Introduction: Space, language, and cognition

Maya Hickmann, Stéphane Robert

To cite this version:

Maya Hickmann, Stéphane Robert. Introduction: Space, language, and cognition. Space in Languages
Linguistic Systems and Cognitive Categories, 66, John Benjamins Publishing Company, pp.1 - 15,
2006, Typological Studies in Language, 9789027229786. �10.1075/tsl.66.01hic�. �halshs-04368743�

HAL Id: halshs-04368743

https://shs.hal.science/halshs-04368743v1
Submitted on 1 Jan 2024

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
In: M. Hickmann, S. Robert (Eds.) Space in languages : Linguistic systems and
cognitive categories. 2006. Typological Studies in Language 66. Amsterdam : J.
Benjamins, 1-15. doi.org/10.1075/tsl.66.01hic

Space, language, and cognition: Some new challenges

Maya Hickmann
CNRS, Université Paris 5

Stéphane Robert
CNRS-LLACAN, INALCO, Université Paris 7

1. Why space?
As illustrated by the Kantian tradition and by a number of more recent
theories, space is often viewed as a universal cognitive primitive, an “a
priori form of intuition”, that conditions all of our experience. From this
point of view, it is of particular interest to study the linguistic expression of
space, since languages seem to capture and to make explicit the constraints
of experience on the construction of spatial reference. At the same time,
language confers to spatial representations the property of referential
“detachability”, that distinguishes these representations from those that are
produced by our perceptual experience of space. This fundamental property
allows speakers to dissociate and to choose among different components of
spatial reference, as well as to use spatial morphemes to express other
and/or more abstract meanings, such as temporal, causal, or argumentative
relations.
Some questions then arise concerning the primitive nature of the category of
space in language. To what extent does space, as it is linguistically encoded,
reflect forms of perceptual experience and which aspects of this experience
do different languages encode? Does space constitute a pure and primitive
category from which other linguistic meanings are then derived? These
questions are recurrently addressed by cognitive grammars. They are
particularly central to metaphor theory, as well as highly relevant in the light
of numerous derivations that have been observed in the history of
languages, often indicating that a given term evolves from a “concrete”
spatial meaning to an “abstract” discourse meaning. What are the cognitive
mechanisms that allow such transitions?
In contrast, some recent linguistic analyses argue that spatial values are
neither basic nor even purely spatial, but rather that spatial terms
intrinsically carry many other values, for example meanings related to the
functional properties of objects, their force or resistance, or the goals
towards which speakers construct spatial relations in their utterances.
According to this conception, space in language is therefore not a primitive
INTRO 2

category, but already the result of some construction that is based on our
experience and that is part of the actions we carry out in interaction with the
world. What evidence can be brought to bear on these different
conceptions?
Finally, in the last twenty years, many studies in linguistics,
psycholinguistics, and cultural anthropology have revealed wide variations
in spatial systems across languages and cultures. These variations concern,
for example, the nature of the linguistic devices expressing spatial
information (e.g., verb roots, affixes, classifiers, particles), the particular
distinctions they encode and highlight the most (e.g., manner of motion,
location, posture, shape and other dimensions of spatial entities), and the
reference systems that are used by speakers (absolute, egocentric, relative).
In addition, various studies show that linguistic and cultural systems
determine - at least partially - the nature and cognitive accessibility of the
information that is selected by speakers. This evidence has cast some doubts
on the supposedly universal properties of spatial categories. It therefore
raises questions concerning the impact of linguistic categorization on
perception, as well as the existence of a single (amodal) spatial system or of
two distinct (linguistic, perceptual/motor) systems of spatial representations.

2. General overview of book contents

The study of space is framed in this volume within an interdisciplinary
perspective in which varied contributions stem from different scientific
traditions, each bringing its own concerns and methodologies: descriptive,
typological and diachronic linguistics, cognitive anthropology, the
philosophy of language, cognitive psychology, developmental psychology,
psycholinguistics, neurosciences and the study of various pathologies. The
book is organized into three main parts, each focusing on a major set of
questions: Universals, variability, and change (Part I); The nature and uses
of space in experience and in discourse (Part II); Space, language, and
cognition (Part III). As shown below, these issues are interrelated
throughout the volume.
Part I proposes typological and diachronic analyses of spatial systems.
Particular attention is placed on universal and variable aspects of these
systems, as well as on historical changes and factors determining these
changes over time. The chapters in this part examine major typological
differences across linguistic systems (Creissels, Grinevald, Slobin) and
propose diachronic analyses of how some of these systems have evolved
through the emergence, reorganization, or disappearance of categories or
through some more general structural changes (Kopecka, Marchello-Nizia,
Peyraube). Some of these papers also address questions that are at the center
of subsequent parts. For example, analyses of deixis in language
(Marchello-Nizia) directly touch on issues related to the pragmatics of
discourse (Part II), while discussions of the notion of “salience” across
spatial systems directly touch on issues related to spatial cognition (Part III).
INTRO 3

Part II concerns the nature and uses of spatial language in discourse and in
relation to our experience of space. The papers in this section examine how
spatial reference is constructed from experience at the discourse level and
show how this construction process might bear on the very nature of space
in language. They discuss how semantic information is distributed across
clauses, how linguistic categories interact, and how informational
components may be explicit vs. implicit and inferred from context. Some
papers also address questions concerning cognition (Part III) by asking
whether “spatial” values in language inherently require taking into account
other values (Vandeloise, Robert, Cadiot et al.) or by comparing spoken and
signed languages (Sallandre, Talmy) along some of the typological issues
discussed at length elsewhere in the volume (Part I).
Part III touches on fundamental issues concerning the relation between
spatial language and spatial cognition. It examines the impact of linguistic
variation on how spatial information is expressed, perceived, and
categorized by adults and children, as well as how spatial representations
may break down in pathology. Some question the assumption that linguistic
variation affects speakers’ perception (Dokic and Pacherie). Others examine
how pathology might inform us about the existence of one vs. several
representational systems for space within the same speaker (Landau and
Lakusta; Denis, Ricalens, Baudouin, and Nespoulous). Yet others
(Hickmann) examine how children acquire spatial systems across languages
that present typological differences (discussed in Part I), suggesting the
potential role of language in structuring spatial cognition. As shown below,
some of these papers also touch on issues concerning Part II, by examining
the organization of spatial information in discourse (Hickmann, Denis et
al.).
The more detailed overview of the volume below shows how each chapter
centrally addresses one or the other set of questions at the heart of each part
in the volume. It also indicates how different contributions substantially
address other questions, pointing to interrelations among questions and to
recurrent links across the entire volume.

3. Part I: Universals, variability, and change

The first set of questions in this volume concerns the variability of spatial
systems across languages: What components of space do linguistic systems
encode and how? What is the extent of linguistic variation and what are its
limits? How do spatial systems evolve over time, sometimes shifting from
one type to another, and what are the causes of these changes? How can the
study of sign language bear on these questions from a typological point of
view?
Linguistic typology studies the types and limits of linguistic diversity. For
example, comparing what could be logically expressed with what languages
actually express allows us to classify languages according to the types of
distinctions they make. Spatial systems include different means of
INTRO 4

expressing location (Creissels, Grinevald, Vandeloise) and motion (Slobin,

Kopecka, Peyraube; also see Hickmann in Part III), as well as different
frames of reference used to locate entities in space (see Marchello-Nizia;
also see Robert for linguistic insights in Part II and Dokic and Pacherie for
epistemological questions in Part III).
Because languages use a limited number of means to express meanings,
they differ in the distinctions they systematically express. As shown by
Grinevald in relation to location (responses to the question “Where is X ?”),
languages vary first in the nature of the spatial information they encode.
Each language selects some information components towards which it
directs speakers’ attention, leaving other components more or less implicit
and to be inferred. In this respect, languages differ widely in their degree of
semantic “granularity”. Thus, languages may vary in the number of spatial
prepositions they provide. Some even provide a unique semantically
vacuous preposition, but express locations indirectly through constructions
that indicate the position of entities (‘the pot is [standing vertical] by the
fire’). Again, languages may distinguish only a few positions (‘lying’,
‘standing’, ‘sitting’, ‘hanging’) or a great number of positions (up to fifty
positions, for example, ‘sitting on bottom’, ‘sitting on one’s haunches’,
‘sitting huddled’….).
Languages also differ in the density of the information they convey through
the phenomenon of “conflation”. Different types of information may be
expressed in a unique form, for example posture can conflate with
verticality, dimensionality (whether the object is 1D, 2D, or 3D), texture,
permanence, animacy, number (e.g., “a horizontal flat object is somewhere
on its back in relation to another entity”, “something is crouching
immobile”). Some distinctions may not be expressed at all. For example,
Creissels shows that some languages use distinct morphemes to express
location (‘to be at a place’), the source of a motion (‘to come from a place’),
and direction (‘to go to a place’), but two or three distinctions may conflate
into a unique morpheme, without further detail. Furthermore, spatial
information may be distributed across various devices and subtle
combinations thereof (verbs, prepositions, postpositions, affixes, particles,
nominal classifiers…) (see papers by Grinevald, by Kopecka, and by
Creissels).
However, as pointed out by Talmy, whereas the spatial lexicon can be quite
rich (particularly because of conflation phenomena), grammatical forms
relevant to space come in a relatively closed set of categories. Speakers
must therefore select among these pre-packaged schemata when depicting
spatial scenes. Furthermore, the universal inventory of fundamental spatial
elements that combine to form whole schemata is relatively limited.
Expressing a spatial scene requires a process of “schematization”, that is the
selection of some characteristics, that relies on some among a relatively
limited set of elements in each relevant category.
INTRO 5

For example, the category of “number” pertains to individual components of

spatial scenes. In closed-class items (i.e. classes with a closed inventory
such as grammatical forms), this category may only include four members
in relation to space (one, two, several, many): the ground may consist of just
one object (near), of two objects (between), of several objects (among), and
of numerous objects (amidst). According to Talmy, this property is a special
characteristics of spoken language as compared to other cognitive systems.
Furthermore, Talmy and Vandeloise both note that classical geometric tools
do not accurately account for the distribution of linguistic spatial
components such as prepositions. In this respect, it is worth noticing that
language is neutral with respect to particular dimensions of Euclidean
geometry. This neutrality makes languages flexible and allows them to
make maximal use of a limited number of components. For example, with
respect to the dimension of “magnitude”, the preposition across can apply to
a situation of any size and the preposition near can describe the distance
between planets in the solar system or between two houses within a
relatively small region.
Thus, languages vary noticeably in the spatial distinctions they explicitly
make, but they also vary in other respects. Interestingly, comparing the
types of distinctions that are found across languages to the set of all logical
possibilities shows three points. First, all types seem to exist most of the
time, but a few patterns are predominant and some are very rare. Second,
existing types often correspond to a common linguistic area or linguistic
family, but this rule is by no means absolute. Different types of languages
can coexist in the same linguistic area. Third, different patterns may be
found within one language, so that it might be best to talk of “strategies”
used by languages rather than of language types.
The same conclusions hold for location and for motion. As Peyraube
reminds us, the expression of a basic motion event in natural languages
involves several semantic components: a figure (or target), that is the entity
in motion and/or to be located; a ground (or landmark), that is the entity in
relation to which the figure is located; the path of motion; the manner in
which motion is carried out; and the cause of motion. Three of these
components are central across languages: manner (e.g., English to run, to
walk, to fly), path (to run in, out, up/down, across), and ground (to run into
the room, to run into the garden). Languages differ in how they encode path
and manner, but also in the attention they pay to manner. In his pioneer
work, Talmy (1985, 1991, 2000) suggests that languages can be divided into
two groups in terms of the ways in which they encode the core feature of a
motion event, namely its path. Verb-framed languages (such as Romance or
Semitic languages) typically convey path information by lexicalizing it in
the main verb (e.g., French entrer, sortir, monter, traverser). In contrast,
satellite-framed languages (such as Germanic and Slavic languages) encode
path in satellites, such as particles, prefixes, or prepositions associated to the
main verb (English to walk into, to climb up, to run across). The use of
INTRO 6

satellites to encode path allows the main verb of the clause in S-languages to
be available to encode other dimensions of motion events, for instance
manner (to walk into, to climb up, to run across).
As pointed out by Slobin, languages differ considerably in their lexical and
morphological means of expressing manner, thereby attributing different
degrees of salience to this dimension. For example, various common
manner verbs in English (to walk, creep, trample, stomp, stumble… on the
plants) can hardly be translated into French. Manner is expressed with more
limited means in V-languages, frequently in subordinate manner verbs, that
are merely optional (entrer en courant, en rampant…), and it is most
frequently not expressed at all. Slobin proposes a third language type,
namely “equipollently-framed” languages, in which path and manner are
expressed by equivalent grammatical forms (also see Slobin 2003, but see a
critique by Peyraube in relation to Chinese). These forms may be serial-
verbs, bipartite verbs (a complex of two verbs, one expressing manner, the
other path) or generic verbs combined with coverbs encoding path and
manner. As we will see (Part II), the nature of these morphological means
has important consequences for the degree to which manner is salient in
discourse.
Now let us turn to frames of reference. Levinson’s major work has shown
the existence of different frames of reference across languages (e.g.,
Levinson 2003). As summarized in this volume (see Robert in Part II and
Dokic and Pacherie in Part III), three kinds of frames of reference can serve
to locate entities: (1) an intrinsic frame of reference, in which coordinates
are determined by the inherent features of the ground object (He’s in front of
the house: the house has an intrinsic orientation defining its front part); (2) a
relative or anthropocentric frame of reference, where the coordinate system
is based on an external viewer or point of view (He’s to the left of the house:
the left of the house is defined relative to the speaker’s position); (3) an
absolute frame of reference using fixed bearings such as cardinal points
(He’s north of the house). When the point of view used as the frame of
reference is the speaker, the relative frame of reference is also called
“egocentric” or more commonly “deictic”. Several authors in this volume
also point out the crucial role of the speaker’s deictic space for language,
that is the space in which the speaker is taken as reference point (Marchello-
Nizia in relation to French, Robert in relation to an African language, and
Vandeloise more generally).
Three chapters add a diachronic perspective to the description of spatial
linguistic systems, providing interesting examples of how systems evolve
through time. These papers illustrate a semantic change in the values of
French demonstratives (Marchello-Nizia), as well as structural changes in
the expression of motion events in French (Kopecka) and in Chinese
(Peyraube). In all three cases, changes were not abrupt, but unfolded in
several stages that took place over centuries. In addition, all three cases
illustrate the existence of some “hybridization” within given languages at
INTRO 7

given points in time and show that this observed language-internal

variability corresponds to the more general variability that can be observed
across languages. That is, during the course of its history, a given language
evolves from one type of system into a different type that is found in other
languages.
For example, spatial systems may undergo structural changes that reflect
typological shifts. With respect to the expression of motion events,
Peyraube shows that Chinese evolved some ten centuries ago from a verb-
framed language encoding path information in the main verb of the clause to
a satellite-framed language encoding path in satellites, namely in directional
complements. Inversely, Kopecka shows that French evolved since about
the 14th century onwards from a satellite-frame language encoding path in
verbal prefixes to a predominantly verb-framed language where path is
lexicalized in the verb. Nonetheless, French has retained a secondary
satellite-framed system, which is a less productive remnant of its previous
state (e.g., verbs such as écrémer ‘to take off cream’, empoter ‘to pot’,
emprisonner ‘to imprison’, atterrir ‘to land on the earth’).
French demonstratives evolved from a personal value (in Latin) to a spatial
semantic value (in Modern French) through a stage during which they
referred to the speaker’s sphere (anything that is linked to the speaker,
whether spatial or not). This change went smoothly through several stages
before reaching its present state. These stages can be summarized as
follows. First, an ambiguous construction plays a pivotal role that allows the
change to occur. Second, at some point the former and the new
constructions (or values) coexist. Third, the frequency of ambiguous
constructions is probably a crucial factor explaining why they are
reanalyzed into a new type. In the case of French demonstratives, the
subjective-pragmatic meaning played the role of the pivotal construction.
The change from a strictly subjective value (referring to the “speaker’s
sphere”) to a more abstract value (any element or object activated in the
speaker’s mind or standing out because it is present in the speech situation)
was facilitated by the frequent use of elements being altogether part of the
speaker’s sphere (in a very broad sense) and present at the time and moment
of speech. The frequent combination of these two features in discourse
allowed the semantic reanalysis.
Interestingly, the spatial value of French demonstratives (which is very
common cross-linguistically) is not primary, but rather appears to be the
result of a long evolution that took place during more than twelve centuries.
Marchello-Nizia argues that the evolution from Latin to Modern French, far
from starting with a spatial meaning and gradually moving further away
from it, seems to have gone through a “cyclic” change, as do some other
morphemes: it moved from spatial to personal meanings (Latin), then to
subjective-pragmatic meanings (Old French), before returning to spatial
meanings (Modern French).
INTRO 8

Finally, Talmy compares how spoken and signed languages represent space.
Signed languages are of particular interest because, in comparison to spoken
languages that are linear, they are spatialized and multidimensional systems,
that use a gestural subsystem of face, head, torso representations, a gradient
subsystem of “bodily dynamics”, and an associated somatic subsystem
including facial expressions. However, according to Talmy, spoken and
signed languages share the property of containing two subsystems, one
“open-class” or lexical subsystem (typically the roots of nouns, verbs, and
adjectives) and one “closed-class” or grammatical subsystem, consisting of
relatively few forms that are difficult to augment. These two subsystems
basically perform two different functions when they combine in the
sentence: open-class forms largely contribute conceptual content, while
closed-class forms determine the conceptual structure of the scene to be
construed by language. Spoken and signed languages can therefore be
considered as two language modalities.
As shown by Talmy, a crucial property that is specific to how signed
language represents space appears to be the structural characteristics of
scene parsing in visual perception. Thus, in comparison to spoken language,
signed language can mark finer spatial distinctions with its larger inventory
of structural elements, of categories, and of elements per category. It can
represent many more of these distinctions in any particular expression. It
also represents these distinctions independently in the expression, not
bundled together into pre-packaged schemata. In addition, its spatial
representations are largely iconic with visible spatial characteristics; with
respect to this last point, Sallandre also shows the central role of highly
iconic structures in discourse. She further demonstrates that signers may use
a variety of different handshapes (proforms), which are all available in
French Sign Language, to denote a given referent, depending on the
particular properties on which they choose to focus, given their relative
relevance in discourse. As further discussed below, these properties have
consequences for how the brain might organize cognitive functions related
to space in different systems (an issue that is at the center of Part III).
However, because both systems represent spatial situations schematically
and structurally, they nonetheless share properties that are central for
language use at the discourse level (at issue in Part II).

4. Part II: The nature and uses of space in experience and in discourse
This volume addresses a second set of questions concerning space in
language: How are linguistic systems used to construct spatial reference at
the discourse level? Is it possible to determine “spatial” values in language
autonomously, without taking into account other values that are relevant to
our experience of space and that are necessary to characterize our use of
language in discourse or, more generally, any action we might perform.
As shown above, spatial information is distributed in different parts of the
sentence and this distribution varies across languages within a certain range
INTRO 9

of possibilities. Grinevald amply illustrates how spatial information can be

distributed across different sentence components. She shows that the
distribution of spatial information can be overt or covert, redundant or
underspecified, and even entirely absent in extreme cases where the location
of an entity must be inferred from its posture or shape. Indeed, discourse
analysis reveals that the semantics of space, as well as meaning in general,
is compositional and distributed, because language inherently involves
linearization and sequencing. In contrast to vision, which is a holistic and
multidimensional process, verbalization imposes the need to break down
information into discreet and successive pre-constructed units.
However, as a counterpart to this constraint, another property of language
gives it a special kind of power, because different sentence elements interact
with each other, thereby creating new meanings. As shown by Vandeloise,
particular ways of combining spatial prepositions with different verbs and
constructions may confer new meanings to the sentence and to the units
within sentences themselves. For example, French contre (‘against’) cannot
be used with intransitive motion verbs (*L’enfant va contre le mur ‘*the
child goes against the wall’), because voluntary motion verbs describe the
will of the mover, who is assumed not to move deliberately into an obstacle,
unless s/he is mad (Le forcené court contre le mur ‘The madman runs up
against the wall’). The sentence construction itself contributes to spatial
meaning. Transitive motion verbs are used when there is a dynamic
exchange between the agent and the patient (John breaks the wood), while
intransitive motion verbs are used when there is no such dynamic exchange
between the subject and the complement (The bird is above the tree).
It is worth noticing that despite some important differences across systems,
spoken and signed language share common properties at the discourse level.
As pointed out by Talmy, both have basic elements that combine in order to
structurally schematize scenes. Both group their basic elements within some
categories that themselves represent particular categories of spatial
structure. Both follow some conditions on the combination of basic
elements and categories into a full structural schematization. Both also
follow conditions on the co-occurrence and sequencing of such schemata
within a larger spatial expression. Both allow speakers to amplify some
semantic elements or parts of a schema by means of open-class lexical
forms outside the schema. And in both subsystems a spatial situation can
often be conceptualized in more than one way, so that it is amenable to
alternative schemata.
Talmy’s analysis, then, shows the extendability of linguistic prototypes and
the existence of processes that deform schemata. In line with this insight,
Vandeloise’s analysis of spatial prepositions in discourse reveals first that
their values vary according to the terms with which they are used. It also
points out that the semantics of spatial terms involve notions that are related
to our experience, rather than to a conception of space in terms of Cartesian
axes. Such notions include, for example, the transmission of energy and
INTRO 10

forces, the cause of motion, control, intentionality, will, and even the
agent’s satisfaction. More generally, some important asymmetries found in
the uses of prepositions (The bird is in front of the house, but *The house is
behind the bird) result from the fact that spatial prepositions are not devoted
to a purposeless description of space, but rather serve as instructions in
order to help locate a specific target. In order to guide the addressee, the
speaker uses the most conspicuous landmark possible and a bird is not a
good landmark to locate a house.
We saw above that the semantics of spatial terms often combine spatial
values with other components because of the phenomenon of conflation.
Another characteristic of language is that spatial terms always have non-
spatial uses. This property is not specific to spatial language, but results
from the more general polysemous nature of linguistic units. This point is
alluded to by Vandeloise’s provocative title “Are there spatial
prepositions?”. His final answer to this question is positive, but as long as
one conceives of space in language as a component of human concrete
external experience, rather than as a geometric tool. According to this
conception, spatial uses of prepositions play a central role in providing the
necessary “impulsion” that determines how these devices are distributed
overall in languages.
Non-spatial uses of spatial markers are also discussed by Robert in the
particular case of deictic space (also see by Marchello-Nizia in Part I). If
deictic elements are used to refer to the space of the speaker, they always
have at least an extended use to refer to the space of discourse, particularly
to designate a term that is close or far away in previous speech. This special
discursive use of spatial terms illustrates another property of language,
namely its reflexiveness, that is the property whereby language can be used
to “talk about” language. In the case of Wolof studied by Robert, the use of
deixis goes far beyond the spatial location of an entity, pervading the entire
language (noun determination, predication, subordination) and playing a
special role in the construction of various relationships of syntactic
dependency. Through a special suffix indicating the absence of localization
in the speech situation, Wolof also provides a striking example of how
“deixis in absentia” plays a central role for linguistic construals.
Finally, as demonstrated by several papers, discourse analysis reveals
another important point concerning linguistic variation. Although different
ways of expressing space may coexist in a given language system, some
may be scarcely used in discourse, while others, on the contrary, may be
obligatory and even overexploited. This variation results from the fact that
languages choose particular strategies about which elements they consider
to be most salient for the description of situations. These choices can be
purely conventionalized or induced by the morphosyntactic constraints of
each system.
Grinevald (in Part I) illustrates this point with two groups of Amerindian
languages that make extensive use of the same morphological devices, but
INTRO 11

that do so in totally different ways. Tzeltalan languages make pervasive use

of positional roots in locative predicates, but also in a very productive
derivational system (such as numeral classifiers and verbs, intransitive and
transitive constructions). Such frequent positional roots therefore
systematically direct attention to spatial and other physical characteristics of
entities, such as the figure of a basic locative construction. In Jakaltek-
Popti’ directionals are also massively used. However, because they have
evolved to express an abstract notion of trajectory in space, they can be
used in the absence of any motion on the part of spatial entities, as shown by
their use with verbs of perception or with verbs of saying (‘He saw her [up]
[away]’ or ‘He said hello [up] [towards] to her’). In these cases directionals
serve to perspectivize scenes, indicating the reference point from which the
scene is to be conceived, somewhat like a camera which takes different
points of view.
Slobin’s analysis of an extensive corpus of texts (in Part I) concludes that
lexicalization and morphosyntactic patterns constrain information focus in
discourse. In contrast to speakers of satellite-framed languages, speakers of
verb-framed languages virtually never mention manner, focusing on
emergence, appearance, or changes of state and showing a strong preference
for marking state changes in the verb root. Although V-languages provide
means of expressing manner, speakers seldom do so in spontaneous
discourse, because such constructions unnecessarily foreground manner,
given that their language selects state changes as the main information focus
and provides no compact construction that allows joint attention to state
changes and to manner. Hickmann’s study (in Part III) provides
developmental evidence for this claim, showing that adults and children
frequently express both path and manner in English, but only path in
French. As a result of verb- vs. satellite-framing, speakers also organize
their discourse in very different ways, compactly expressing information
within utterances in English, but distributing it across several utterances in
French, particularly at young ages. Finally, in addition to these strong cross-
linguistic differences, particular discourse factors (such as presuppositions,
contrastive contexts, world knowledge of particular event properties)
influence discourse organization in both languages by inviting speakers to
express or to ignore the manner of motion.
As will be discussed at length below (in relation to Part III), it is at this point
worth noting a general problem that concerns the precise nature of language
effects on cognition and the ways in which such effects can be
demonstrated. For example, when speakers’ internal representations are
inferred on the basis of their productions, it is somewhat circular to argue -
in the absence of any independent evidence - that these behaviours reflect
different modes of cognitive organization. In this respect, some recent views
propose that linguistic mediation should be particularly efficient in
situations that require organizing the flow of information in discourse for
communicative purposes. Depending on the authors, this organizational
INTRO 12

process is said to be universal or to vary (partially or even substantially) as a

function of language-specific properties from the youngest age onwards,
despite developmental progressions that can be observed in all languages. It
is also shown to remain intact or to break down in various pathologies, in
addition to other deficits that may be otherwise observed.
Finally, in sharp contrast with all of these views, Cadiot et al. argue that the
various theoretical frameworks that underlie such analyses are insufficient
to account for the ways in which language intertwines spatial and other
values that arise from our experience during the construction of meaning. In
particular, they argue against the predominant view in cognitive linguistics
that space should be reduced to topological properties and provide the basis
from which all other values should be derived. In order to illustrate the
problems that arise in such a framework, they argue against the validity of
the typological distinction between verb- and satellite-framing by analyzing
a number of French path-verbs (partir ‘to leave’, sortir ‘to exit’, monter ‘to
go up’). This analysis shows the numerous dimensions that are involved in
contextualized verbal uses, such as mecanicity and correct functioning in the
case of marcher ‘to walk’ (cf. 1), or surprise and non-control in the case of
tomber ‘to fall’ (cf. 2), as illustrated in examples such as: (1) Le moteur
marche (‘The engine is running’), Ca marche bien, ton affaire ? (‘Is your
business going well?’), Il nous a fait marcher ! (‘He put us on!’); (2) tomber
dans les pommes (‘to pass out’), tomber amoureux (‘to fall in love’). Such
uses, they argue, cannot be accounted for in the currently available
frameworks of cognitive linguistics, except by postulating secondary
processes of deriving “metaphorical” meanings in various artificial and
counter-productive ways, that do not capture the very nature of language
and of experience in context. In contrast, Cadiot et al. defend a holistic view
of semantics and of perceptive experience. In line with phenomenology and
Gestalt theory, they claim that perception is a dynamic field of experience
involving not only spatial perception but also dimensions pertaining to
action (such as manner, gesture or attitude) and qualitative evaluation (such
as surprise, telicity, intentionality, anticipation). These “praxeologic” and
subjective dimensions of perception are also present in the core semantics of
motion verbs but activated to different degrees as a function of the situation
and discourse context, as is also the case for the spatial value of these terms.

5. Part III: Space, language, and cognition

The systematic cross-linguistic study of space raises central questions
concerning the relation between language and cognition. Although this
fundamental question is not specific to space (see similar issues in other
domains, for example in Gentner and Goldin-Meadow 2003; Gumperz and
Levinson 1996; Lucy 1992; Nuyts and Pederson 1997), it has been
particularly debated in relation to this domain across all disciplines among
the cognitive sciences. As we saw, much research in linguistics and
anthropology has been devoted to uncovering and interpreting universal and
INTRO 13

language-specific properties of spatial language, the implications of which

have led to a renewed attention for the relation between language and
cognition. Recent research in neurosciences has been also concerned with
determining the neural substrate underlying spatial behaviour, leading some
to postulate the existence of two distinct anatomical systems in the brain
(what/where-systems) devoted to cognitive functions centered on the
recognition of objects or on their localization in space (see Jackendoff 1996;
Landau and Jackendoff 1993). Finally, space has always been at the center
of much research in developmental psychology, constituting even one of the
building blocks for cognitive development in Piagetian theory (Piaget and
Inhelder 1947). From this point of view, cognition emerges from early
sensori-motor schemata during ontogenesis (e.g., displacements in space,
object manipulation), following a sequence of general and universal stages.
The special status of space in this respect may reside in the fact that it is one
of the most basic domains of behaviour, which is essential for survival in all
species, but also a domain that displays considerable and striking variations
across spatial systems in human languages. In the face of such linguistic
diversity, a number of major questions must be addressed on the basis of
empirical evidence. As we saw, linguistic representations result from
particular spatial systems that display very different types of internal
organization. However, general perceptual or cognitive processes
contributing to how we construct spatial representations have been
traditionally assumed to be universal and independent of language. How
autonomous or intimately related are our linguistic and non-linguistic
representations of space? Do these two forms of representation necessarily
interact in everyday processing? Does one derive from the other
ontogenetically and, if so, through what developmental processes?
Different answers to these questions have been proposed. In particular, a
major debate opposes two contrasted views, although variants of each
position and intermediate views between these two extremes are also
available. The first position assumes that linguistic and non linguistic
representations are entirely independent from one another and that they do
not interact in any substantial way. In this view, the fact that these two types
of representations may have similar structures would merely result from
functional requirements of spatial behaviour. In sharp contrast, it has also
been proposed that spatial language and other forms of spatial cognition are
intimately related and that they fully interact with one another. A further
variant of this position, which is thoroughly discussed in this volume,
postulates that our linguistic spatial representations partially structure our
non linguistic representations and that they do so in variable ways across
systems.
More generally, developmental research shows the existence of some
universal perceptual and cognitive constraints, as well as the impact of
general and language-specific properties on the development of human
cognition. With respect to cognitive constraints, a growing number of
INTRO 14

results shows that infants display numerous capacities from a few days or
months onwards in a variety of knowledge domains. However, the relation
between this precocious knowledge and subsequent development after the
emergence of language still remains entirely mysterious. Depending on the
theoretical framework that is adopted, the child’s “initial state” at birth
comprises some innate “core” knowledge (Spelke 2003) and/or a strong
propensity to discover perceptual invariants (Mandler 1998), either of which
(or both) might constitute the first universal foundation of cognition. In
either case, it is assumed that the child’s task is to match his/her initial
representations (or at least some of them) with the ones that are provided by
language. In turn, language implies a new representational format that
allows abstraction and/or interconnections among knowledge components
during later development.
A very different approach puts forth the idea that language structures
cognition itself. One version of this approach is best represented by
Vygotsky’s writings during the twentieth century (Vygotsky 1962; also see
revivals in Hickmann 1987; Wertsch 1991), but also by more recent
developmental research addressing related issues in a new light (for
example, Gentner 2003). It proposes that human language provides a
semiotic medium that has a major implications for ontogenetic and
phylogenetic development. Thus, during ontogenesis language invites the
child to construct new forms of cognitive organization, though some of its
general properties, such as its multifunctionality, the propositional and
temporal constraints it imposes on information processing, its meta-semiotic
or self-reflexive capacity, and the potential it offers the child to extract
invariants and to construct concepts through particular forms of reasoning.
A second version, most known through the writings of Whorf (1956) but
recently revived by neo-whorfian approaches (Bowerman and Choi 2003;
Gumperz and Levinson 1996; Lucy 1992; Nuyts and Pederson 1997; Slobin
2003), goes further by postulating that language-specific properties also
result in a partial transformation of our representations, thereby leading to
particular patterns of behaviour during language acquisition (its rhythm, its
course) and in cognitive organization itself. As can be well illustrated in the
domain of space, these new approaches suggest that languages “filter” and
“channel” the flow of information by biasing the salience of some
informational components, bringing the child to focus his/her attention on
some aspects of reality, which then become cognitively more available and
accessible. Language thereby invites the child to construct a particular
system of categories (Bowerman and Choi 2003) and to follow a particular
“mold” when organizing information in discourse (thinking for speaking,
Slobin 1996).
Opponents (e.g., Jackendoff 1996; Landau and Jackendoff 1993; Landau
2003; Clark 2003) reject this approach on several grounds, highlighting
especially the problem of circularity in empirical attempts to demonstrate
the impact of language on cognition merely on the basis of data concerning
INTRO 15

language use. Although some admit that language-specific properties (or at

least some of them, such as obligatory markings) might influence our
language behaviors, they argue against the idea that such properties can
influence our representations beyond language itself and that our variable
language behaviors should reflect variables modes of cognitive
organization. The crucial point of disagreement, then, revolves around the
impact of language on non-linguistic representations. This question
constitutes perhaps the most difficult challenge for future research in
psychology, that will have to address at least two of its facets: To what
degree is the fully developed cognitive system of adults affected by the
properties of their mother tongue? When and how do children’s initial
representations “connect” with their later representations during
ontogenesis?
These different issues are discussed at length in this volume. First, from a
typological point of view, Talmy’s analysis (in Part II) of substantial
differences between spoken and signed languages leads him to challenge
Fodor-Chomsky’s model, according to which there would be a special
language module or “linguistic organ” in our brain. He admits the existence
of a “core” language system, that would be responsible for common
properties of spoken and signed languages. However, he argues that this
system is more limited in scope than the one proposed by Fodor and
Chomsky, and that it further connects with different systems responsible for
properties that are specific to signed vs. spoken languages. With respect to
signed languages, this system would connect with particular parts of the
neural system that are responsible for visual perception, accounting for finer
granularity, iconicity, gradience, and aperture limitations. With respect to
spoken languages, it would connect with neural systems responsible for
particular properties, such as the packaging of spatial elements into a stable
closed set of patterned combinations, as well a system for generalizing or
deforming these packets.
Second, from a philosophical epistemological perspective, Dokic and
Pacherie challenge Levinson’s neo-Whorfian claim that the (egocentric,
absolute, or relative) frame of reference that is dominant in a given language
should infiltrate spatial representations in non-linguistic modalities.
Admitting the fact that languages need not adopt a relative or egocentric
frame of reference, they critically discuss the further assumption that one
and the same frame should coordinate language and perception. Their first
step is to argue against the assumption that frames of reference are
necessary at the most basic level of perception and that this assumption
leads to a logical flaw or infinite regress. According to this analysis,
geometrical properties of objects can be perceptually encoded independently
of intrinsic frames of reference, for example perceiving a bottle in front of a
chair is tantamount to perceiving it near a particular (front) side of the chair.
Similarly, with respect to absolute frames, the perceptual identification of
directions across contexts presupposes their non-absolute identification
INTRO 16

within a given context, for instance using demonstratives (this direction).

Finally, perception need not use relative frames in order to distinguish
directions in the perceptual scene (left/right vs. right/left), since the relevant
distinction can be drawn in each perceptual context by demonstrative means
(from here to there).
Furthermore, relative frames of reference can be implicit in perception. Our
perception of spatial entities exploits our situation relative to them, which
need not be represented as such. Whereas a relative frame of reference in
language implies the explicit representation of relations among referent,
relatum, and point of view, such a frame is necessarily implicit (not
explicitly represented) in perception. Implicit frames can explain
information transfer across modalities without any appeal to additional
assumptions. Thus, it is not necessary to assume that the target state can
only exploit explicit representations in the source state, since relevant
information can be implicitly nested in or associated with the source state.
When we perceive a bottle next to a chair, we need not perceptually
represent the bottle and the chair as bearing different spatial relations to
parts of my body. Spatial information can be carried over, even though it is
not perceptually represented, thereby producing perspective-bound
representations, despite the fact that explicit representations may be
incompatible. Dokic and Pacherie conclude that frames of reference best
characterize high levels of cognitive processing, whereas perception may be
perspective-free at the most basic level.
Third, two papers (Landau and Lakusta, Denis et al.) partially support the
same view, arguing for the relative autonomy of language-based and other
behaviours, on the basis of empirical evidence from pathologies (Williams
syndrome, Altzheimer) showing dissociations – and therefore some
independence - between linguistic and cognitive representations. Thus,
patients with Williams syndrome (WS) have been classically described as
showing a dissociation between linguistic and non-linguistic spatial
knowledge, since they seem to suffer from severe non-linguistic spatial
impairments, while displaying relatively spared language. Landau and
Lakusta ask to what degree and in what way spatial language depends on
non-linguistic spatial representation or can emerge autonomously. More
specifically, they examine whether spatial language is necessarily affected
when one or more aspects of non-linguistic spatial representation is
impaired.
They first note that these questions have not been convincingly resolved in
the available literature, since the only available evidence for a deficit in
spatial language among WS speakers is far from conclusive. Although this
evidence seems to point to a common general deficit in language use and in
non-verbal behaviours, which would support the hypothesis of a relation
between linguistic and non-linguistic spatial knowledge, it does not allow us
to distinguish among many different mechanisms that may cause apparent
linguistic deficits. They then present further evidence from two studies
INTRO 17

concerning how WS speakers represent motion events and static location.

First, when WS subjects describe voluntary motion that is carried out in
different manners (e.g., fly, slide, jump) and over different paths (FROM,
TO, VIA), their responses show that the structure of their spatial language is
preserved overall and that they only differ from matched controls in the
frequency with which they express source information (“Source
Vulnerability”). Source vulnerability, however, seems to be a general
characteristic of cognitive architecture, also applying to normal populations
(adult and children), who prefer to mention goals over sources.
Second, static location was examined in three tasks. A non-linguistic task
required subjects to judge whether an array was the same as a target array
after small changes in the original location of an entity (on or off the
extension of a square’s axis for the relations above, below, left, and right).
In a language production task they had to complete sentences of the form
“The circle is [where?] to the square” in order specify the location of a
target entity by (which also fell on or off the extension of the square’s axis
for the same relations). A language comprehension task examined their
responses to requests of the form ‘Put a dot… to the square’ (in which 14
different spatial terms were inserted). The non-linguistic task shows that WS
and control subjects both perform better with on-axis than with off-axis
locations, suggesting that they can impose mental axes on the reference
entity. In both linguistic tasks WS and control children respect cardinal axes
(above/below or over/under and next to or beside) and neither population
correctly produced or comprehended horizontal directional terms
(right/left). However, errors in direction on the vertical axis were more
frequent among WS children than among controls (misusing/misinterpreting
vertical positives for vertical negatives, e.g., above, over instead of below,
under). This directional fragility in the WS population seems to reflect other
deficits in their non-linguistic representations.
Landau and Lakusta’s conclusion highlights the fact that our view about the
relation between language and cognition largely depends on “where we
look”. They propose a complex two-fold answer to their initial questions.
Spatial language emerges with normal structure, despite the presence of
other impairments, if we look at performance in tasks where language can
only encode the spatial world in a coarse manner. However, linguistic
impairments echo non-linguistic deficits if we look at tasks where spatial
language encodes the spatial world in a fine-grained manner.
Similarly, Denis et al. examine spatial discourse in Alzheimer patients, who
are known to present impairments in their ability to navigate in space. Their
main aim is to determine the nature of these patients’ difficulties, and more
specifically to disentangle different cognitive components in their ability to
generate spatial discourse. In order to do so, they compare the verbal
performance of patients and of control subjects across three complementary
studies involving spatial tasks that placed different demands on them. In the
first study, the participants were asked to provide oral route directions in a
INTRO 18

familiar urban environment. In the second they were invited to describe

familiar environments from memory, but without any request to transform
their knowledge into navigational instructions, while the third study allowed
them to rely on a map when generating their spatial discourse.
The patients' route directions in the first study contained far less relevant
spatial information than that of control participants, providing in particular
virtually no reorienting instructions locations that could guide a moving
person's displacements. Although the second study did not require
navigational instructions, their descriptions also provided less information
than those of the controls. Furthermore, they also produced substantially
more modalizing expressions, suggesting that visuo-spatial knowledge was
less accessible to them (e.g., I don't know very well, I can't remember the
name of that street, It is quite difficult to explain). However, their deficit in
generating spatial discourse virtually disappeared in the third study, where
they were allowed to use maps. For example, like controls, they were
sensitive to the relative relevance of actions vs. landmarks in different
segments of their descriptions. Thus, their main difficulty in generating
route directions stems from their difficulty in retrieving spatial information,
rather than from any underlying (purely) linguistic disturbance.
With respect to the discourse issues discussed above (in relation to part II),
note that the patients’ verbal performance in the first two studies seem to
indicate that they may also have had some difficulties in organizing spatial
information in discourse. As noted by the authors, these patients’ spatial
discourse not only contain less relevant referential content, but it is also less
coherent. For example, it frequently consists of a series of successive
statements providing unrelated spots with little information about the
surroundings or about relevant actions and they infrequently position
landmarks relative to each other or relative to the observer.
Finally, some chapters propose quite a different view of the relation
between spatial language and spatial cognition on the basis of cross-
linguistic developmental evidence. These chapters examine the linguistic
and cognitive factors determining how children acquire spatial language
across typologically different systems, suggesting that acquiring different
languages implies acquiring different forms of cognitive organization. As
noted above (in relation to Part I), Slobin proposes that typological
properties of languages determine how speakers talk about space, leading
them to pay more or less attention to the manner of motion in discourse and
to construct particular modes of cognitive organization. Similarly,
Hickmann shows that the typological properties of French and English (as
verb-framed and satellite-framed languages) strongly influence how adults
and three- to five-year-old children talk about voluntary motion
(upwards/downwards motion, crossing events, arrivals, departures). English
speakers express path and manner in compact structures (e.g., to run in, to
run up/down, to run away), while French speakers do so less systematically,
frequently focusing on path alone (e.g., descendre ‘to descend’) or
INTRO 19

distributing path and manner information across utterances (e.g., elle fait du
vélo […] et elle traverse la route ‘she is biking […] and she crosses the
road’), depending on various factors (event properties, discourse context,
lexicalization patterns). These differences result from verb- vs. satellite-
framing, that leads speakers to pay less attention to manner in French than
in English. In particular, they partly result from different lexicalization
patterns across these languages. In the controlled experimental situations
examined, verbs typically encode path alone in French (monter ‘to ascend’,
descendre ‘to descend’, traverser ‘to cross’, arriver ‘to arrive’, partir ‘to
leave’), with the only exception of one common verb that describes path and
manner in relation to upwards motion (grimper ‘to climb up’). And it is
indeed with this event type that the responses of adults and of children most
frequently contain both path and manner.
However, similar developmental progressions can be observed in both
languages. Regardless of language, children tend to encode path alone at the
youngest ages, then to increasingly encode both manner and path with age.
This result reflects the impact of general cognitive factors. Encoding one
information component is obviously simpler than encoding more
information components from a cognitive point of view. Furthermore, when
only one type of information is encoded, it is predominantly the path of
motion, which is more basic than its manner, particularly with changes of
locations that have implications for discourse organization. Nonetheless,
dense manner+path utterances are significantly more frequent at all ages and
with all event types in English than in French. Furthermore, this
developmental progression is most striking in French, where speakers are
increasingly able to use complex subordinate structures in order to jointly
express path and manner (e.g., Il descend/traverse en courant ‘He
descends/crosses by running’). Finally, French children display some gaps
in their verbal lexicon. In particular, with crossing events, adults typically
use the verb traverser (‘to cross’), but this verb does not seem to be part of
the repertoire of young children, who focus on manner alone (Il court ‘He’s
running’), rather than on path alone (Il traverse ‘He crosses’), as might have
been expected on the basis of French typological properties.
Along with other developmental studies (Choi and Bowerman 1991; Slobin
2003, this volume), this research raises questions concerning universal vs.
language-specific determinants of first language acquisition. Children’s
language of motion seems to be language-specific, reflecting typological
properties, despite similar developmental progressions across languages,
that result from general cognitive determinants. From early on children
construct a spatial language that tightly fits the adult system and they then
further tune into this system during language and cognitive development.
Finally, Cadiot et al.’s proposal (in Part II) is strongly opposed to all of
these views. They argue for a holistic conception of both language and
cognition that heavily relies on a number of fundamental principles
borrowed from phenomenology and from Gestalt psychology. They object
INTRO 20

to both a pre-built space that is independent from other dimensions of

perception and the primacy of spatial meaning in the semantics of motion
verbs. Language, in their view, reflects perceptual experience in which
space (like time) is constantly reconstructed by the perspective of an active
subject and consequently must anticipate, accompany, and record these
perpetual adjustments.

6. Concluding remarks
Space has been and remains a rich source of intriguing and challenging
questions for the cognitive sciences. As shown in this volume, it provides
the grounds for debates concerning the existence and implications of
universal and variable aspects of linguistic and cognitive systems. It
furthermore highlights issues concerning the very nature of language and of
cognition, which are perhaps best illustrated by one of the most central
questions in the cognitive sciences, namely the relation between human
linguistic and cognitive processes. Across disciplines various approaches
based on complementary descriptive and experimental methods have
converged or diverged with respect to these issues, reaching conclusions
that have led to extremely different theoretical frameworks.
The contributions in this volume present two sorts of data. They provide
general analyses of space in language, as well as specific empirical evidence
showing the diversity of spatial systems across languages and during their
evolution in history. They also present theoretical discussions and empirical
evidence concerning human spatial behaviors, both verbal and non-verbal,
their evolution in ontogenesis, and their break-down in pathology. On the
one hand, some of this evidence points to the diversity of spatial systems,
raising questions about some previously postulated universals of language.
On the other hand, some analyses in different perspectives ague that, beyond
this diversity, the data show common patterns in how human languages are
organized and evolve over centuries.
These debates have two major types of implications. They first renew old
questions concerning the nature of language, which has been viewed either
in terms of distinct and entirely autonomous levels of organization or as an
integrated semiotic system relating forms, functions, and meanings in
communicative context. Such renewed questions also have implications for
our understanding of cognitive processes, viewed either as entirely
autonomous from linguistic processes or as intimately related to them,
whether in the developed system of the adult or in the developing system of
the child. Evidence from both linguistic and psycholinguistic analyses
indicate that varied modes of cognitive organization are (at least partly)
associated with varied modes of linguistic organization, but also that
linguistic and cognitive organization may be (at least partly) dissociated, for
example in the cases of speakers suffering from various pathologies.
Many empirical questions remain open and require further evidence across
disciplines. Among them, future research will need to be particularly
INTRO 21

attentive to two problems. As shown in this volume, one fundamental point

of disagreement concerns the relation between our linguistic and non-
linguistic representations. In this respect, the revived Whorfian hypothesis
of linguistic relativity that is presently debated across the cognitive sciences
needs to be tested on the basis of evidence that directly relates speakers’
verbal and non-verbal behaviours in such a way as to go beyond two
pitfalls: circular arguments that aim at showing the impact of language
merely on the basis of language behaviours; arguments that aim at
dissociating language and cognition merely on the basis of a small range of
low-level perceptive processes that cannot capture the complex nature of
human behaviour. Related to this point are a number of controversial
questions, both theoretical and methodological, concerning how to define
and capture non-linguistic representations, as well as demonstrate their
linguistic mediation: Are non-linguistic representations pervasive and
implicit to any behaviour? Should they be assessed in situations that are not
immediately mediated by language and, if so, how? If they are linguistically
mediated, what is the nature of the resulting changes? How permanent must
these changes be? How central are attentional processes and do they result
in different categorical systems?
A related point concerns the debated specificity of human language and
cognition in relation to the systems that are found in other species. In this
respect, empirical evidence from phylogenesis is necessary to provide
evidence for or against the significant role of human language in shaping
human-specific cognitive processes during evolution. Few answers are
available and they show again the methodological and theoretical
difficulties facing any attempt to specify the qualitative cognitive changes
that might result from human language (see recent discussions in Gentner
and Goldin-Meadow 2003). Although this volume does not directly tackle
phylogenetic development, the contributions therein provide developmental
(diachronic, ontogenetic) analyses concerning systemic evolution, that point
to some of the directions to be pursued. For example, more evidence must
come from initial phases of child development, which are typically assumed
to be under major biological influences, in order to provide strong support
for the views that cognition is linguistically mediated from very early on or
rather that linguistic mediation is but a secondary phenomenon that
characterizes only subsequent phases, without any major impact on the
child’s pre-existing representations.
These questions require an interdisciplinary approach that can fully spell out
and empirically address the many problems that still remain to be solved.
The joint interdisciplinary enterprise that is illustrated in this volume shows
the invaluable merits of crossing the boundaries that have long prevented
researchers from going beyond the limits of their scientific traditions in
order to construct general theories of human language and cognition. Such
theories must be continuously renewed and revised in the light of
INTRO 22

theoretical, methodological, and empirical advances, that present recurrent

challenges across the cognitive sciences.
References
Bowerman, M. and Choi, S. 2003. Space under construction: Language-
specific categorization in first language acquisition. In Language in
mind: Advances in the study of language and thought, D. Gentner and S.
Goldin-Meadow (eds), 387-427. Cambridge, MA: MIT Press.
Choi, S. and Bowerman, M. 1991. “Learning to express motion events in
English and Korean: The influence of language-specific lexicalization
patterns.” Cognition 41: 83-121.
Clark, E. V. 2003. “Language and representations.” In Language in mind:
Advances in the study of language and thought, D. Gentner and S.
Goldin-Meadow (eds), 17-24. Cambridge, MA: MIT Press.
Gentner, D. and Goldin-Meadow, S. (eds). 2003. Language in mind:
Advances in the study of language and thought, Cambridge, MA: MIT
Press.
Gentner, D. 2003. “Why we’re so smart.” In Language in mind: Advances in
the study of language and thought, D. Gentner and S. Goldin-Meadow
(eds), 195-235. Cambridge, MA: MIT Press.
Gumperz, J.J. and Levinson, S.C. (eds). 1996. Rethinking linguistic
relativity, Cambridge: Cambridge University Press.
Hickmann, M. (ed.) 1987. Social and functional approaches to language and
thought. Orlando, FL: Academic Press.
Jackendoff, R. 1996. “The architecture of the linguistic-spatial interface”. In
Language and space, P. Bloom, M. Peterson, L. Nadel and M. Garrett
(eds), 1-30. London: Cambridge, MA: MIT Press.
Landau, B. and Jackendoff, R. 1993. “What and Where in spatial language and
spatial cognition.” Behavioral and Brain Sciences 16(2): 217-38.
Levinson, S. 2003. Space in language and cognition [Language, culture and
cognition 5]. Cambridge: Cambridge University Press.
Lucy, J. 1992. Language diversity and thought: A reformulation of the
linguistic relativity hypothesis. Cambridge: Cambridge University Press.
Mandler, J.M. 1998. “Representation.” In Handbook of child psychology,
W. Damon, D. Kuhn and R. S. Siegler (eds), vol. 2, 255-308. New
York: Wiley.
Nuyts, J. and Pederson, E. (eds) 1997. Language and conceptualization.
Cambridge: Cambridge University Press.
Piaget, J. and Inhelder, B. 1947. La représentation de l'espace chez l'enfant.
Paris: Presses Universitaires de France.
Shay, E. and Seibert, U. (eds). 2003. Motion, direction and location in
languages, Amsterdam/Philadelphia: John Benjamins.
Slobin, D. I. 1996. “From ‘thought to language’ to ‘thinking for speaking’.”
In Rethinking linguistic relativity, J.J. Gumperz and S.C. Levinson
(eds), 70-96. Cambridge: Cambridge University Press.
INTRO 23

Slobin, D. I. 2003. “The many ways to search for a frog: linguistic typology
and the expression of motion events.” In Relating events in a narrative:
Typological and contextual perspectives, S. Strömqvist and L.
Verhoeven (eds), 219-257. Mahwah, NJ: Lawrence Erlbaum Associates.
Spelke, E. S. 2003. “What makes us smart? Core knowledge and natural
language.” In Language in Mind: Advances in the study of language and
thought, D. Gentner and S. Goldin-Meadow (eds), 277-311. Cambridge,
MA: MIT Press.
Talmy, L. 1985. “Lexicalization patterns: semantic structure in lexical
forms.” Language typology and lexical description: vol. 3. Grammatical
categories and the lexicon, T. Shopen (ed.), 36-149. Cambridge:
Cambridge University Press.
Talmy, L. 1991. “Paths to realization: a typology of event conflation.”
Proceedings of the Berkeley Linguistics Society 17: 480-519.
Talmy, L. 2000. Toward a cognitive semantics: vol. II – Typology and
process in concept structuring. Cambridge, MA: MIT Press.
Vygotsky, L. S. 1962. Thought and language. Cambridge, MA: MIT press.
Wertsch, J. V. 1991. Voices of the mind: a socio-cultural approach to
mediated action. London: Harvester Wheatsheaf.
Whorf, B. L. 1956. Language, thought and reality. Cambridge, MA: MIT
Press.