Evans and Pourcel present a collection of very diverse articles in the field of
Cognitive Linguistics, which grew out of presentations at the first conference
of the Cognitive Linguistics Society of the UK. Naturally, such a volume does
not focus on one specific topic or approach. The editors have nevertheless
identified five major strands of research in Cognitive Linguistics, which are
reflected in the different sections of the book, some of them thematically
connected (Part I: Approaches to semantics: Theory and method, Part II:
Approaches to metaphor and blending: Theory and method, and Part III: Approaches
to grammar: Theory and method), some of them similar in their attempts to
integrate Cognitive Linguistics more broadly in the general Cognitive Sciences
and neighboring disciplines (Part IV: Language, embodiment and cognition: Theory
and application and Part V: Extensions and applications of cognitive linguistics).

In the first part, Peter Harder's article ''Meaning as input: The instructional
perspective'' discusses the theoretical problem of a usage-centered perspective
on language. One major claim of Cognitive Linguistics has always been that the
focus on ''competence'' at the expense of ''performance'' is misleading, since
linguistic abilities are not based on the existence of a language module, but
arise from general cognitive processes plus exposure to actual occurrences of
language - hence the idea of a ''usage-based model'' as an explanation of the
human ability to learn and use language(s). Although generally agreeing with
this view, Harder points out that adopting a ''usage-fundamentalism'' (p. 16)
would be inappropriate as well. Specifically, he is concerned about a change of
the term ''meaning'' that tends to only include ''meaning'' in actual utterances. He
argues that in assuming only encyclopedic knowledge (to the exclusion of
linguistic knowledge) would mean that there is no ''semantic pole'' in the meaning
of linguistic entities anymore, and that we are not able to understand the
genuine semantic contribution of linguistic forms apart from their function in
actual utterances. He therefore proposes a ''tripartition'' of language
understanding and production into input, processing, and output. ''Linguistic
meaning'' and ''linguistic knowledge'' would then be placed in the input realm,
being the knowledge about the ''input properties'' of linguistic entities. This
enables speakers and hearers to choose appropriate linguistic forms for the
encoding and understanding of a message.

Vyvyan Evans, in his article on ''Semantic representation in LCCM Theory'', deals
with a similar problem, namely the nature of linguistic meaning in contrast to
concepts. Evans, however, proposes two distinct systems, the linguistic and the
conceptual system. Therefore he names his semantic theory the ''Theory of Lexical
Concepts and Cognitive Models.'' These two systems contain different types of
semantic representations: lexical concepts and cognitive models, respectively.
The linguistic system serves as an ''executive control function'' (p. 53) that
connects the two levels in order to enable the speaker to communicate with
language. Evans discusses the consequences of such a view for some concepts of
cognitive semantics, such as Talmy's distinction between open- and closed
class-elements (Talmy 2000) and Barsalou's frame semantics (Barsalou 1992).

The third article in this first part of the book, ''Behavioral profiles: A
corpus-based approach to cognitive semantic analysis'' by Stefan Th. Gries and
Dagmar Divjak, is concerned with a more concrete problem in semantics. The
authors provide an empirical account of the problem of polysemy and near
synonymy, starting with a short critique of existing empirical studies in
cognitive lexical semantics and their often vague criteria to distinguish
different senses of a word from slightly different conceptualizations in
differing usage events. Adopting a corpus-based perspective, they point out that
corpus data can be used for much more than just establishing collocational and
colligational behavior of single lexical items, but can yield results relevant
to cognitively inspired lexical semantics. The basic idea behind this claim is
that relations between senses of a word should be reflected in corpus data,
because the likelihood of similar behavior of linguistic elements is an index
for the strength of the connection between different senses in a relational
network. Their case studies on the polysemous English word 'run' and nine
near-synonymous Russian words meaning something close to 'try' show how their
notion of behavioral profiles and different quantitative methods for their
evaluation leads to a more precise description of polysemy and synonymy, two
phenomena that have been crucial for the development of some core concepts in
cognitive linguistics like radial networks and prototypical meanings.

Dylan Glynn's article on ''Polysemy, syntax, and variation'' starts out with a
very similar problem, the need for a method that allows the proposal of semantic
networks on the basis of linguistic data instead of native speakers' intuition.
After a discussion of existing approaches to this problem, Glynn argues for the
adoption of a usage-based perspective - which is very much in line with the
theoretical presuppositions of cognitive linguistics. Quantitative corpus
analyses are one appropriate way to observe natural language use, and therefore
the author suggests a multifactorial analysis of corpus data that takes
regional, social, and register variations into account as a possible solution
for the problem. Applying this method in a case study on the English word
'hassle,' he shows how this approach reveals variational patterns not accessible
through intuition.

The second part of the book, ''Approaches to metaphor and blending,'' offers some
new perspectives on a by now classic topic of Cognitive Linguistics.

Mimi Ziwei Huang's article ''Solving the riddle of metaphor: A salience-based
model for metaphorical interpretation in a discourse context'' discusses the
activation of metaphorical meanings in discourse. Based on the Graded Salience
Hypothesis (Giora 1997), which claims that the most ''salient'' (here meaning the
most frequent and familiar and thus most of the time literal) meaning of a
polysemous lexical item is always activated during metaphor comprehension. Huang
argues that contextual information has to be included in the notion of salience,
since certain meanings can be made salient in discourse, even if this meaning is
not the most frequent and most prominent in the mental lexicon. By analyzing a
short story, the author aims to show how this ''salience'' of a certain meaning
can be developed in discourse. However, her implicit understanding of ''metaphor''
and her example differ considerably from typical metaphors that have been the
topic of a large number of theoretical and empirical studies in Cognitive
Linguistics, since there seems to be no systematic mapping from one domain onto
another. Still, the psycholinguistic implications of an extended Graded Salience
Hypothesis as proposed by Huang seem nevertheless worth more systematical
testing, since the results would be a valuable contribution to the discussion of
conceptualizations and processing of metaphors.

In his contribution ''When is a linguistic metaphor a conceptual metaphor?'',
Daniel Casasanto questions one of the core assumptions of Conceptual Metaphor
Theory (Lakoff & Johnson 1980, 1999) - the conceptual basis of linguistic
metaphors. Casasanto points out the explanational gap between purely linguistic
data and assumptions about cognitive processes and conceptualizations. Taking
this theoretical concern as a starting point for an experimental study on the
conceptual metaphor SIMILARITY IS NEARNESS, he shows empirical evidence for
conceptualizations coherent with metaphorical patterns in language and thus
confirming the predictions of Conceptual Metaphor Theory on the one hand, and on
the other hand results of a very similar experiment that run counter the
observed linguistic patterns. These seemingly incoherent results lead to serious
doubts about the very general tenets of Conceptual Metaphor Theory. Casasanto
explains his results with a difference between high- and low-level cognitive and
perceptual processes and argues that the theory can be partly preserved, but
that is is absolutely necessary to base claims on conceptualizations on
experimental empirical evidence.

The chapter by Gilles Fauconnier with the title ''Generalized Integration
Networks'', argues that his approach to conceptual integration is able to make
appropriate distinctions in the description of language and cognition. The aim
of his paper is to take the systematic study of integration as a cognitive
routine beyond the phenomenon of obvious and visible ''blends'' in order to create
a much more general notion of this cognitive process. Fauconnier mentions three
assumptions that could be used as arguments against such generalization, namely
the assumption that different surface products result from different cognitive
operations (1), the assumption that something ''new'' is always connected to
higher cognitive costs (2), and the assumption that a very wide-ranging
cognitive operation explains ''too much'' and is unconstrained (3). He discusses
these in turn and shows why he thinks that they are wrong.

In the following article on ''Genitives and proper names in construction blends,''
Barbara Dancygier presents an application of the Conceptual Integration approach
to the phenomenon of the English genitive. Describing constructions as blends,
Dancygier attempts to explain how concepts must be integrated in the GEN-XYZ
construction (for example ''Cambodia has become Vietnam's Vietnam'') and the One
person's X is another person's Y construction (for example ''One person's trash
is another person's treasure''). She shows how Conceptual Blending Theory can
account for the meaning relations between the different slots in these
constructions. This approach also allows an explanation for the constraints of
similar expressions.

The third section of the volume deals with ''Approaches to Grammar.'' The author
of the first article in this part of the book, Arne Zeschel, asks ''What's (in) a
construction? Complete inheritance vs. full-entry models.'' Zeschel compares
approaches which treat constructions as one linguistic unit among others with
theories that assume that language consists only of constructions of different
complexity and abstractness, and argues that this distinction is more than a
terminological argument but crucial for empirical approaches. His own corpus
study on the [good NP] construction (a conventionalized instantiation of this
construction would be ''good for a laugh'') shows that speakers' knowledge about
possible and impossible applications and variations of certain constructions is
coherent with the assumptions of a usage-based model that does not base the
postulation of a construction on the non-predictability of this structure.

Eva Dabrowska, in her article ''Words as constructions'', discusses the
acquisition of an elaborated lexicon. Given the fact that many words are not
encountered in contexts that give a direct clue to their meaning, the syntactic
frame a word occurs in gives must hints to the general meaning, but there is no
way of extracting the exact meaning directly from the context. Dabrowska
presents two studies on the understanding of English manner-of-motion verbs. In
the first study, Dabrowska shows on the basis of elicited data that nearly
synonymous verbs occur with semantically different collocates (for example, some
manner-of-motion verbs are strongly connected to plural subjects or certain
paths). The results show that elicited sentences exaggerate tendencies also
found in corpora, which is evidence for the fact that even though speakers are
often not able to give clear definitions of word meanings, they have
considerable implicit knowledge on typical usage and collocation patterns. The
results of the second study, where participants had to match verbs with
dictionary definitions, constructions, and motion events in video clips, support
the hypothesis that collocational knowledge is psychologically basic. Dabrowska
therefore argues that words are constructions and can thus be learned from a
purely linguistic context. She suggests that this is possible because typical
collocation patterns are memorized.

Ronald W. Langacker's article on ''Constructions and constructional meaning'' aims
at a clarification of the question if whether certain aspects of clausal meaning
can be ascribed to the predicate or to constructional meaning alone. He argues
that a sharp distinction between these two options is artificial. If a
compositional structure is constructed by a rule, this means that a
constructional schema is instantiated. The complex cognitive process of
instantiating and activating connected extracompositional meaning can be a more
or less entrenched cognitive routine. Langacker claims that different linguistic
phenomena like grammatical composition, and conceptual metaphor can all be
explained by general cognitive processes like recognition, categorization, and
apprehension of something as something. Therefore he claims that the difference
between syntagmatic and paradigmatic relationships, constructions and
instantiations as well as other traditional distinctions like lexicon vs.
grammar, derivation vs. extension are better described in a non-dichotomous way.

Edith A. Moravcsik's article ''Partonomic structures in syntax'' is a study in
cognitive metalinguistics. The author asks why it is useful to assume partonomic
relationships in linguistic structures and shows that the assumption of ''wholes''
made from different ''parts'' such as phrases and clauses enable the linguist to
formulate syntactic rules, often by paying the price of overgeneralization or
creating unappropriate complexities. Because of that, she suggests a general
solution to this problem, namely positing different layers for complex
structures, which she argues is a normal argumentative cognitive process.

Part four of the book, ''Language, embodiment, and cognition: Theory and
application,'' starts with an article by Chris Sinha, ''Language as a biocultural
niche and social institution.'' Sinha asks how it might be possible to accept and
describe language as being both an important feature of human biology and a
socio-cultural institution, and thus to overcome the old dichotomy between these
two view. He offers a biocultural view on language and language acquisition,
showing how a certain genetic and biological features of organisms can yield
cultural behavior, for example rituals or semiotic systems, which can be
inherited in evolutionary processes. This view places language and grammar
alongside with other culturally induced behavior that nevertheless rests on
abilities residing in human biology. Sinha's approach therefore shows how to
avoid the pitfalls of purely nativist vs. culturalist theories of language in a
very general way.

Madga Altman describes how traditional medical models reflect folk theories of
the body and bodily processes and functions. In her article ''Understanding
embodiment: Psychophysiological models in traditional medical systems,'' she
shows that prominent conceptualizations of body functions in traditional Chinese
medicine correspond with Talmy's Force Dynamics. Therefore, she argues that
these medical models can be a relevant source in determining universal as well
as culture-specific conceptualizations. According to the author, this could help
to understand the nature of embodiment. How this should be done exactly,
however, remains rather unspecific.

Paul Chilton's article '''Get' and the grasp schema. A new approch to conceptual
modelling in image schema semantics'' presents an analysis of the various uses of
the English verb 'get'. Chilton proposes that the different construction
meanings associated with get he describes in detail are a conceptual category of
related senses, all derived from a prototypical instantiation, the meaning of
which is embodied in an image schema, the ''prehension image schema.'' The
author's analysis rests on an innovative formalization of spatial concepts such
as viewpoint, distance, and direction. It would have been interesting, however,
to make the differences between existing diagramatic notations (such as
Lakoff's, Talmy's, and Langacker's, which Chilton criticizes in passing) more
explicit, discussing which differences are crucial and which are notational
variants.

The last article of this section of the book, ''Motion scenarios in cognitive
processes,'' by Stéphanie Pourcel, aims at an empirical validation of certain
assumptions following from Talmy's (1985, 2000) and Slobin's work on typological
differences in the encoding of motion events (see for example Slobin 2004).
She explores if speakers are influenced by their native language when speaking
and inferencing about, or recalling motion events. Comparing speakers of French
and English, two typologically different languages, Pourcel reports differences
in speaking about and recalling motion events. The differences correspond to the
preferred lexicalization patterns (Talmy 1985, 2000) in their respective mother
tongues. Pourcel also finds differences in inferencing about motion between the
two groups of speakers, however, these differences are a little more difficult
to explain. Pourcel concludes that Germanic and Romanic languages do indeed
differ in their preferred narrative styles, which could also explain linguistic
relativity effects.

The last section of this volume is called ''Extensions and applications of
cognitive linguistics,'' and combines work in the linguistic analysis of
dialogical interaction and narratives, and Cognitive Poetics, taking Cognitive
Linguistics theoretical framework and - in the case of the former - methods
beyond lexical and syntactic phenomena.

William Croft's contribution points out that Cognitive Linguistics must include
the social aspects of language in order to become a serious, full-fledged
theory. Following the title of his article, ''Towards a social cognitive
linguistics,'' he discusses possible ways of integrating existing research from
pragmatics and sociolinguistics as well as language acquisition and psychology.
His main argument is that most language use is situated and therefore unique.
The ascription of meaning to utterances is a joint action and therefore social
activity. Croft argues that Cognitive Linguistics should not only focus on what
is ''in the head'' of individual speakers, but explain what happens ''outside the
head'' in interaction and communication.

Ruth Berman and Bracha Nir develop in their article on ''Cognitive and linguistic
factors in evaluating text quality'' an analytical tool for the evaluation of
texts. Introducing several measurements for text production proficiency, they
combine quantitative evaluations of local linguistic expressions (like word
length, lexical density, register, and semantic abstractness) with ''global text
quality'' (measured through the presence of top-down and bottom-up
generalizations, overt categorial structure, and relations between beginning and
ending of a text). Interestingly, although both measurements correlate with age,
suggesting that the ability of text production develops well into adulthood,
they are not directly connected to each other, as the authors show with their
case study. Berman and Nir therefeore conclude that the question how cognitive
and linguistic factors interact in the development of discourse abilities
deserves more in-depth research in Cognitive Linguistics.

In her article on ''Reference points and dominions in narratives,'' Sarah van
Vliet aims at an extension of Van Hoek's (1997) sentence-level analysis of
reference maintenance to the level of whole narratives. She proposes that
reference point organization in narratives can be described as a part of
attention framing, which contributes to the construal of referent salience and
discourse connectivity and explains constraints of anaphoric relations within
the narrative as a whole.

Johanna Rubba's article ''The dream as a blend in David Lynch's Mulholland Drive''
applies Blending Theory to the interpretation of a film. Showing the relations
of concepts and frames in the complex blends in that film, she aims at
developing an analytical tool for film and literature in