Review of The Composition of Meaning
Date: Mon, 09 May 2005 23:15:55 -0400
From: Catherine R. Fortin <firstname.lastname@example.org>
Subject: The Composition of Meaning: From lexeme to discourse
EDITORS: ter Meulen, Alice G.; Abraham, Werner
TITLE: The Composition of Meaning
SUBTITLE: From lexeme to discourse
SERIES: Current Issues in Linguistic Theory 255
PUBLISHER: John Benjamins
Catherine R. Fortin, Department of Linguistics, University of Michigan
This volume contains eight new papers which explore issues relating
to interfaces with syntax; the first four deal with phenomena on the
syntax-semantics interface, while the second four are concerned with
syntax-pragmatics interface. Following an introduction by editors ter
Meulen and Abraham in Chapter 1, Part I of the book contains four
papers dealing with issues on the syntax-semantics interface,
including the composition of meaning of morphologically complex
compounds (Olsen); aspect and its syntactic properties, specifically
with respect to the diachronic morphophonological changes in English
infinitives (van Gelderen); the issue of unintelligibility within Optimality
Theory (OT) (de Hoop); and unaccusative verbs (Abraham). Part II of
the book, which contains four papers dealing with issues on the
syntax-pragmatics interface, is motivated by the question of how best
to integrate syntax and discourse-sensitive phenomena, with the goal
of eliminating information-structural features (e.g. [+foc]) in favor of
relying on the interface between the grammar itself and more general
cognitive processes. Topics covered include English either, both and
neither (Hendriks); the interaction of case morphology and word order
in Middle Bavarian (Weiß); the way information structure is encoded in
a sentence (von Heusinger); and negative concord in Afrikaans
In 'Coordination in morphology and syntax: the case of copulative
compounds', Susan Olsen addresses the issue of the composition of
meaning of a certain type of morphologically complex compound, the
so-called copulative compound (e.g. bartender-psychologist).
Copulative compounds are semantically unique in that they represent
a coordinative (not subordinate) relation between two (or more)
constituents, which are predicated equally of the same referent.
Olsen argues that the semantic composition of a morphological
copulative compound is not equivalent to its syntactic analogue,
the 'coordinative apposition', as shown in (1) (her (4)), and hence is
governed by different principles.
a. Henry Kissinger, diplomat and lightning rod, returns to the corridors
b.* The diplomat-lightning rod returns to the corridors of power.
Olsen proposes a cognitive principle, the Principle of Ontological
Coherence (POC), which states that 'a complex concept as the
denotation of a morphological object picks out a coherent individual
from one of the domains of individuals' (p. 19). The POC holds only in
the morphological component - not in the syntax - accounting for the
grammaticality contrast shown in (1). Olsen proposes a 'compound
template' according to which the meaning of determinative compounds
(those characterized by a subordinate relation, e.g. computer monitor)
is computed. The semantics of this template consists of a context-
independent level containing an open parameter that is instantiated by
a specific relation at a context-dependent level. This context-specific
relation is suggested to be the relevant difference in interpretation
between the different types of determinative compounds, and that
copulative compounds are simply a type of determinative compound;
the copulative interpretation is the 'most neutral' and is obtained as a
default, when the context does not provide another meaning.
In 'Aspect, infinitival complements and evidentials', Elly van Gelderen
argues against Giorgi & Pianesi's (G&P) (1997) analysis of eventive
infinitives lacking the ending -en (as in English, and unlike German
and Dutch) as inherently perfective. G&P argue that these verbs carry
a perfective feature, because perception verb complements (PVCs) in
English (but not German and Dutch) are perfective. van Gelderen
uses diachronic changes in English eventive infinitives to support her
claims that they do not carry a perfective feature, and that the
incipience of the perfective interpretation of bare eventive infinitives
did not coincide with the loss of -en (contra G&P). van Gelderen
hypothesizes that bare eventive infinitives in English became
perfective as the result of two diachronic changes: the reanalysis of -
ing as a checker of the imperfective feature in Asp° (following the loss
of ge- as a marker of perfectivity in Asp°), and a parametric change in
unmarked aspect (from unmarked imperfectivity to unmarked
van Gelderen argues that not all PVs uniquely require perfective
complements. She claims that there are three kinds of perception
verbs (PVs) in English, and hence, three kinds of PVCs in English.
More specifically, there are three kinds of 'see': a (rare) activity see,
as in (2a) (her (79)); a stative see, as in (2b) (her (3)); and a modal
see, as in (2c) (her (1)), which van Gelderen argues has become
grammaticalized as an auxiliary/evidential.
a. Poirot was seeing the face of a girl with red hair.
b. I see him crossing the street.
c. I saw/*see him cross the street.
The difference between these three types of see is reducible to their
complements: activity 'see' selects CP; stative 'see' selects AspP; and
auxiliary 'see' selects vP. If perfectivity in English is unmarked, and
the PVC selected by the evidential see does not contain Asp° (the
locus of (im)perfectivity), the unavailability of an imperfective reading
in (2c) is accounted for.
Helen de Hoop's 'The problem of unintelligibility' is the only paper
included in this volume couched within a non-Chomskyan theoretic
framework. de Hoop uses a bi-directional Optimality Theoretic
framework (which assumes that both speaker and hearer take into
account the other's perspective) to explore unintelligibility, which
obtains when a syntactically well-formed expression fails to give rise to
a felicitous interpretation, such as (3) (her (18)).
(3) Most female professors are men with beards or glasses.
Within OT, unintelligibility arises when there is not an optimal
interpretation within a candidate set of interpretation outputs for a
given syntactic expression. (The counterpart of intelligibility is
ineffability, which obtains when a given semantic input does not yield a
well-formed syntactic expression as its output.) de Hoop argues that
the problem of unintelligibility with OT can be circumvented by simply
extending the candidate set of outputs to include contradictory
interpretations, which may be calculated to be the optimal candidate.
She posits a set of five (pragmatically-oriented) ranked constraints
which govern the matching between a syntactic input and its semantic
output, and suggests that ranking a constraint 'Be Informative' first,
over other constraints such as 'Avoid Contradiction', will yield the
correct result in cases such as (3) (i.e., the number of female
professors who are men with beards or glasses is greater than the
number of female professors who are not).
Werner Abraham's 'VP-internal subjects as "unaccusatives":
Burzio's "Object Account" vs. the "Perfectivity Account"', is the second
of two papers in this volume to deal with perfectivity. Abraham argues
that both unaccusativity and Burzio's Generalization (BG), which links
a verb's ability to assign accusative case to its ability to license an
external argument, are epiphenomena with roots in Aktionsart
perfectivity (the 'Perfectivity Account'). Abraham refutes one of BG's
empirical tests - the use of existential clauses to test whether a
predicate's single argument is internal - as inadequate. Compare (4a)
(his (5a)), containing an unaccusative verb, to (4b) (his (5c)),
containing an unergative; if the single argument is definite, an
existential clause is similarly ungrammatical (5) (his 5b).
a. There appeared a/some/many/few/three dog(s) in the garden.
b. * There ran a/some/many/few/three dog(s) in the garden.
(5) * There appeared the/all/most/both/every dog(s) in the garden.
Abraham adopts Diesing's (1992) Mapping Hypothesis, which states
that while definite object NPs must move to positions outside of the VP
by LF to take scope, indefinite object NPs remain within the VP.
Specifically, Abraham proposes that definite NPs must move to [Spec,
AgroP], a position by hypothesis unavailable in the clause structure of
existentials. Consequently, he argues that existential clauses are not
effective as a diagnostic of unaccusativity, and that semantic
properties such as telicity, change of state, and the possibility of an
agentive reading must be used as diagnostics instead. Abraham
proposes that the relevant difference between (4a) and (4b) is
perfectivity: namely, appear (and other so-called unaccusatives) is a
perfective verb, while run is not. Ultimately, the question of how to
represent the syntactic difference between verbs such as appear and
verbs such as run remains somewhat open. Abraham assumes that
the syntactic representation of perfectives contains a small clause,
whence the single argument's logical subject status; for details, the
reader is referred to earlier work. He also assumes that all structural
cases except for nominative are structurally/lexically inherent.
In 'Either, both and neither in coordinate structures', Petra Hendriks
advances the argument that these three lexical items are focus
particles quantifying over a set of relevant alternatives, contra the
traditional analysis of these lexical items as conjunctions. She
considers each lexical item individually, in comparison with clear focus
particles such as 'only', with respect to their distribution, their
interaction with sentential intonation, and their contribution to the
interpretation of the sentence. The second and third criteria - that
focus particles must c-command the element in the first conjunct
receiving contrastive stress, and that sentences with focus particles
entail the sentence without the focus particle - most clearly
demonstrate that these three lexical items are subject to the same
constraints and receive the same interpretation as other focus
particles. Hendriks concludes that 'either' and 'neither' (akin to 'only')
are restrictive focus particles, quantifying over alternatives that are
excluded, while 'both' (similar to 'also') is an additive focus particle,
quantifying over alternatives that may be included. (She further notes
that 'both' is not unambiguously a focus particle; in some contexts -
i.e., where 'both' does not c-command the contrastively focused
element - 'both' is a floated quantifier.)
In 'Information structure meets Minimalist syntax: On argument order
and case morphology in Bavarian', Helmut Weiß advances an analysis
of two types of short scrambling in double object constructions in
Middle Bavarian: the obligatory scrambling of a definite direct object
over an indefinite direct object, and the optional object inversion for
reasons related to focus (wherein the unfocused element precedes
the focused element). Weiß argues that, because the Bavarian case
system is morphologically impoverished, the first type of (obligatory)
scrambling is feature-driven, occurring within the narrow syntax;
however, the second type of (optional) scrambling is a 'stylistic'
operation occurring in a post-syntactic component.
To capture the typological distinction between 'free' and rigid word
order languages (e.g. Bavarian and English), Weiß proposes the
Principle of Strong Morphology (PSM), which supposes that overt
morphology and feature checking are related; namely, strong
morphology (e.g. case in Bavarian, but not in English) delays feature
checking until LF. He further assumes a distinction between core
features (e.g. case) and peripheral features (e.g. focus); although
both types of features drive movement, only the core features must be
checked within the narrow syntax; peripheral features may, possibly,
instead be checked within the phonological component (cf. Chomsky
2000). Specifically, Weiß proposes that definite NPs carry a [D]
feature which indefinite NPs lack; since in Bavarian, this feature is not
morphologically encoded (and, according to the PSM, must be
checked prior to LF), the definite DO must raise to AgrP overtly, over
the indefinite IO, which remains in situ.
In 'Focus particles, sentence meaning and discourse structure', Klaus
von Heusinger proposes an alternative to semantic theories of
information structure (e.g. Alternative Semantics and Structured
Meanings) which hold that sentences are semantically composed of
two disjunctive units (e.g. the 'background', or given/presuppositional
information, and 'focus', new information). von Heusinger's
Foreground-Background Semantics holds that the information
structure of sentences is composed of two overlapping units: the
foreground (the new information, which for von Heusinger is the entire
sentence) and the background (the foreground minus focused
expressions). He seeks to demonstrate that focus-sensitive
expressions, such as focus particles like only and adverbs of
quantification like usually, are operators which take two arguments,
the foreground (not the focus alone) and the background.
von Heusinger provides support for his proposal by showing that
approaches which rely on a distinction between focus and background
(instead of foreground and background, as his does) are unable to
account for complex NPs, as in the following example (his 6a).
(6) Sam only talked to the SWISS artist.
Here, Swiss receives contrastive focus. Theories of focus semantics
make the incorrect prediction here; namely, they hold that the two
arguments of only would be the focus (Swiss) and the background
(Sam talked to --- artist) This would presuppose a uniqueness
requirement on the set of alternatives which is not met if the set
contains, for example, two German artists. Using a framework of
Segmented Discourse Representation Theory, von Heusinger shows
that his theory, supplemented with the additional assumption the
uniqueness requirement of the definite article is presuppositional, not
semantic, is able to account for these examples.
In 'On the interpretation of multiple negation in spoken and written
Afrikaans', Laszlo Molnárfi seeks to account for a phenomenon -
negative concord (NC) - that is manifested differently in written and
spoken varieties of Afrikaans. In written Afrikaans, NC forms a
negation bracket, which is interpreted semantically as a single
sentential negation; the first negative particle opens the scope of
negation, while the second marks the right periphery of the sentence,
as in (7) (his (23c)). In spoken Afrikaans, however, additional copies
of the negative particle can appear within the scope of negation, as in
(8) (his (2)). In spoken Afrikaans, additional copies of the negative
particle do not effect the interpretation of negation, while in written
Afrikaans, additional copies are interpreted as the logical cancellation
Ek het niemand gesien nie.
I have nobody seen not
'I have not seen anybody'
Ek het niemand nie gesien nie.
I have nobody not seen not
'I have not seen anybody' (Spoken Afrikaans)
'I have seen everybody' (Written Afrikaans)
Molnárfi argues that NC should not be captured in terms of a formal
operator-licensing/feature-checking mechanism, and proposes instead
that NC follows top-down percolation of the NEG feature, whereby
silent copies of NEG percolate downward to all terminal nodes. The
spoken and written varieties differ only according to the constraints
that govern Spell Out of intermediate copies of negation: the written
variety requires a rule stipulating that only the lowest copy of negation
is Spelled Out, while the spoken variety allows for more liberal Spell
Out of additional copies within the phonological component,
to 'facilitate parsing of the negation bracket' (p. 201). To account for
the different interpretations available for (8), Molnárfi argues that
spoken and written languages employ different strategies for
information processing, arising from 'different communicative needs'
with respect to sentence planning and processing, noting that in
texts 're-reading is always possible' (p. 214) if the negation bracket is
parsed incorrectly, a strategy which is unavailable in spoken language.
The papers collected in this volume present a more coherent selection
than might have been expected of a wide-ranging volume of this type,
which considers issues on both the syntax-semantics and syntax-
pragmatics interfaces. There are two papers (Abraham's and van
Gelderen's) which discuss perfectivity; two which discuss focus
particles (those of Hendriks and von Heusinger); and two which touch
on definiteness effects (Abraham's and Weiß's). Each pair of papers
is complementary, considering the phenomenon in question from
different angles. Overall, the papers grapple with deep issues (how
semantic and pragmatic/discourse-related phenomena are or are not
rooted in the syntax) in a principled and meaningful way by closely
considering specific Germanic phenomena. For example, Weiß' study
on object scrambling in Middle Bavarian supports the Minimalist thesis
that pragmatic considerations do not constrain or govern syntactic
operations, and he effectively argues that pragmatically well-formed
derivations are the 'side effects' of syntactic operations.
Not all of the papers contained herein commit to a specific theory of
syntax, and it is not immediately clear how some of the proposals
advanced would be compatible with recent developments within
generative syntax. For example, it's not clear if Molnárfi's proposed
downward percolation of the negation feature could be captured in
e.g. a phase-based theory, given that this percolation can cross
clausal boundaries and has phonetic consequences. (This is not
necessarily a criticism, of course; but given that these papers deal
with interfaces with syntax, it would be interesting to see how the
proposals are implementable within syntactic theory.) Also, given the
cross-disciplinary nature of this volume, some readers may wish that
papers representative of a range of syntactic theories had been
included; only one was explicitly framed within a non-Chomskyan
theory (de Hoop's, in bi-directional OT).
The papers in this volume do contain some minor errors, which range
from misspellings (e.g. 'suboordonated' for 'subordinated' on page
221; 'prefigation' for 'prefixation' on p. 101) to incorrectly numbered
footnotes to unclear citations. For example, on page 2, Abraham,
Epstein, Thráinsson & Zwart , eds. (1996) is cited as a 1995
publication; in the list of references this same work is titled 'Studies in
Minimalism' where 'Minimal Ideas: Syntactic Studies in the Minimalist
Framework' was intended. There are a few instances where the lists
of references are incomplete; for example, Chapter 5 cites 'Burzio
1993' several times, yet this work is not found in the list of references
for this chapter. Nonetheless, these typographic errors should not
detract from the high quality of the papers contained in this volume, a
welcome addition to the body of research on the syntax-semantics
and syntax-pragmatics interfaces.
Abraham, Werner, Samuel Epstein, Hoskuldur Thráinsson & Jan-
Wouter Zwart, eds. (1996) Minimal Ideas: Syntactic Studies in the
Minimalist Framework. Amsterdam: John Benjamins.
Chomsky, Noam. (2000) Minimalist inquiries: the framework. In
Martin, Michaels & Uriagereka (2000), 89-155.
Diesing, Molly. (1992) Indefinites. Cambridge, MA: MIT Press.
Giorgi, Alessandra and Fabio Pianesi. (1997) Tense and Aspect.
Oxford: Oxford University Press.
Martin, Roger, David Michaels and Juan Uriagereka, eds. (2000)
Step by step: Minimalist essays in honor of Howard Lasnik.
Cambridge, MA: MIT Press.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Catherine Fortin is a Ph.D. candidate in the Department of Linguistics
at the University of Michigan. Her research interests primarily concern
the interface of syntax and discourse/pragmatics, including the syntax
and interpretation of nonsententials. She is also interested in the
syntax of argument structure of Austronesian languages, most
especially Indonesian and Minangkabau.