Icelandic has constructions where verbs agree with dative arguments in person
but with nominative arguments in number.) Section 3 argues that structural cases
are interpretable/meaningful in the sense that structural accusative is
preconditioned by the appearance of nominative. The structural cases are treated
as features distinguishing event participants. The notion of uninterpretable
features is called into question.

Section 4 attempts to explain the fact noted earlier that Person (rather than
Case) is a central feature relating DPs to clausal syntax. Sigurdsson's
explanation is based on his claim that one of the central jobs of grammar is to
relate propositional events to speech events. Person and Tense are crucial in
this regard: Tense relates event time to speech time and Person relates event
participants to speech participants. The author discusses the connections
between propositional event and speech event, relating it to various empirical
phenomena and offering a number of suggestions on clausal architecture.

Although the article addresses many issues, with the consequence that not all of
them can be discussed with the requisite amount of detail, I think it is safe to
say that anyone interested in minimalist syntax will find this article intriguing.

8. ''On theta role assignment by feature checking'' by Tista Bagchi
This brief essay examines the feature checking analysis of thematic roles that
has been adopted in some minimalist studies. The author critically examines
Fanselow's (2001) proposal for a base-generation analysis of scrambling (i.e.
one that attributes scrambling phenomena to different possibilities in merging
arguments within VP rather to than movement), arguing that the analysis does not
successfully show that thematic roles are assigned by checking. The author also
provides a critique of Manzini & Roussou's (2000) proposal to replace standard
assumptions about control and A-movement with a mechanism connecting a DP with
the argument selection features of one (or, in the case of control, two)
predicates. Finally, it is argued that, if theta features exist, they are very
different objects from the features motivating Agree and movement in minimalist
theories.

9. ''Argument prominence and the nature of superiority violations'' by Tanmoy
Bhattacharya and Andrew Simpson
This study discusses multiple wh-questions, particularly as they relate to
argument structure. One relevant phenomenon is superiority, as illustrated in
(10). Such contrasts are commonly attributed to a principle such as Shortest
Move, which blocks derivations if there is an alternative derivation involving a
shorter movement. The connection to argument structure is that arguments of V
will typically not be projected equally close to the landing site of a movement
operation. This, coupled with Shortest Move, is commonly taken to suggest that
superiority facts are a window into argument structure. (The same goes for
languages with multiple wh-fronting inasmuch as the order of the fronted
wh-items might be taken as indicating the relative prominence of arguments as
they are initially projected.)

(10) a. Who saw what?
b. *What did who see

The authors argue against the above assumptions. Firstly, regarding languages
(and constructions within particular languages) not exhibiting superiority
effects, the authors show that the data should not be attributed to flexibility
in argument projection in the relevant languages, or to distinctions between
genuine (Shortest-Move-respecting) wh-movement and a distinct operation such as
focus fronting, which is not triggerred by a [+wh] C-head and therefore need not
respect Shortest Move.

Where apparent superiority effects are found, they are shown to be due to a
number of semantic, pragmatic and prosodic factors, not to Shortest Move. For
instance, data like (11) indicate that a Shortest Move analysis of (10) is a
byproduct of a failure to control for animacy. The authors also show that their
conclusions for English hold for Bangla.

(11) Who did what upset?

I found this essay extremely enlightening. It is a must-read for linguists
interested in wh-movement and for indeed anybody who has appealed to data like
(10) as a motivation for principles like Shortest Move.

10. ''Look across: The paradigmatic axis and Bangla causatives'' by Probal Dasgupta
This essay has as its main empirical focus two facets of causatives in Bangla.
The first is the fact that regular causative morphology is excluded in certain
cases. The author argues against attributing this to blocking by irregular or
suppletive causative verbs. Thus, the non-occurrence of 'cause to go' in Bangla
is not due to blocking by the 'send' verb. This result is said to argue against
''formalist'' theories and for a Substantivist approach taking the paradigmatic
axis seriously.

A second aspect of Bangla causatives discussed by the author is the 'sarcastic
causative' construction, which can be illustrated very roughly with an English
gloss like (12). The second sentence features a causativized verb whose
interpretation is sarcastic inasmuch as the utterance dismisses the claim that
the car can drive on these roads.

(12) Your car goes on these roads, does it? I'll make it go!

Sarcastic causatives use regular causative morphology even with verbs like ''go''
which otherwise reject it. Relevant here is a sort of cross-sentence dependency.
The causative somehow depends on the verb's having been previously uttered.
Dasgupta claims that similar factors are at work in English echo questions, and
in the word ''bigness'', which is normally unacceptable, but can be used in
certain contexts where ''big'' is pre-mentioned. The author proposes a device
called Look Across which tries to capture the sensitivity of such violations of
the normal use of an expression to the presence of this expression in the prior
discourse.

Despite making some interesting points, this article does not belong in a
collection on argument structure. Moreover, I must say that much of the
discussion in the first half of the article should not have passed through a
review process in its present form. I found the anti-formalism/pro-Substantivism
discussion regarding the blocking issue (and a related scene-setting discussion
in the introduction) well-nigh incomprehensible. It cites no literature on
blocking and no ''formalist'' literature from after 1981. Moreover, the challenge
to the relevance of blocking to Bangla causatives does not invalidate the notion
in general. It is known that morphological processes can vary in their degree of
productivity and may impose semantic, phonological or morphological constraints
on their bases. Unless it is shown that none of these factors explains the
unacceptability of 'cause to go' in Bangla, it is not clear that the data have
any relevance at all to ''blocking'' (or underspecification or Elsewhere effects,
which seems to be what the author means by ''blocking'').

On the upside, Dasgupta's data on sarcastic causatives raise challenging
questions. To aid further research on this, I will add a few relevant examples
to those noted by Dasgupta:

-The prior occurrence of a cranberry morpheme sometimes allows an otherwise
illicit de-cranberrification of this morpheme. Thus, ''ept'' normally only appears
in ''inept'', but is jocularly acceptable as a free morpheme with prior mention of
''inept'': ''John is inept, and Mary is not exactly ept either''.

-Bad cases of ''-ee'' affixation (''*killee'') sound better if the base is
pre-mentioned in an '-er' nominal (''The soldier decided that it's better to be a
killer than a kill-ee.'')

-''*un-nice'' is normally deviant, but a prior ''nice'' can rescue it: ''She says
that her theory has some 'nice' consequences, but I find them decidedly un-nice.''

Some of these data involve conscious manipulation of language rather than normal
language use. Even so, one still wonders why, and under what precise conditions,
prior use enables morphological constraints to be violated.

EVALUATION
In sum, the book contains some excellent work, but also some work with clear
flaws which should have been corrected in the review process. A way to avoid
this problem in future volumes might be to ensure that each essay is reviewed in
such a way that the decision to publish articles is not treated as the default
option, and is contingent on (a) the elimination of obvious oversights and (b)
empirical and/or theoretical usefulness.

I hope others will join me in protesting against the publisher's failure to
issue books in this series in paperback format, resulting in prohibitive prices.
(The book's website charges 110 Euros.) The much higher cost of hardcover format
is justified if a book is likely to become dog-eared through very frequent use,
but for most users this problem will affect only textbooks and reference works,
but not relatively specialized works such as the book under review.
Consequently, many private buyers (and institutional buyers with limited budgets
or a conscience about the use of taxpayer money) will not buy this book unless
the publisher finds a way to halve the price.

REFERENCES
Alsina, A., Bresnan, J. & Sells, P., (eds) 1997. _Complex Predicates_. Stanford:
CSLI.

Bowers, J. 1983. The Syntax of Predication. _Linguistic Inquiry_ 24: 591-656.

Butt, M. & W. Geuder (Eds.) 1998. _The projection of arguments_. Stanford: CSLI.

Carrier, J., & Randall, J. 1992. The argument structure and syntactic structure
of resultatives. _Linguistic Inquiry_, 23, 173-234.

den Dikken, M., 1997. The syntax of possession and the verb 'have'. _Lingua_
101: 129-150.

Emonds, J. 1985. _A unified theory of syntactic categories_. Dordrecht: Foris.

Fanselow, G. 2001. Features, theta-roles, and free constituent order.
_Linguistic Inquiry_ 32: 405-437.

Freeze, R. 1992. Existentials and other locatives. _Language_ 68, 3:553-596.

Goldberg, A. 1995. _Constructions_. Chicago: University of Chicago Press.

Hale, K., & Keyser, S., 1997. On the Complex Nature of Simple Predicators. In
Alsina et al. 1997. 29-66.

Harley, H. 1995. _Subjects, events and licensing_. MIT Dissertation.

Jackendoff, R. 1990. _Semantic Structures_. Cambridge (MA): MIT Press.

Kayne, R. 1993. Toward a modular theory of auxiliary selection. _Studia
Linguistica_ 47, 3-31.

Levin, B., & Rappaport Hovav, M. (1995). _Unaccusativity_. Cambridge, MA: MIT Press.

Manzini, M. R. & Roussou, A. 2000. A minimalist theory of A-movement and
control. _Lingua_ 110: 409-447.

McIntyre, A., 2004 Event Paths, Conflation, Argument Structure and VP Shells.
_Linguistics_ 42(3):523-571.

Pinker, S. (1989). _Learnability and cognition_. Cambridge, MA: MIT Press.

Rappaport Hovav, Malka; and Levin, Beth 1998. Building verb meanings. In Butt
and Geuder (eds.), 97-134.

Talmy, L. 2000. _Toward a cognitive semantics_, Vol. 1. Cambridge, MA: MIT Press.

Wunderlich, D, 1997. Argument extension by lexical adjunction. _Journal of
Semantics_ 14, 95-142.

ABOUT THE REVIEWER
Andrew McIntyre (www3.unine.ch/andrew.mcintyre) is an assistant professor for
(English) linguistics at the University of Neuchatel, Switzerland. He mainly
works on issues of the syntax-semantics interface in the VP domain.