LINGUIST List 12.2911

Tue Nov 20 2001

Review: Allan, Natural Language Semantics (2nd rev)

Editor for this issue: Terence Langendoen <terrylinguistlist.org>


What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Simin Karimi at siminlinguistlist.org or Terry Langendoen at terrylinguistlist.org. <p> Subscribe to Blackwell's LL+ at http://www.linguistlistplus.com/ and donate 20% of your subscription to LINGUIST! You get 30% off on Blackwells books, and free shipping and postage!

Directory

  1. Eleni Koutso, Allan, Natural Language Semantics

Message 1: Allan, Natural Language Semantics

Date: Tue, 20 Nov 2001 09:38:46 -0800 (PST)
From: Eleni Koutso <e_koutsoyahoo.com>
Subject: Allan, Natural Language Semantics

Allan, Keith (2001) Natural Language Semantics. Blackwell
Publishers, xix+529pp, hardback ISBN 0-631-19296-4, $64.95.

Reviewed by Eleni Koutsomitopoulou, ABD Computational
Linguistics Georgetown University

Another review of this book can be found at
http://linguistlist.org/issues/12/12-1709.html

OVERVIEW
Allan's "Natural Language Semantics" is an introductory book on
semantics and as such its "intended readership is undergraduate
and graduate students of linguistics and relevant areas of
psychology, philosophy, anthropology, sociology, communications,
language studies, and education." The book indeed discusses
"fundamental concepts for linguistic semantics." What makes it
special is the author's encompassing consideration that meaning
is "cognitively and functionally motivated." even though the
tools for its description are formal. The author's main goal is
to offer his intended readership the basic tools and skills
needed in order to exercise proper and linguistically acceptable
semantic analysis. To this aim, Allan takes pains with his
presentation of varied linguistic methods of semantic analysis.
The traditional formalist approach has been elucidated in as
great depth and clarity as the cognitive and functionalist
approach.

Each chapter includes a standard "Where we are heading"
introductory part, as well as a "Summary" section and "Notes on
further reading". Throughout a chapter the reader will find
exercises, the answers of which readers may find online at:
http://www.arts.monash.edu.au/ling/nls/getans.shtml This user-
friendly layout as well as the presence of a section explaining
the "Symbols and Conventions" used in the book and the quite
comprehensive Index makes this book a very accessible textbook
for both undergraduate and general graduate work on semantics.

Each of the 13 chapters may be considered fairly autonomous.
Below I will present the contents of each chapter along with my
critical comments on certain theoretical practices and views
adopted and presented in each chapter. In closing, I will proceed
into a short general critical evaluation of the book.

In Chapter 1 Allan describes basic semantic concepts such as the
metalanguage of semantic theory, inference, the nature of speech
events, context, the principles of cooperation and relevance, and
compositionality. These concepts are presented informally and
non-technically by way of introduction and they are interwoven
around the epicenter of such fundamental and elementary notions
as that of a metalanguage for semantic analysis and context. The
author presents some criteria for the selection of a metalanguage
appropriate for semantic analysis. It is pointed that a
metalanguage is an expressive and comprehensive formal language
that ideally possesses the same descriptive and explanatory power
as natural language. Some further assumptions about semantic
theory and the interdependence between semantics and the other
levels of linguistic analysis are also informally discussed in
this chapter. Context is defined as "a model of the world and
time spoken or written of", or the discourse model or discourse
world, but also as a "co-text", i.e. the discourse preceding and
following a given linguistic point in discourse. Context is also
meant as the pragmatic "situation of utterance" and "situation of
interpretation" corresponding to the factors of place and time of
the speaker and the "hearer" respectively during the utterance.
Other general assumptions about semantics and meaning in human
languages are also discussed in this chapter. Because of its
informal style this first chapter offers an excellent opportunity
for students in undergraduate classes to exercise their written
and oral argumentation skills. The summary section recapitulates
the main concepts and new terms discussed in the chapter. The
summary along with the extensive references to related literature
in the "notes on further reading" make the first chapter a
comprehensive introduction to semantics for both beginners and
more advanced linguistics students.

Chapter 2 starts with a crucial trifold differentiation between
sentence meaning, utterance meaning and speaker meaning. Then the
focus shifts to an informal description of extension and
intension, reference and anaphora. The latter are brought up by
way of explanation of the distinction between "dictionary
meanings" and meanings of utterances in the discourse model or
discourse world. The reference to "dictionary meaning" is merely
lexicographic and the author does not offer a much-needed
linguistic criticism of the lexicographic practices followed
regarding homonymous and polysemous terms (but see next chapter
for further discussion of homonyms). Thus he uses a rather
problematic example from Webster's New Collegiate Dictionary 1977
to demonstrate homonymy and polysemy. The term "canine" is
separated in two "homonyms" canine1 and canine2 that basically
differ in part of speech identity rather than in meaning. The
noun "canine" bears two quite distinct senses that in a different
dictionary could have been cataloged as homonyms themselves. No
comments on the fuzziness of the example (and the related
lexicographic practices) have been made by the author. A more
obvious example of homonymy and polysemy would have better served
the purposes of this section. Leaving the details of the lexicon
for the next chapter, chapter 2 builds on the notion of "world
and the time spoken of" that was presented in the previous
chapter. First, speaker's meaning is defined as the speaker's
intended meaning of an utterance. Utterance meaning is defined as
the meaning of an utterance as it is perceived by a hearer in the
context of its utterance. Sentence meaning is defined as the
informative content of a sentence expressed in terms of an
abstract system. The next step is what is noted as an informal
description of the terms "denotation", "extension", "intension"
and "reference". Also, the term "proposition" is used in this
chapter for the first time. The final sections of the second
chapter get into the specifics of generic NPs and anaphora. This
chapter is packed with technical terms and sophisticated
concepts, and the author's attempt to define each and every one
of them so early in the book and in the same chapter leaves the
reader with the impression that they were intended to be "brushed
away" rather than studied in depth.

Chapter 3 focuses on the lexicon and its basic structure. Words
within the lexicon are designated "listemes". In the same
chapter, the lexicon is seen as part of a bigger construction
that of the encyclopedia. Basic differences between the lexicon
and the encyclopedia are outlined, the most highlighted being the
distinction between meaning and denotation. The perspective
adopted in this chapter is that lexicography is the construction
of "models of the mental lexicon". In this line of thought, the
author points out Lyons's 1977 suggestion that lexical entries
have been assigned indices for easy access during the formal,
morphosyntactic and semantic specification of an entry. Allan
then goes as far as to discuss Jackendoff's 1995 and 1997 lexical
licensing theory and explain its significance in justifying
traditional lexicographic practices that distinguish for instance
the word "canine" in two homonyms given only their
morphosyntactic specification, or part-of-speech information:
"member of the dog family (N)" vs. "pertaining to the dog family
(Adj)". The gist of this presentation is the important lesson
that each lexical entry is a complex network of relations between
formal (i.e. pertaining to form), morphosyntactic and semantic
specifications, i.e. we derive semantic information for an entry
from both its form and its part-of-speech information, and
conversely we discover the appropriate form and part-of-speech of
a lexical entry given the intended meaning. Another interesting
point in the 3rd chapter was made about proper names. These are
taken to be partly or wholly defined in the lexicon where form
features like spellings and pronunciation as well as gender are
thoroughly specified and from which speakers of a language draw
upon when baptizing new members of a community.

Chapter 4 touches upon morphology and its connection to meaning
and semantics. The issue of "polyword listemes", i.e. compounds,
idioms and phrasal verbs, and their seeming non-compositionality
is being explored. Listemes are described morphologically and in
relation to each other. English inflection (affixes) and
nonderived lexemes are listemes in the lexicon. Languages with
richer morphology may follow different rules. The general rule is
that if the meaning of an inflectional/derived listeme is
transparent (compositional) then it should exist in the lexicon
as two separate listemes (affix and root). Derived lexemes
(including lexemes that are products of zero-derivation) present
different cases in that they usually are independent from their
"derivational origins", which makes them good lexical entries.
Phrasal verbs are compounds (often discontinuous) and as such are
listemes. Idioms, even "polyword" ones, are lexemes of a certain
morphosyntactic class. For instance, nominal idioms are likened
to NPs. Sound symbolism is the phenomenon in natural languages
during which, through certain lexemes, a conventional (thus
language- specific), yet arbitrary, connection between the form
and the meaning is adopted. A typical example of sound symbolism
are onomatopoetic words that mimic sounds of the denotatum. Allan
presents three different types of sound symbolism: phonesthesia,
synesthesia and ideophones, and he suggests --for a good reason--
that synesthetic words do not belong in the lexicon. This chapter
is rich in examples from English and other languages when it
describes sound symbolism, but it lacks a similar crosslinguistic
depth when it accounted for compounds and their presence in the
lexicon.

Chapter 5 delves into connotation and jargon. The author bridges
the two by regarding connotation as "what is being naturally
believed about the meaning of a word" and jargon as the
"realization of the special meanings of the language of a group".
Connotation is seen as a source of inspiration for the creation
of new vocabulary. Connotation is what differentiates synonyms in
certain context in such a way that no two words can substitute
each other entirely and in every context in natural languages
(this is what Allan has earlier baptized as the "blocking
principle"). Subtle connotations of names and terms of address
follow different styles of naming and addressing, and different
languages exploit this source in various ways. (see Joos 1961 for
style in English). At this point, Allan explains the connotations
of discriminatory language expressed with "-IST dysphemisms", and
other newly enlisted listemes. He then describes "taboo" words
and the generation of substitute euphemistic words or
expressions. He also explains "pejorization" or negative
connotations and their societal persecutions. When he discusses
gender-specific language ("words denoting women") and how it is
pejorized, he calls forth the issue of polysemous words with
taboo senses and how natural language users resort to semantic
narrowing to eliminate the taboo senses. On the same topic,
dysphemisms are used to show the speaker's disapproval and often
separate the Speaker's (and his/her audience's) opinion from that
of the opponent. Insults are analyzed in detail and Allan offers
five types of "dysphemistic terms of insult". The cognitive-
functional attitude of the author is becoming apparent when he
reminds us that euphemisms and dysphemisms promote semantic
change. Other ways for language to evolve semantically are:
remodelling, phonetic similarity, acronyms, abbreviations, verbal
play, circumlocution, hyperbole, understatement, metonymy,
substitution, synecdoche, borrowing. A big absent here is
metaphor, however, the author categorizes metaphor ("metaphorical
extension") along with "lexical confusion" and "semantic
transfer" as semantic processes rather than linguistic realities.
These processes are responsible for bringing about new listemes
or new meanings to existing listemes. Next, Allan focuses on
jargon. Jargon is often associated with specialized vocabulary,
special syntactic forms, and special typographical or other
norms. Its main function is to serve as a technical language or
the language of a specialist. The second function is to promote
group consciousness between people who share similar interests
and use the jargon. Like the groups they represent, jargons can
be highly prestigious. Allan also explains how jargons have taken
a bad name: they often only serve the image of the group they
represent with no real or substantial need for a specialized
language. Even "social chit-chat" is a type of jargon. Jargons in
a language may become quite elaborated to the point of displaying
discrete styles.

Intersentential semantics is investigated in Chapter 6. After a
brief sketch of propositional calculus, the author discusses
conventional and conversational implicatures, presuppositions,
and illocutions. The basic tools for truth-conditional semantic
analysis are presented here in their most elementary form.
However, the author is -- justifiably -- guided by the belief
that natural language is less than a ideal subject for truth-
condition analysis: implicature, especially as it comes up in
discourse (conversational implicature) is very important when we
account for natural language meaning. Related terms like
implication, entailment, and conventional implicature are being
examined. Also, semantic presupposition and pragmatic
presupposition are examined in this chapter, especially in regard
to conversational implicature. Interesting points in this chapter
are also the use of truth- conditional or formal semantics in
general to elucidate differences between entailment and synonymy,
as well as conventional vs. conversational implicature. For
instance, entailment is regarded as semantic implication, and
synonymy as semantic equivalence, whereas conventional
implicature is a non- truth conditional semantic implication that
is a lexical rather than propositional assignment and part of the
linguistic system rather than the speaker or listener and their
language expectations and regardless of whether we flout or
observe the cooperative principle or the maxims.

An example of conventional implicature is the two aspects of the
maxim of Quantity: there is a maximum quantity with a negative
implicature, and a minimum interpreted quantity, and the clues
for each one are either lexical or grammatical. The issue of
presuppositions and their defeasibility is analyzed next. In sum,
Allan defends the following position: The fact that
presupposition can be made void renders them pragmatic or at
least at the borders of the semantic with the pragmatic realm.
Although both a proposition A and its negation should entail B in
order for B to be a presupposition, the linguistic reality of
"presupposition negation" undermines the semantic nature of
presupposition itself: a presupposition negation negates the
presupposition proposition as well as the presupposition behind
it. When the presupposition is negated but not the presupposition
proposition that sustains it, the proposition becomes
underspecified or indeterminable (neither true or false). In
illocutions, presuppositions are those "preconditions" (or
"preparatory conditions" in Allan's terms) that help speaker and
listener communicate felicitously. Since preconditions, and
therefore presupposition, are subject to the maxim of quality,
both are part of conversational implicatures. Accepting
presuppositions as conversational rather than conventional
implicatures would explain the defeasibility of presupposition.

Traditional formal semantics starts with Chapter 7, where the
author presents predicate logic, set theory, functions, and the
lambda operator. A simple proposition and its constituents are
analyzed semantically using formal tools. Ways for the semantic
analysis of clauses and sentences are also reviewed. The notion
of "scope" of quantifiers and operators is first presented here.
The set theory is linked to certain lexical relations as "hyponym
of" ("subset of") and "hypernym of" or "superordinate of"
("superset of"). The interpretation and assignment functions are
being explained and distinguished: assignment is mechanical and
enables the proposition with its predicates and arguments to be
evaluated for truth, whereas interpretation defines the
conditions for the satisfaction of a predicate for a specific
domain. An interesting methodological stance of the author
becomes clear in this chapter: the author chooses to use the
lambda operator in this book only as a "convenient extension to
predicate calculus", instead of delving into the advanced issues
of its use in model theoretic semantics by type categorial
languages (languages that only use two types: t for truth value
and e for entity to define every other type.)

Chapter 8 makes a leap from formal to cognitive and lexical
semantics and the concepts of scripts, frames and semantic
fields. The notion of compositionality is revisited in light of
"componential analysis" and the search for "semantic primitives".
A more formal definition of "listeme meaning" is given in this
chapter. The most prevailing cognitive trademark in this chapter
is Allan's view of semantic properties (the equivalent of
"features" in formalistic approaches to semantic analysis) and
semantic relation among entities: Allan points out that both
properties of entities and relations among entities are greatly
constrained by the perceptual categories and relations that the
speakers have internalized. Thus, intensions and their extensions
in space and time are critical for any account of senses and
sense relations. Following is a basic differentiation between
scritps, frames and fields as explicated by Allan: Scripts apply
to text in order to fill in the gaps of the textual description
of dynamic event structures. The speaker assumes a lot of details
that s/he knows that the Hearer can deduce from their own
internalized knowledge of scripts. Frames apply to listemes in
order to determine essential characteristics (attributes and
functions) of their denotata. Frames encompass all the relevant
"arguments" of a listeme (in fact, of its denotatum). Frames can
be formalized if we think of them as grids with certain
predefined slots that get their values (i.e. morphosyntactic and
semantic information) once they apply to specific listemes.
Certain selectional restrictions may apply, but the task of
identifying a full set of them is regarded by Allan as
"impracticable and at worst impossible." In an attempt to
reconcile lexical and generative semantic approaches, Allan at
this point acknowledges Pustejovsky's work (e.g. 1995) on using
frames as the basis for the "co-occurrence constraint" when
building the lexical semantic structures of a generative lexicon.
Semantic fields refer to certain listemes and are defined by the
conceptual field in which the denotatum of the specific listeme
occurs. Conceptual fields are said to be in the same semantic
field if the listemes corresponding to each denotatum in this
field are semantically related (e.g. if the different denotata
bear certain similarities). This notion of conceptual field (vs.
the notion of semantic field) as presented here seems to be on
more of Allan's elaborated reconciliatory attempts to bridge the
gap between lexical and conceptual semantics on one hand, and
abstract semantic structures on the other. An example of a
semantic field encompassing different conceptual fields in
different languages is discussed: color. Some elementary
semantic concepts are now revisited in light of the notion of
scripts, frames and fields: hyponymy (Chaffin 1992), antonymy,
and meronymy. The focus then shifts into componential analysis
(Wilkins 1668) and a reminder of older "contrastive and
distributional" analyses (Harris 1951) in phonology and
morphology. And right when we thought that we had dispensed with
"semantic primitives", Allan comes back to this notion in the
frame of "compositional semantic analysis". Theories of J. Katz
(e.g. 1972), Weinreich (1966) and Wierzbicka (1980) are presented
and criticized as attempts for a valid semantic metalanguage.

After the introduction to cognitive semantics in Ch. 8, Chapter 9
presents some case-studies traditionally explored by cognitive
semanticists: the meanings of "back", colors, and classifiers. It
becomes clear that semantic categories are essentially
experiential: for instance, the way humans perceive the human
body (anatomically and physiologically) affects their cognitive
model for extensions of meaning of "back" to animate and
inanimate objects, spatial uses of time, direction and space
attributes, etc. Similarly, the way we see and categorize what we
see affects our way of naming and categorizing colors (MacLaury
1997, but also E. Rosch 1978 et seq.). Semantic (lexical)
categories are constructed in terms of "vantages", windows of
perception we select to pay attention to and in which similarity
and difference are being balanced or measured against a steady
factor (a fixed point of reference). Vantage theory starts with
"natural" categories and accounts for their perception, their
extensions and mappings into more synthetic classes: red, green,
blue are basic natural color categories, but, say, purple is not,
and the question for Vantage theory (and most of cognitive
linguistic theory) is how humans come to linguistically
categorize complex (i.e. composed of other elementary) entities.
Vantage theory is not an innovation, although Allan presents and
analyzes it at length: it's rather a sum up of older theories of
experiential subcategorization starting with Eleanor Rosch (or
even before Rosch, the Sapir-Whorf hypothesis). This chapter
quickly reviews the long-standing dispute on "linguistic
relativity". Classifiers are examined in light of linguistic
relativity and semantic crosslinguistic universalia. Even though
all the world's languages seem to adopt the same eight general
groups of classifiers for entity classification Allan proposes
(material make-up, function, shape, consistency, size, location,
arrangement and quanta), each one may choose in specific
occasions to focus on different salient features (attributes) of
the referent as perceived in various time and space coordinates.
Allan often uses the term "typical denotatum" denoting exactly
the use of exemplars in prototype semantics for recognition and
(sub)categorization.

Chapter 10 is a glimpse at prototype semantics touching upon
topics like the "typical denotatum", "gestalten", (cognitive
linguistic) stereotypes, intension and the intended referent.
Allan criticizes the prototype theory making a good point about
its low representational power. Emphasis on prototypes, Allan
says, does make respectable the correlation of semantics with
cognitive processes, but -- and this is his main point of
criticism -- prototype semantics does not reveal anything new
about the meaning of the listemes they examine. The question
Allan poses is the following: How does the cognitive model
speakers have internalized about "tomato" (a vegetable and fruit)
contribute to the semantic frame of this listeme? And --he
concludes-- prototypes does not answer this question. However,
Allan fails to see how semantic/conceptual prototypes define
linguistic constructions that -- in turn -- enhance
lexical/semantic frames. The Berkeley school of thought, that of
Construction Grammar (Charles Fillmore and Paul Kay 1993 et seq.)
examine how our internalized knowledge of prototypes and their
related categories and subcategories affects our lexical/semantic
frames for lexemes occurring in constructions of real-time
discourse. For instance, in account of polysemy A. Goldberg
points that "lexical items are typically associated with a set of
related meanings rather than a single abstract sense" (1995:44)
essentially decentralizing the cognitive process of composing and
decomposing semantic frames via a parallel process of discovering
the radial, experiential rather than abstract related categories.
Allan in his effort to balance cognitive and formalistic
approaches to semantics appears to have missed out in this turn
of cognitive linguistics. To remedy the perceived inadequacy of
prototype theory, Allan introduces Putnam's theory of gestalt, or
mental stereotype (Putnam 1979). A stereotype is a mental image
of a listeme that includes the attributes of its
exemplar/prototype (or "typical denotatum" in Allan's own
idiolect). In this image we often are called to reconcile
lexical/semantic incongruities, in semantically anomalous
expressions like e.g. "blue tomato". Instead of seeking to
categorize the referent (which prototype semantics does), gestalt
theory would attempt to resolve the perceive incongruity in the
expression. In spite of what Allan believes, Putnam's theory does
not necessarily negate or compete prototype theory. In fact, it
presupposes the existence of prototypical categorization with a
focus on the semantic incongruities often allowed in natural
language. Just because "red tomatoes" are typical, we can't deny
the potential existence of "blue tomatoes". Gestalts are being
perceived by means of a mechanism for individual member
subcategorization, although it is undoubtedly true that the whole
lies beyond the sum of its parts. The lens of perception and
experience and previous learning may give a gestalt different
colors. Natural language semantics should deal with what is
meaningful to humans, the dynamic semantic network that resonates
with their perceptions, cognition, learning and experiences,
instead of accounting for an abstract, hypothetical, and non-
experiential gestalt.

In Chapter 11 mood is explored "as primary illocution", and along
with it tense and modals and thematic roles are analysed. Mood
may indicate six primary illocutions in English (and most other
languages): declarative, expressive, imperative, interrogative,
interrogative subjunctive and subjunctive. Tense locates
situations in time with regard to a deictic center. Grammatical
aspect "captures some aspect of the development of an event."
Modals are also analyzed in their root, epistemic and deontic
meanings. Thematic roles are carried out by the arguments of
predicates and they depend on the roles the corresponding event
participants play in real discourse. Allan emphasizes the fact
that thematic roles are theory- dependent, and he goes on to
identify the roles of the effector (includes agent), the force,
the instrument, the experiencer, the locative (includes
recipient), the theme and the patient. The internal structure of
effectors is hierarchical, and Allan claims the same for other
roles. As a closing topic, Allan revisits GB's theta-criterion,
critically reminding that Chomsky based that on an earlier
criterion in Fillmore's case grammar, 1968). The theta-criterion
states that there is one thematic role per clause and no NP can
hold more than one thematic roles. Allan argues that verbs like
give-take and sell-buy allow their goals and sources to also be
agents, and that symmetric predicates and resumptive pronouns
also show that there can be more instances of a thematic role in
one clause. Because of the fuzziness in the definition of theta-
roles, Allan suggests the sole use of two "macroroles", that of
the "actor" and that of the "undergoer" in the grammar.

Chapter 12 focuses on clause predicates, a chance for the author
to present Jackendoff's "lexical conceptual structures" (1995,
1997) and Role and Reference Grammar's logical predicate
structures (Van Valin 1993). The latter seems to be presented
more extensively than the former. The main difference between
these two theories from Allan's perspective lies in the issue of
whether syntax is autonomous or not. In Jackendoff's theory
syntax is autonomous and semantics needs to integrate with a
rather dominant syntactic theory, whereas in Van Valin's theory
syntax is understood "with reference to its semantic and
communicative functions." However, Jackendoff's system includes
conceptual structures that Van Valin seems to ignore. Allan feels
the need to compare these two theories with Wierzbicka's Natural
Semantic Metalanguage (1980) discussed in earlier chapters, and
make it clear that Wierzbicka's theory was purely lexicographical
and does not seek to integrate with a syntactic theory. The
metalanguages Van Valin and Jackendoff use are --for Allan--
ingenious in that they bridge the gap between natural language
use and "a formal language such as predicate calculus and
intensional logic". Van Valin's logical structures and
Jackendoff's lexical conceptual structures are criticized as more
economical and precise metalanguages than Wierzbicka's.
Jackendoff's system is pretty comprehensive. It draws from the
generative schema of I-language and adds to it a particular
component, that of conceptual structure, that integrate with
syntax and phonology. Conceptual structure includes linguistic,
sensory and motor information for mental representation. Nouns
are regarded as semantic primitives, but verbs are decomposed
into function- arguments structures, as in predicate logic.
Argument fusion is the combinatorial process of combining the
semantic content of separate lexemes/constituents in a phrase
marker (by A-marking arguments). Argument linking is a
tranferring process of attaching A-marked constituents in the
lexical conceptual structure to appropriate nodes in the phrase
marker consistent with thematic and syntactic hierarchies. Theta-
roles are assigned to arguments via argument binding. Actor-
patient thematic relations are separated from motion and
location. In Role and Reference Grammar predicates are classified
into activities, states, achievements, accomplishments, and
causatives. Some discussion on the distinctive features of the
above classes are discussed as well as the fuzziness in the
distinction between achievements and accomplishments. In Van
Valin's system, theta roles have been summed up to the macroroles
of actor (roughly, logical subject) and undergoer (roughly,
logical object). Each of the macroroles hierarchically correspond
to thematic roles. More in depth analyses are offered for
ditransitive verbs (which are postulated as having two logical
structures), copular sentences and possessive constructions. The
notion of decomposition is revisited in light of the discussion
on metalanguages. Allan's persistence with the notion of semantic
decomposition is indicative of the importance he allocates to
metalanguage and the epistemology of semantic representation via
formal, formalistic and natural language methods.

Chapter 13 is devoted to English quantifiers and the semantics of
NP. Although it is just preliminary to the topic, it goes a long
way into analyzing "countability" in English language,
compositional quantification, relationships among quantifiers and
the semantics and pragmatics of simple sentences with
demonstratives, and other identity statements. Some
crosslinguistic observations on quantifiers are also included in
the analysis. Definite and indefinite articles are presented as
having other primary functions than quantification. Quantifiers
in different (but successive) propositions may entail each other
(monotonicity) and Allan notes that usually quantifiers
demonstrate predicate rather than NP monotonicity (scope and
compositionality of quantifiers were also examined). A typical NP
monotonic quantifier is "all". Definite article "the" is a
universal quantifier. The chapter closes with an exhaustive
analysis of a simple sentence that draws on theories,
terminology, tools and ideas presented throughout the book. In
the Epilogue of the book, Allan goes as far as answering the
question regarding the psychological reality of the semantics of
quantification.

OVERALL CRITICAL EVALUATION
Individual chapters have been commented on in the turn in which
they were presented above. As a general observation, a big asset
of Allan's book on natural language semantics from this
reviewer's point of view is the multitheoretical perspective
about natural language, the wealth of evidence and argumentation
from different (sometimes even contradictory) theories while
focusing on a particular topic, and the always present
informative "notes on further reading" for the curious reader.
This homo-universalis approach of Allan's exploration of
semantics causes only a minor issue for the more demanding
reader: certain topics are worth a deeper presentation, analysis
and argumentation that a multitheoretical textbook simply does
not have enough space to allow for. This book won't let you go
"out there" without a good overall grasp of the field, although
the beginner might be left with a bulk of often critical
information to absorb.

BIBLIOGRAPHY
Chaffin, Roger (1992) The concept of a semantic relation, In
Frames, Fields, and Contrasts, Lehrer and Kittay (eds.),
Hillsdale: Erlbaum, pp. 253-88.

Fillmore, Charles J., and Paul Kay (1993) Construction Grammar,
Unpublished manuscript, University of California, Berkeley.

Goldberg, Adele E. (1995) Constructions: A Construction Grammar
Approach to Argument Structure, Chicago: Chicago University
Press.

Harris, Zellig S. (1951) Methods in Structural Linguistics,
Chicago: University of Chicago Press.

Jackendoff, Ray (1995) The boundaries of the lexicon, In Idioms:
Structural and Psychological Perspectives, M. Everaert, Erik-Jan
van der Linden, A. Schenk, and R. Schreuder (eds.), Hillsdale:
Erlbaum, pp. 133-65.

Jackendoff, Ray (1997) Architecture of the Language Faculty,
Cambridge, MA: MIT Press. 	 Joos, Martin (1961) The Five
Clocks, New York: Harcourt, Brace and World.

Katz, Jerrold (1972) Semantic Theory, New York: Harper and Row.

Lyons, John (1997) Semantics, 2 Vols, Cambridge: Cambridge
University Press.

MacLaury, Robert (1997) Color and Cognition in Mesoamerica:
Constructing Categories as Vantages, Austin: University of Texas
Press.

Pustejovsky, James (1995) The Generative Lexicon, Cambridge, MA:
MIT Press.

Putnam, Hilary (1979) 2nd ed., Mind, Language and Reality,
Philosophical Papers, 2 Vols., Cambridge: Cambridge University
Press.

Rosch, Eleanor (1978) Principles of Categorization, In Cognition
and Categorization, Eleanor Rosch and L.L.Lloyd (eds.),
Hillsdale: Erlbaum, pp. 27-48.

Van Valin, Robert D. Jr. ed. (1993) Advances in Role and
Reference Grammar, Amsterdam: John Benjamins.

Weinreich, Uriel (1966) Explorations in Semantic Theory, In
Current Trends in Linguistics 3, Thomas A. Sebeok (ed.), The
Hague: Mouton, pp. 395-477.

Wierzbicka, Anna (1980) Lingua Mentalis: the Semantics of Natural
Language, Sydney: Academic Press.

Wilkins, John (1668) Essay toward a Real Character and a
Philosophical Language, London: Royal Society, Fascimile, 1968,
Menston: Scholar Press.

ABOUT THE REVIEWER
Eleni Koutsomitopoulou is currently completing her PhD in
Computational Linguistics at Georgetown University (Washington
D.C.) Her research interests include natural language processing,
knowledge acquisition and engineering, cognitive semantics,
metaphor and conceptual modeling with neural nets especially
designed for natural language. Her dissertation project is a
simulation of a biologically-inspired neural network system for
conceptual modeling of basic natural language constructs. She is
a research analyst at LexisNexis (Dayton, OH).
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue