LINGUIST List 11.1672

Tue Aug 1 2000

Review: Bloom et al. Language and Space

Editor for this issue: Andrew Carnie <carnielinguistlist.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 Andrew Carnie at carnielinguistlist.org

Directory

  1. Zouhair Maalej, Review of Bloom et al (1999)

Message 1: Review of Bloom et al (1999)

Date: Sun, 16 Jul 2000 23:29:24 +0200
From: Zouhair Maalej <zmaalejgnet.tn>
Subject: Review of Bloom et al (1999)

Bloom, Paul, Mary A. Peterson, Lynn Nadel & Merrill F. Garrett 
(eds.) (1999). Language and Space. Cambridge, Massachusetts/London, 
England: The MIT Press (597 pages, ISBN: 9 780262 522663 90000). 

Reviewed by Zouhair Maalej, University of Tunis I 

The book is a collection of papers from the Tuscon, Arizona, 
conference held in 1994, and edited by a team from the same 
university. The collection is interdisciplinary by including papers 
from verbal and signed languages, from neuropsychological and 
computational approaches, from formal and cognitive perspectives. 
For a complementary review, colleagues are advised to consider the 
one written by Anne Reboul for the same book published in 1996 (cf. 
Linguist, 10.1950, 16 December, 1999). 

SYNOPSIS 

1. 'The Architecture of the Linguistic-Spatial Interface'

Jackendoff proposes to ground language and spatial cognition in 
Representational Modularity (RM) and Interface Modules, which are 
assumed to resolve the lack of trade-off between modules in the 
Fodorian model of mind. Conceptual structure (CS) is language-
independent, therefore non-propositional. In spite of the 
universality of CS, languages tend to differ in (i) packaging 
concepts in lexical items, (ii) spelling out concepts in syntax 
(e.g. expressing social meaning in French through the tu-vous 
distinction)and (iii) using some syntactic constructions to encode 
specific conceptual notions. CS is different from spatial 
representation (SR) in that it is image-schematic (i.e. cognitive). 
The CS-SR interface contributes to economy by disallowing the 
duplication of information, and assuming 'a criterion of 
interfacing' that warrants a requisite overlap for modules to 
intercommunicate. Other dimensions of economy include the criterion 
of 'grammatical effect' and the criterion of 'non-spatial 
abstraction.' Thus, word meaning is a composite of linguistic 
structure (phonology & syntax) and non-linguistic knowledge (CS & 
possibly SR), where linguistic structure is ancillary to the 
conceptual-cum-spatial knowledge which it serves to express. 
Reacting to the literature on frames of reference that postulates a 
binary distinction between an intrinsic (or object-centered frame) 
and a deictic (or observer-centered) frame, Jackendoff assumes the 
existence of four frames under intrinsic and four others under 
deictic. Intrinsic frames include geometric, motion, canonical 
orientation, and canonical encounter frames, while deictic or 
environmental frames include gravitational, geographical, 
contextual, and observer frames, with only the geometric and motion 
frames free from the influence of the environment. Drawing on 
Narissiman (1993), Jackendoff defends a SR hypothesis, whereby 
spatial information is generated in SR rather than in CS. 

2. 'How Much Space Gets into Language' 

Bierwisch is much concerned with the natural language components 
that support spatial information. Although working within a lexical 
semantics framework, he observes that the same syntax can express 
spatial and non-spatial contexts, and identical spatial contexts may 
be expressed by different syntactic constructions, without the 
further claim that phrase structure reflects spatial information. 
Grounding lexical semantics in Chomsky's I-language linguistics 
based, Bierwisch adopts 'semantic primes.' Unlike Jackendoff who 
assumes CS to be extralinguistic knowledge, and who argues for the 
autonomy of modules and for the existence of interface systems, 
Bierwisch holds semantic form (SF) to be a module of I-language, and 
claims going back to Chomsky's modularity, which he calls 
'modularity of knowledge' (as an alternative to Jackendoff's RM). On 
this model of knowledge, SR is called I-space (after I-language). SR 
is assumed to interface with (i) perceptual and motor modalities, 
and (ii) the conceptual system C-I. Being transmodal, depictive, and 
domain-specific, SR is based on locations determined by 
dimentionality, topological structure, metrical structure, 
orientational structure as based on gravitation, and canonical 
position or motion. Thus, SR is distinct from CS, in that it is a 
level of representation drawing on different systems of mental 
organization (with visual perception providing it with fundamental 
support) rather than a separate module. Thus, at the level of 
representation SR does not interrelate with CS but belongs in C-I. 
The SR-CS interface is reduced to matters of shape, place, size, and 
paths of moving objects. Bierwisch posits two hypotheses, the second 
of which he believes to be more accurate: basic structures of 
spatial organization must either function as 'a general schema of 
conceptual knowledge,' or as 'an intrinsic condition for I-space.' 
The distinction Bierwisch makes between 'implicit transfer' (where 
the conceptualization of time and social hierarchy is implicit) and 
'explicit transfer' of spatial structures (which is at the origin of 
the blurred distinction between spatial and non-spatial terms). The 
way CS relates to I-space determines (i) the shape and size of 
objects, (ii) the fact that in relation to the spatial properties of 
objects encyclopaedic knowledge works with basic-level categories 
but not with superordinates, and (iii) which units involve only 
spatial information and which ones include further conceptual 
knowledge. This suggests a classification of concepts as (i) 
aspatial (emotions, time), (ii) extrinsically spatial (superordinate 
categories), (iii) intrinsically spatial (basic-level categories), 
and (iv) strictly spatial (some of the implicitly and explicitly 
spatial categories). The same classification is applied to colours 
and time. Concepts correlate with different syntactic and 
morphological categories and different conceptual spatial 
categories. Concepts that combine spatial and encyclopaedic 
knowledge (like most basic-level categories) are called 'rich 
concepts,' while those involving restricted domains are called 
'spare concepts.' SF relates to CS by being simply embedded in it in 
order to be able 'to recruit pertinent configurations.' 

3. 'Perspective Taking and Ellipsis in Spatial Descriptions' 

Levelt argues that speech production involves macroplanning (i.e. an 
informative content whose expression reveals intentions best) and 
microplanning (i.e. a manner of lexically conceptualising content). 
This process of abstraction is termed by Levelt as 'perspective 
taking,' which involves: (i) focusing on some portion of the scene 
('referent'), (ii) focusing on the referent's location ('relatum'), 
and (iii) expressing the referent's relatum ('perspective system'). 
Perspective taking (PT) is multiple (deictic, intrinsic, absolute) 
and language independent, with cross-cultural PT as a confirmation 
for this freedom of conceptual PT and linguistic expression of the 
choice of a given PT. However, rather than being randomly made, the 
choice is pragmatically motivated. With regard to the converseness 
and transitivity properties, the deictic and absolute systems work 
but not the intrinsic one, making the latter's inferential potential 
low in spatial communication. Misunderstandings (as in the case of 
alignment and pre-emption) between communicators may arise when they 
are not on the same wavelength as to from what perspective system 
spatial communication is conceptualized, or when systems conflict 
with one another. For that, Levelt posits a 'principle of canonical 
orientation,' and argues for the primacy of conceptualization of a 
given PT ('thinking for speaking,' borrowed from Slobin, 1987) over 
the choice of a lexical item. To capture the language-cognition 
interface, Levelt discusses the ellipsis of spatial expressions, and 
distinguishes two types of ellipsis: deep ellipsis (DE) and surface 
ellipsis (SE). The dichotomy DE (occurring before PT) and SE (taking 
place after PT) only arises with the intrinsic PT. Levelt concludes 
that PT systems are 'interfaces between our spatial and semantic 
'modules'' in Jackendoff's sense. 

4. 'Frames of Reference and Molyneux's Question: Crosslinguistic 
Evidence' 

 Levinson starts his paper by reviewing the notion of frame of 
reference, and argues that, apart from the classical distinction 
between 'intrinsic frame' and 'relative frame' which seemed to 
capture spatial representations, an 'absolute frame' is also 
possible in some cultures like that of Tzeltal. In a series of 
experiments contrasting Dutch and Tzeltal, the Tenejapans are 
described as consistently adopting the absolute frame of reference 
across modalities while the Dutch adopted the relative frame, quite 
in line with the 'semantics of spatial description in the two 
languages.' Quite amazingly, such results confirm the long-standing 
Whorfian tradition, and shake the modularity of mind theory. 
Levinson distinguishes many binary frames of reference that we make 
use of, namely, relative vs. absolute, egocentric vs. allocentric, 
viewer-centered vs. object-centered, orientation-bound vs. 
orientation-free, deictic vs. intrinsic, and the more complex 
ternary distinction between viewer-centered vs. object-centered vs. 
environment-centered (note that these crosscut across disciplines), 
but ends up collapsing the linguistic frames of reference into the 
more canonical frames of intrinsic, relative, and absolute, where 
intrinsic and absolute are said to be allocentric frames, and 
absolute and relative are orientation-bound frames. However, in 
spite of this correlation the frames remain ''untranslatable' from 
one to the other,' which leads to the question of whether spatial 
perception and conceptualization are modality-bound. Levinson's 
answer seems to be negative, implying that subjects use the same 
frame of reference across modalities, although no evidence exists as 
to how such a cross-modal transfer takes place. 

5. 'The Confluence of Space and Language in Signed Languages' 

 Emmorey's thesis is that, since signed languages (SLs) combine 
space and language to signify spatial locations, object orientation, 
and point of view, they can be used as an index to the language-
spatial representation (SR) interface. SR is found at the 
phonological level when different lexical items are signed in 
relation to the body, at the morphological level when complex forms 
are signed 'by nesting a sign stem within dynamic movement contours 
and planes in space.' The visual-gestural combination in SLs to 
encode space enables signers to simultaneously use the intrinsic and 
deictic frames. Emmorey reports on evidence whereby the right-
hemisphere, which is known to monitor the visual-spatial function, 
is more involved in processing spatial information. 

6. 'Fictive Motion in Language and 'Ception.'' 

 Talmy focuses on fictive motion (FM), which is defined in terms of 
its relation to 'the imaginal capacity of cognition.' One important 
form of FM is 'emanation,' which includes four types of path: (i) 
orientation path (OP), (ii) radiation path (RP), (iii) shadow path, 
and (iv) sensory path. The OP of a FM is a linear intangible entity 
originating in the front of some object and moving away from it, or 
moving laterally with regard to it. The radiation path emanates from 
an energy source, and moves away from it. The difference between a 
RP and an OP is that the latter's line of motion is imperceptible. 
The shadow path describes the shadow of an object as moving away 
from that object. Including visual paths, the sensory path 
establishes a straight path between Experiencer and Experienced or 
vice versa. The most privileged cases of sensory paths in language 
have to do with the senses of sight, audition, and olfaction. 
Cognitively, the less dependent entity is conceptualized as source, 
which is called 'active-determinative principle.' The reason why 
this principle holds has to do with agency, even though the latter 
cannot be ascertained to be either learned or innate. Underlying 
agency are intention and its realization. Intention is reminiscent 
of Searle's (1983) Intentionality, and can be captured in the 
Intentional State of desire for the existence of a state of affairs, 
with the realization component as the felicity condition of agency. 
Agency being active and determinative, its realization depends on 
matching the desire, and affecting a distal physical object, for 
instance. The emanation category of fictive motion in language has 
its analogue in other cognitive systems like perception, where more 
psychological research is needed. There is, thus, more to cognitive 
organization than the modular view, where various modules specialize 
in different tasks a cognitive system not only has its own 
specificity, but also shares structural properties with some or all 
other systems. Talmy extends his analysis of emanation to cultural 
concepts, with which they share a deeper cognitive connection. 
Beside emanation, there exist five other fictive motions, which are 
said to have parallels in visual perception, namely, pattern paths, 
frame-relative motion, advent paths, access paths, and coverage 
paths. For that, Talmy offers a general cognitive domain he calls 
'ception' to account for conscious and unconscious cognitive 
representations that are the output of both conception and 
perception. The proposal is a mixture of cognitive and pragmatic 
phenomena, and includes: palpability, clarity, strength, ostention, 
objectivity, localizability, identifiability, structure, type of 
geometry, accessibility, certainty, actionability, and stimulus. 
Talmy establishes a relation between fictivity and the Lakoffian 
theory of metaphor, where the discrepancy between target domain (TD) 
and source domain (SD) accounts for factive and veridical and 
fictive and less veridical, respectively. Talmy ends his paper by 
arguing for 'cognitive dynamism' as a more default mode of ception 
than cognitive staticism. 

7. 'The Spatial Prepositions in English, Vector Grammar, and the 
Cognitive Map Theory' 

 Even though O'Keefe adopted a Kantian view of space as absolute, he 
grounded the prepositional system of English in physical experience. 
The structure of his cognitive map theory is possible to study if 
the contention that the left human hippocampus, as a spatial mapping 
system, is transformed to store linguistic information. A vector 
grammar realises this spatial mapping through prepositions, whose 
role is to 'provide the spatial relationships among a set of places 
and objects and to specify the movements and transformations in 
these relationships over time.' Prepositions are defined as 
essentially spatial, and their non-spatial dimensions are acquired 
metaphorically. Satisfying a more nominal scale (vs. ordinal or 
interval), the below relation is transitive, and allows for both 
allocentric (universal gravity) and egocentric (object-centered) 
uses. The down relation involves a 'reference plane' (plane or line) 
and a 'reference entity' (place or object). O'Keefe explains the use 
of influence, control, and social status through vertical 
prepositions metaphorically. 

8. 'Multiple Geometric Representations of Objects in Languages and 
Language Learners' 

 Landau is interested in how children represent space, and argues 
that 'a crucial part of learning the mappings is properly 
representing objects in terms of their distinct relevant geometrical 
descriptions.' The reading of reference objects in abstract 
geometric descriptions can be motivated by the preposition coercing 
us into a process of schematization, where virtual volumes (e.g. 
birds in a tree) or virtual lines (e.g. customers in a line) are 
conceived as possible. Finding Paige's topological representation of 
space too weak to account for children's spatial knowledge, Landau 
adduces evidence to this effect. The representations of objects by 
English children are of three types: (i) coarse , (ii) axial, and 
fine-grained. In one experiment, Landau & Stecker (1990) concluded 
that three-year old children are capable of representing the figure 
object in a coarse way, ignoring shape completely. In another 
experiment, children tended to ignore detailed shape, and 
concentrated on its principal axis. In yet other experiments, it was 
evidenced that children adopted a richer perspective, where shape, 
objects parts, their spatial relationships, and their motion combine 
into a fine-grained account. 

9. 'Preverbal Representation and Language' 

 Mandler is interested in preverbal conceptual representation (PCR), 
which is held to be the basis of language and learning. Her main 
argument is that 'language is structured in spatially relevant ways 
because the meaning system of the preverbal language learner is 
spatially structured.' PCR is spatial not in the sense of spatial 
perception, but in the sense of 'meaning packages' that retain some 
spatial characteristics. Unlike Talmy who uses 'ception' to unite 
perception and conception, Mandler conceives of perceptual and 
conceptual categories as different because they follow two different 
developmental paths. Spatial information is easier for children to 
use and process than temporal data, which explains their using 
spatial conceptions in talking about time. In this sense, the 
concept of space is a primitive one, with the concept of time 
deriving therefrom in the guise of image-schemata. Other derivative 
schemata are dynamics and internal feelings. PCR as encapsulating 
perceptually spatial phenomena is the basis for the distinction 
between basic-level categories (e.g. cat and dog), which get 
linguistically differentiated by the sounds they make. Presumably, 
this is a precursor for the noun-verb distinction in the grammar. 
Shape at this level is not helpful for such a distinction. The 
notions of agent and patient, on the other hand, are arrived at by 
perceiving in the environment objects that initiate motion in space 
and other objects against which force is exerted, which underlie the 
transitive-intransitive distinction. Thus, the language learner 
comes to the learning task equipped with preverbal meanings or 
schemata such as containment, contact, and support. Such a 
conception of language learning necessarily holds 'language to be 
mapped onto a meaning system that forms an interface between analog 
and digital forms.' 

10. 'Learning How to Structure Space for Language: A Crosslinguistic 
Perspective' Adducing cross-linguistic evidence, Bowerman argues 
that nonlinguistic and linguistic understanding of space do not 
entertain a one-to-one relation. While agreeing with Mandler that 
the former lays the foundation for the acquisition of spatial 
language, she is convinced that this is not enough. Evidence exist 
in favor of the temporal primacy of nonlinguistic spatial knowledge, 
of the reliance of children on this knowledge in learning spatial 
words, and of the use of such knowledge as a learning strategy. What 
also plays a role in reinforcing spatial language is linguistic 
input, i.e. children's exposure to adults' talk about space. 
Crosslinguistic evidence from English and Korean suggests that, in 
spite of the assumed preverbal cognitive spatial structures, English 
children grow up expressing path notions with a separate preposition 
or particle which they juxtapose to the verb, whereas Korean 
children express path in the verb itself. Disagreeing with Mandler 
on the total mapping of linguistic over prelinguistic information, 
Bowerman argues that image-schemata cannot be relied upon to account 
for the linguistic representation of space in children if only 
account is to be taken of productivity in the use of spatial 
morphemes. A more likely explanation is that 'children build spatial 
semantic categories in response to the distribution of spatial 
morphemes across contexts in the language they hear.' This process 
is not even for all children across languages as semantic packaging 
of spatial notions varies from one language to another. What seems, 
however, to play a more primordial role in this is semantic 
primitives. 

11. 'Space to Think' 

 Johnson-Laird argues that spatial reasoning is model-based. Spatial 
representation through models is more commonsensical than that with 
formal rules of logic. According to him, the advantages of a model 
are the fact that: (i) it yields a 'decision procedure,' and (ii) it 
extends 'to the informal arguments of daily life,' which resist the 
application of formal rules of inference. The apparatus required to 
reason model-wise includes the procedures that construct models, the 
formulation of true conclusions in those models, and the testing of 
conclusions in them. Extrapolating from work on spatial reasoning, 
Johnson-Laird experimented with reasoning by space for time. His 
conclusions point to the fact that 'reasoning about temporal 
relations depends on mental models of the sequence of events.' 
Contrasting work with formal rules and work with models, Johnson-
Laird concludes that subjects err because they misapply a rule; 
however, reasoners with models err because they fail to apply all 
the models of the premises. The time allotted for reading premises 
is longer with multiple models and shorter with one-model problems. 
Johnson-Laird ends his paper by pointing out the importance to 
thinking of diagrams, which facilitate thinking via their iconic 
nature. Reasoning from verbal information is, thus, less easy than 
from diagrams. Reasoners with models not only use spatial models to 
reason about space, but also use such models to reason at large. 

1 'Spatial Perspective in Descriptions' 

 Tversky's paper is reminiscent of Levelt's 'perspective taking,' 
but with a difference. The spatial perspective (SP) Tversky is 
interested in is rather the one that the viewer borrows from other 
points of view: 'Perspective in language use is of particular 
interest because language allows us to use perspectives other than 
those given by perception.' Experiencing the world from someone 
else's perspective can be cognitively and socially rewarding. 
Tversky reviews some of the views with regard to perspective taking 
such as in object recognition, environmental cognition, 
neuropsychological support, and spatial language. These are 
cognition-dominated. The social categories of spatial reference are 
no less important. They include: (i) the speaker's (ii) the 
addressee's - or (iii) other's (neutral) perspective. Whatever 
perspective is adopted, consistency (of a default perspective) and 
order of presentation (regardless of perspective) have been noted. 
The criteria for choosing one SP are pragmatic, mostly having to do 
with (i) cognitive ease or difficulty for the addressee to follow 
the speaker's spatial descriptions, (ii) audience (where speakers 
try to adopt the addressee's perspective). To illustrate this, 
Tversky describes three experiments on narratives and PT, where 
subjects were asked to identify perspective(s), and reaction time 
patterns were identified. Motivated by social concerns, Tversky's 
conclusion is that (i) perspective subdivides into personal and 
neutral perspectives, with the personal in turn subdividing into 
speaker and addressee and (ii) not only is there no default 
perspective, but adopting different perspectives on one occasion is 
not uncommon. 

13. 'A Computational Analysis of the Apprehension of Spatial 
Relations' 

 Gordon & Sadler present a psychological theory of apprehension of 
spatial relations computationally. After reviewing basic, deictic, 
and intrinsic relations, Gordon & Sadler argue that this 
distinction has two implications for computation: (i) since the 
reference frame is central to the meaning of spatial relations, 
therefore the reference frame is also central to the reference frame 
computation in the process of apprehension and (ii) since reference 
objects and located objects are distinct, spatial relations must be 
individuated by adopting spatial indexing. Intrinsic relations are 
computationally more complex than deictic relations. Gordon & Sadler 
propose spatial templates as determining the goodness of fit of 
spatial relations to spatial objects. The computational theory of 
apprehension works by combining various levels of representation: 
(i) perceptual (including implicit information about objects and the 
spatial relations between them), (ii) conceptual (being addressable 
linguistically, this representation identifies the spatial relation, 
individuates its arguments, identifies its reference frame and the 
relevant spatial template), (iii) reference frame (being flexible 
and giving direction to perceptual space and conceptual 
representation, a frame serves as a map between the perceptual and 
conceptual representations), and (iv) spatial template (taking care 
of the zones of acceptability associated with a given spatial 
relation). Apart from representations, a computational theory of 
apprehension involves processes such as spatial indexing, reference 
frame adjustment, spatial template alignment, and computing goodness 
of fit. The execution of representations and processes in an ordered 
combination is called 'programs or routines.' Gordon & Sadler devote 
the second part of their paper to evidencing their theory through 
experiments. 

14. 'The Language-to-Object Perception Interface: Evidence from 
Neuropsychology' 

 Shallice believes that to Jackendoff's model of linguistic 
representation (namely, phonological, syntactic, and 
semantic/conceptual structures) should be added the level of visual 
structures, which fact battles for the separability of 'visual 
semantics' and 'verbal semantics.' Drawing on some pathologies, 
Shallice argues through category specificity and sensory quality and 
functional aspects of different categories that the semantic system 
is not as unitary as many might think. Shallice reports experiments 
where patients were less able to perform equally visually with 
regard to identifying animals, plants, and foods. His argument is 
that patients score lower on animal categories because of the less 
frequent availability of animals in their environment as against 
man-made objects and foods. Patients, however, tend to perform 
better with functional aspects rather than sensory aspects of 
biological categories. Evidence from lesioning of the 'visual' 
semantic units points to the fact that 'the functional 
representations of the living things were less adequately retained 
than those of artifacts.' However, the living/nonliving distinction 
is graded and not absolute since the pattern might be found reversed 
in some patients. Evidence from category-specific pathology 
indicates that 'knowledge of the characteristics of objects is based 
on representations in more than one type of system.' The second 
difficulty with Jackendoff's conceptual/structural description as 
unitary comes from optic aphasics, who show a dislocation between 
'visual semantics' and 'verbal semantics.' Optic aphasics have been 
demonstrated to perform less accurately in naming from drawings than 
from tactile and auditory stimuli. 

15. 'Space and Language' 

 Peterson et al are interested in how space is structured by humans. 
Their main argument revolves around the extent to which development 
and culture shape the language-space interaction and alter spatial 
cognition. Drawing on Jackendoff's model of space representation 
(SR) to substantiate claims about the language-space interface, 
Peterson et al mention evidence from studies of the brain and of 
perception. Evidence from the former points to the fact (i) the 
existence of spatial maps in the brain makes it unlikely that a 
single amodal SR determines the set of spatial primitives (ii) not 
all neural representations of space include detailed representations 
of objects and (iii) there is a partial separation between what an 
object is and its location. Evidence from the latter, however, 
points to the fact that (i) there exist independent modules for SR 
and (ii) modules specialize in objects, spaces or the interaction 
between both. Reacting to Jackendoff's prepositions and nouns 
correlating with different sorts of SRs in language, Peterson et al, 
following Bloom (1994), argue that nouns do not map onto a 'what 
system that encodes objects in terms of shape,' but rather 'nouns 
map onto CRs that are nonspatial (and thus can include notions like 
joke and day).' Peterson et al argue that different languages and/or 
cultures do have an effect on how SRs are coped with in cognition, 
but insist that this is not espousing the Whorfian hypothesis 
according to which language determines our manner of thinking about 
space. In the same line of thought, they maintain that experience 
does not alter perceptual and cognitive processes. They end their 
paper by pushing for research in spatial language and its relation 
with motion, recognition of spatial relationships, and inferential 
system about objects in space. 

CRITICAL EVALUATION 

Bierwisch's proposal is too formal to be used to account for how SR 
is grammaticalized. Levelt's model of SR, however, is more in line 
with the attentional cognitive school (Ungerer & Schmid, 1996), 
where events are conceptualized as focused and selective. Levelt is 
making this attentional perspective as PT, with a systematic 
characterization of spatial representation as absolute, intrinsic, 
and deictic. Talmy's motion in space or 'fictive motion' involves 
the Lakoffian linguistic metaphor, which is judged in terms of truth 
and falsity by assimilating the TD to factivity and the SD to the 
mapping of fictivity on the TD. As is well-known, this conception of 
metaphor suffers from much the same drawbacks as those of Grice's 
(1975) truth vs. falsity account and Searle's (1979) word meaning 
vs. speaker meaning conception. Mandler's paper echoes Talmy's 'how 
space structures language.' Tversky's pragmatic criteria for 
choosing one SP, which are 'cognitive difficulty' and 'audience,' 
may be collapsed as 'cognitive difficulty' because when speakers 
try to adapt their descriptions to their addressee's perspective, 
they are already attending to the question of audience, which is 
presupposed under 'cognitive difficulty.' 

This collection of papers on SR and language is important in more 
than one respect: (i) across modalities, it makes it clearer that 
frames of reference are not modality-specific, (ii) across spoken 
and signed languages, the recognition of SR is effected through the 
sensorimotor system specialized for each type, and (iii) research in 
spoken and signed languages strongly suggests that SR, although 
language-specific, is not language-bound as evidenced by SLs. The 
experiments reported by Emmorey to the effect that SLs use space and 
motion to encode space via the right hemisphere point to the fact 
that SR is a cognitive operation rather than a linguistic one. 
Talmy's 'ception' program is a shortcut to perception and 
conception, where the distinction between language and other 
cognitive systems almost disappears. Optic aphasics' difficulty with 
keeping 'visual semantics' and 'verbal semantics' together is 
another difficulty for Jackendoff's CS-SR interface model. In her 
December 1999 review of the same book, Anne Reboul wrote that the 
book could have been entitled something like: All You Need to Know 
about Space. Although the book offers an impressive variety, yet it 
may be worth noting that not a single paper in this volume devoted 
any space to SR and mental spaces a la Fauconnier (1994) and 
Fauconnier & Sweetser (1996), which I think have their word to say 
on space, language, cognition, and culture. 

BIBLIOGRAPHY
 Fauconnier, Gilles (1994). Mental Spaces: Aspects of Meaning 
Construction in Natural Language. Cambridge: CUP. 

Fauconnier, Gilles & Eve Sweetser (1996) (eds.). Spaces, Worlds and 
Grammar. Chicago/London: The University of Chicago Press. 

Grice, H. P. (1975). 'Logic and Conversation.' In: P. Cole & J. L. 
Morgan (eds.), Syntax and Semantics (vol. 3) Speech Acts. New 
York/London: Academic Press, 41-58. 

Landau, B. & D. Stecker (1990). 'Objects and Places: Syntactic and 
Geometric Representations in Early Lexical Learning.' Cognitive 
Development 5, 287-312. 

Searle, John R. (1979). 'Metaphor.' In: A. Ortony (ed.), Metaphor 
and Thought. London/New York: CUP, 92-123. 

 (1983). Intentionality. An Essay in the Philosophy of Mind. 
London/New York: CUP. Ungerer, Friedrich & Hans-Jorg Schmid (1996). 
An Introduction to Cognitive Linguistics. London/New York: Longman. 

REVIEWER 
Zouhair Maalej, Assistant professor of Linguistics, University of 
Tunis I. Research interests include: metaphor, cognitive 
linguistics, pragmatics, cognition-pragmatics interface, cognition-
culture interface, (cognitive) stylistics, (critical) discourse 
analysis, functional linguistics, translation studies, etc. 
Publications include voice, perception, and metaphor. 

Dr Zouhair Maalej Department of English, Chair Faculty of Letters 
and Human Sciences, Tunis-Manouba, 2010, Tunis, Tunisia. 

Office phone: (+-216) 1 600 700 Ext. 136 Office Fax: (+-216) 1 520 
910 Home Telefax: (+-216) 1 362 871 E-mail: zmaalej+AEA-gnet.tn URL: 
http//:simsim.rug.ac.be/ZMaalej 

 
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue