Advanced Review

Linguistic relativity
Phillip Wolff∗ and Kevin J. Holmes

The central question in research on linguistic relativity, or the Whorfian

hypothesis, is whether people who speak different languages think differently.
The recent resurgence of research on this question can be attributed, in part,
to new insights about the ways in which language might impact thought. We
identify seven categories of hypotheses about the possible effects of language on
thought across a wide range of domains, including motion, color, spatial relations,
number, and false belief understanding. While we do not find support for the
idea that language determines the basic categories of thought or that it overwrites
preexisting conceptual distinctions, we do find support for the proposal that
language can make some distinctions difficult to avoid, as well as for the proposal
that language can augment certain types of thinking. Further, we highlight recent
evidence suggesting that language may induce a relatively schematic mode of
thinking. Although the literature on linguistic relativity remains contentious, there
is growing support for the view that language has a profound effect on thought.
 2010 John Wiley & Sons, Ltd. WIREs Cogn Sci 2011 2 253–265 DOI: 10.1002/wcs.104

INTRODUCTION rise to several logical paradoxes.6 However, a recent

resurgence of research in this area has uncovered
F olk psychology tells us that human cognition
depends on language, and further, that this
dependency creates differences in thought across
subtle and intriguing interactions between language
and thought, leading to a number of more nuanced
versions of the proposal.
language communities. Although often mistaken,
folk psychology appears to be at least partially
correct in this case. In academic circles, such LINGUISTIC RELATIVITY AND ITS
intuitions are referred to as linguistic relativity, MANIFESTATIONS
the Whorfian hypothesis, or the Sapir–Whorf
It has often been claimed that linguistic relativity
hypothesis. Linguistic relativity comprises three
is a weaker form of linguistic determinism. But
main ideas.1–3 First, it assumes that languages can
the strong–weak distinction oversimplifies the more
differ significantly in the meanings of their words
complicated picture that is emerging in recent research
and syntactic constructions—an assumption that is
on the relationship between language and thought.
strongly supported by linguistic, anthropological, and
Linguistic relativity can now be said to comprise a
psychological studies of word and phrasal meaning ‘family’ of related proposals that do not necessarily
across languages.3–5 Second, the proposal holds that fall along a single strong-to-weak continuum. In this
the semantics of a language can affect the way in which article, we examine the arguments and evidence for
its speakers perceive and conceptualize the world, and several branches of the ‘family tree’ shown in Figure 1.
in the extreme, completely shape thought, a position Our overall conclusion will be that the proposals we
known as linguistic determinism. Finally, given that call language as language-of-thought and linguistic
language can affect thinking, linguistic relativity holds determinism can be rejected on both theoretical and
that speakers of different languages think differently. empirical grounds, but that recent findings support
In the early 1990s, linguistic relativity was all a range of alternative ways in which language might
but given up for dead, especially after it was realized have significant effects on thought, leading to possible
that the proposal, as it was often understood, gave differences in thought across language communities.

∗
Correspondence to: pwolff@emory.edu Language as Language-of-Thought
Department of Psychology, Emory University, Atlanta, GA, USA Language surely affects thought if the units of
DOI: 10.1002/wcs.104 thought are words from natural language. This

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 253

 Advanced Review wires.wiley.com/cogsci

Language affects thought

Thought is language Thought is separate from language

(Language as language-of-thought)

Thought and language are structurally parallel Thought and language differ structurally
(Linguistic determinism)

Thinking before language Thinking with language Thinking after language

(Thinking for speaking)

Linguistic representations Linguistic representations Language makes certain Language primes certain
compete with nonlinguistic extend/enable nonlinguistic properties highly salient types of processing in
representations representations in nonlinguistic thinking nonlinguistic thinking
(Language as meddler) (Language as augmenter) (Language as spotlight) (Language as inducer)

FIGURE 1 | Classes and subclasses of hypotheses on how language might affect thought.

version of how language influences thinking has been to specify the meanings of words and constructions in
advanced by a number of theorists, including Plato, language (see Refs 6,11,12,21).
‘[T]he soul when thinking appears to me to be just
talking. . .’7 (p. 252), and Kant, ‘Thinking is speaking
to ourselves’8 (p. 278). Max Müller perhaps stated the Linguistic Determinism
position most directly when he asserted, ‘Language is The concept of linguistic relativity was championed
identical with thought’9 (p. ii). In psychology, the in the 1950s by the amateur linguist Benjamin
conflation of language with thought is exemplified by Lee Whorf.22 Whorf argued for what has come
the views of behaviorist John Watson, who proposed to be known as linguistic determinism, the view
that thought should be equated with the production that language determines the basic categories of
of (subvocal) speech.10 thought and that, as a consequence, speakers of
Clearly, this version of the language–thought different languages think differently.a In linguistic
interface cannot be right.6,11,12 As argued by Pinker,6 determinism, the shaping role of language is held
people can have thoughts that are difficult to express, to be so strong that it can even overwrite pre-
but this would never be the case if thoughts were existing perceptual and conceptual capabilities,23 in
represented entirely in natural language. People a manner analogous to the way infants lose the
can also understand linguistic expressions that are ability to notice phonetic distinctions absent in their
ambiguous, such as Kids make nutritious snacks, but native language.24 Linguistic determinism differs from
their very ability to recognize this ambiguity implies language as language-of-thought in that it separates
a finer level of representation than that encoded in language from the conceptual system.25,26 While this
the meanings of words.6 If people thought entirely distinction represents an important advance over the
in words, words expressing new concepts could previous proposal, linguistic determinism can still be
never be coined because there would be no way of rejected because it makes untenable predictions about
imagining their meanings. Further, research indicates the relationship between language, thought, and the
that infants and nonhuman primates are capable of world.
relatively sophisticated forms of thinking, even in the Linguistic determinism holds that differences
absence of language.13–20 These arguments point to a in language cause differences in thought. This view
medium of thought for categorization, reasoning, and implies a relatively tight connection between language
memory—conceptual representations, or mentalese— and thought and a loose connection between thought
that is independent of the kinds of representations used and the world (see the left side of Figure 2). This

254  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

Language (semantic system) Language (semantic system) THINKING BEFORE LANGUAGE

One type of thinking that might be influenced by
Thought (conceptual system) Thought (conceptual system)
language is the thinking that occurs immediately prior
to using language—that is, the thought processes
World World associated with producing speech. Such an influence
might be expected to produce differences in thought
Linguistic determinism View from the cognitive sciences across languages because languages differ with respect
FIGURE 2 | In linguistic determinism, the relationship between to the aspects of experience to which their users
language and thought is tight, while the relationship between thought must attend. In English, for example, but not in
and the world is loose. Research in the cognitive sciences suggests the Indonesian or Mandarin, verbs must specify tense,
opposite pattern: a loose relationship between language and thought, so presumably English speakers must attend to when
and a relatively tight relationship between thought and the world. an event occurred. In Turkish, descriptions of past
events must indicate whether they were witnessed or
nonwitnessed.38
pattern of relationships is expected because thought is
held to be determined by language, not the world. Thinking for Speaking
Research from the cognitive sciences suggests In using language, then, speakers must engage in a
a different pattern of relationships, namely that special kind of mental activity—attending to certain
the connection between thought and the world is aspects of experience—that Slobin38,39 has called
tighter than the connection between thought and thinking for speaking. Thinking for speaking has
language12,27–30 (see the right side of Figure 2), been observed in people’s attentional patterns and
especially in the case of nominal concepts.31,32 memory for motion events.36,40 The effect stems from
Evidence for a loose connection between language and the well-known phenomenon that certain languages
thought comes from cross-linguistic studies showing (e.g., English, German, Russian, Chinese) tend to
that differences in word meanings across languages encode manner in the main verb (e.g., jog, roll,
are greater than differences in the underlying concepts. march) and path in a variety of other linguistic
For example, as shown by Malt et al.,33 speakers of structures, while other languages (e.g., Greek, Spanish,
English, Spanish, and Chinese diverged significantly French, Japanese) do just the opposite. In a study
when labeling pictures of common storage containers by Papafragou et al.,40 eye-movement patterns of
but performed comparably when making similarity native speakers of English and Greek were monitored
judgments about the same containers. A similar as they watched motion events. When participants
pattern of results was observed in an experiment were instructed to watch the events in preparation
comparing French-speaking and Dutch-speaking for describing them verbally, Greek speakers were
Belgians.34 The two groups of Belgians resembled much more likely than English speakers to focus on
each other in their similarity judgments for storage path over manner. When participants simply watched
containers but diverged in their naming patterns, the motion events freely, however, the eye-movement
despite sharing essentially the same culture. This patterns were largely the same for the two language
kind of disconnect between language and thought groups (except at the very end of the events; see the
has been observed in other domains as well. For section ‘Motion’ under Language as Meddler below).
example, Munnich et al.35 found that Korean and In similar research by Gennari et al.,36 similarity
English speakers’ memory for spatial locations was ratings for motion events by English and Spanish
far more similar than their naming patterns, and speakers conformed to language-specific patterns
Gennari et al.36 observed that English and Spanish when they were instructed to verbally describe the
speakers’ memory for motion events was more similar events, but not when there was no verbal encoding.
in comparison to how they named such events (see
also Ref 37).
The lack of alignment between language THINKING WITH LANGUAGE
and thought raises severe problems for linguistic In thinking for speaking, the effect of language on
determinism, but these problems do not imply that thought occurs immediately before the production
language cannot have an effect on thought. Indeed, it of language. However, much recent research points to
is because language and the conceptual system differ another kind of effect of language on thought, namely,
that we might expect a tension between them, driving one in which processes associated with language are
each system to exert an influence on the other. activated along with nonlinguistic processes. Thus,

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 255

 Advanced Review wires.wiley.com/cogsci

in this kind of effect, thinking occurs with language. Color

One of the hallmarks of this kind of language effect is The color domain has been of central interest in
that it can be eliminated by having people engage in research on the relationship between language and
a verbal interference task—that is, a secondary task thought. In the 1970s, such research cast a pall over
that recruits verbal processing. There appear to be the possibility that language might influence thought
two general classes of this kind of language effect. with the findings that inventories of color terms
shared significant commonalities across languages42
and that any linguistic differences did not correspond
Language as Meddler to differences in categorization behavior.43
In one class, the effects of language occur from Several recent studies indicate, however, that
the spontaneous recruitment of linguistic codes in language may have an influence on color cognition.
tandem with nonlinguistic codes. Linguistic codes, Work with the Berinmo, a small tribe in New Guinea
in effect, meddle with nonlinguistic codes in the whose language has 5 basic color terms (compared
process of making a decision. When the linguistic and to 11 in English), is a case in point. Controlling
nonlinguistic codes are consistent with each other, for a confound in previous research,44 Roberson
speed and accuracy are facilitated, but when they and colleagues found that the Berinmo’s recognition
conflict, speed and accuracy may be compromised.41 memory was better for the focal colors of their own
language than for those of English.44–47 In a similar
Motion line of research, Winawer et al.41 found that an
An effect of linguistic meddling is suggested by obligatory color distinction in Russian between siniy
Papafragou et al.’s40 study, described in the previous (dark blue) and goluboy (light blue) led to differences
section. The eye-movement patterns of the English in color discrimination. Russian speakers, but not
and Greek speakers differed not only in the linguistic English speakers, performed faster on a matching
condition, but also in the non-linguistic condition, at task when the colors belonged to different linguistic
the very end of the animations. Participants shifted categories than when they belonged to the same
their attention to aspects of the scenes not typically category. Moreover, these cross-linguistic differences
encoded in verbs in their language, perhaps, as disappeared under conditions of verbal interference.
suggested by Papafragou et al., to compensate for A similar effect was found for English by Gilbert
relatively greater early attention to typically encoded et al.;48 participants were faster to locate a target
aspects. Thus, at the end of the animations, English when its linguistic category differed from that of the
speakers attended preferentially to path (from an surrounding distractors (e.g., a green among blues),
earlier relative preference for manner) and Greek and slower when the target and distractors shared
speakers attended equally to both manner and path the same linguistic category (e.g., a green among
(from an earlier preference for path). Papafragou other shades of green), but only when the target was
et al. speculated that these cross-linguistic differences presented in the right visual field. This lateralization
reflected different approaches to the linguistic coding effect was presumably due to the fact that presentation
of the scene in memory. Importantly, these attentional in the right visual field entails that the stimulus will
differences do not constitute a type of thinking for initially be processed in the left hemisphere, the side of
speaking because the very point of the nonlinguistic the brain where language processing typically occurs.
condition was to examine what people would do Further, as in the study by Winawer et al., the effect
when they were not asked to put the scenes was eliminated by a verbal interference task.
into words. Instead, these differences exemplify The findings described above suggest that
an effect of language on thought that occurred language can meddle with cognition via the interaction
from the unprompted, spontaneous generation of of perceptual and linguistic codes.41,47,49 The results
linguistic codes, which consequently meddled with do not imply a change to the underlying perceptual
how participants attended to the scenes. Greek machinery or memory representations because the
speakers also had poorer memory for the events than effects of language on cognition were disrupted by
English speakers, a difference the authors attribute verbal interference tasks (but see Ref 50). It should
to processing costs associated with trying to attend be noted that, the effects in these studies cannot be
to both manner and path. But once again, the attributed to task demands. Some linguistic relativity
effect on memory was not due to any promptings experiments have used tasks with questions for which
from the experimenter or to task demands, but there were no objectively correct answers; under these
rather, apparently, from the spontaneous generation conditions, participants might use linguistic codes to
of linguistic codes. choose between two alternatives because they have

256  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

no other basis for making a decision.11,28,36 However, allows for the specification of exact quantities, such
the color tasks described above all had objectively as the quantity 31. An exact number system must
correct answers. Despite the fact that language was be explicitly learned, and several sources of evidence
not needed to solve these tasks, linguistic codes were suggest that this third system depends on language.
generated nonetheless. In a study by Dehaene et al.,59 Russian–English
bilinguals were trained on exact and approximate
number addition in either Russian or English. After
Language as Augmenter training, performance on the exact addition problems
In the case of language as meddler, a decision was faster in the trained language than in the untrained
can be made on the basis of either linguistic language, suggesting that the results of training were
or nonlinguistic representations. In certain cases, stored in a language-specific format. Performance on
however, linguistic representations may combine with the estimation problems, in contrast, was unaffected
nonlinguistic representations to enable people to by the language of training, suggesting the use of
perform tasks that could not be completed with mental codes that were independent of language.
either type of representation alone. In such cases, as A functional magnetic resonance imaging (fMRI)
argued by Gentner51,52 and Frank et al.,53 language study supported these results by showing that the
may augment thought by offering new conceptual exact number task recruited neural networks typically
tools. This idea can be illustrated by problems like associated with language processing, whereas the
the one in Figure 3. If the first gear in Figure 3 turns estimation task recruited areas in both parietal lobes
clockwise, in which direction will the last gear turn? not typically associated with language processing.
This problem could be solved by mental simulation; While the results from Dehaene et al.59 suggest
that is, by imagining the first gear turning to the right,
an effect of language on cognition, they do
then the second gear turning to the left, and so on.54
not demonstrate linguistic relativity. Exact number
Alternatively, people might notice that each successive
calculation was no better or worse in Russian or
gear turns in the opposite direction from the previous
English. To make the case for an effect of linguistic
one and generate the parity rule that ‘odd and even
relativity per se, it needs to be shown that the effects
gears turn in different directions’.55 This rule, which
differ across languages. Such evidence has recently
may depend on linguistic coding, can then be applied
been found in work on languages with so-called
more quickly than the laborious process of mentally
one-two-many number systems. Number words in
rotating each gear. In problems like this one, the
Pirahã, a language spoken by a small tribe in Brazilian
constant meddling of language may pay off because it
Amazonia, map only roughly onto the quantities ‘one’
gives rise to a new way of representing the problem,
and ‘two’. As observed by Frank et al.,53 the Pirahã’s
allowing for quicker and more accurate answers.
word for one, hói (falling tone), may be used to
Number described as many as 6 items, the word for two, hoí
A similar kind of re-representation seems to occur in (rising tone), as many as 4–10 items, and the word
the domain of number. There appear to be three main for many, baagi, between 7 and 10 items. Effectively,
systems for representing numerical quantities. One of then, Pirahã lacks words for exact quantities.
these systems involves a fast ‘subitizing’ procedure that Gordon25,60 investigated the potential conse-
allows everyone from infants to adults to recognize quences of this absence of number words on tasks
small numbers of items (≤4) automatically without requiring exact quantities. In one such task, Pirahã
having to count.56 A second system allows animals, speakers were asked to line up batteries on a table
infants, and adults to discriminate larger quantities, across from a set of nuts arranged in a line. In a more
but only approximately, such as the rough amount difficult version of this task, they were asked to match
of sand in a bucket or fish in a net.57,58 These two the items along an axis orthogonal to the one used by
systems are thought to be innate. The third system the experimenter. The main result was that the Pirahã
were unable to perform the task accurately, but their
responses were not random: as the number of items
increased, the Pirahã tended to put out more items,
though rarely the exact number. The results suggest
that the Pirahã tried to solve the task using an approx-
imate number system. Gordon’s findings align well
with Dehaene et al.’s59 study in indicating that exact
FIGURE 3 | Series of gears in which the first turns clockwise. In magnitude calculation requires language. The most
which direction will the last gear turn? critical findings in Gordon’s study were replicated in

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 257

 Advanced Review wires.wiley.com/cogsci

a set of studies by Frank et al.,53 who conclude, like say I know that she thinks X, but what is actually true
Gordon,25,60 that the Pirahã’s conceptual gaps were is Y. The development of false belief understanding
due to gaps in their language. may be rooted in more fundamental executive
An important question raised by Gordon’s and functioning abilities (e.g., dimension switching),66
Frank et al.’s findings concerns the exact way in which which undergo marked developmental change during
language might impact number cognition. Pinker the preschool years.67–69 Interestingly, children’s
and Jackendoff61 and Bloom62 have suggested that performance on dimension-switching tasks (e.g.,
children may learn the number system by co-opting sorting first by shape, then by color) improves
the mental machinery of language used for iterative substantially when language is used to highlight
and recursive processing. Concepts like five and six, the conflicting dimensions.68,69 As with false belief
and odd and even, depend on a system of generative understanding, language may improve performance
rules to give them meaning. The fact that exact number by enabling children to represent embedded rule
is learned much later than language, and is effectively structures—in this case, the higher-order rules that
unique to humans, supports the view that language govern when each individual sorting rule should be
might be used to reason about numbers. Among applied.70
other capacities, language may support the formation
of rules, and perhaps even more importantly, the Spatial Analogies
embedding of rules within other rules. If language aids in the formation of embedded
knowledge structures, it is likely to have a significant
role in many other cognitive activities. As argued
False Belief Understanding by Gentner,51,52 hierarchically structured relational
This type of language effect is potentially significant
knowledge allows people to discover abstract
because rule embedding appears to be essential in
commonalities that can lead to more explicit and
a number of domains, including, for example, the
uniform units of thought. In a spatial analogy study,
representation of false beliefs.63–65 An especially Loewenstein and Gentner71 had 3.5-year-old children
compelling case for the role of language in false watch as a star was placed behind a card on the
belief understanding has recently been made with top, middle, or bottom shelf of a small shelving unit.
Nicaraguan adults who learned Nicaraguan Sign They were then asked to find the star in an almost
Language (NSL).64 NSL first appeared in 1970 with identical shelving unit nearby. Children’s performance
the creation of special-education schools. The first on the task improved if, during the hiding event,
cohort learned an early form of the language, which the experimenter indicated the star’s location using
was elaborated by the second cohort. Although the relational words such as on, under, top, or bottom.
language skills of these two groups differed, the The results suggest that relational language helped the
sociocultural history of the two groups was essentially children align the two sets of spatial relations.
the same: both attended the same schools for the
same number of years, had the same teachers, and Category Learning
had comparable social networks. Pyers and Senghas64 Beyond supporting the representation of embedded
found that the second cohort knew significantly more rule structures and relational knowledge, language
mental state signs (e.g., think, know) than the first may extend nonlinguistic cognition in several other
cohort, indicating a more developed lexicon for this ways. As suggested by Waxman and Markow,72
domain. False belief understanding was measured language may serve as an invitation to form new
using a low-verbal task in which participants were categories (see also Ref 73). It may also facilitate
shown a sequence of picture cards and asked to choose category learning. This type of influence was
the last card to complete a story. The surprising finding demonstrated in a set of studies by Lupyan et al.,74
was that false belief understanding was significantly in which participants learned to distinguish between
stronger in the younger second cohort signers than approachable and nonapproachable alien creatures.
in the older first cohort signers. As emphasized These categories could be learned on the basis of
by Pyers and Senghas, the participants in the two visual information alone, so learning labels for the
groups were comparable, except in their language categories was not necessary for completing the task.
ability, suggesting that the difference in false belief Nevertheless, category learning was much faster when
understanding was due to language. Following de it was accompanied by auditory or written labels than
Villiers and de Villiers,63 Pyers and Senghas suggest not. In a subsequent work, Lupyan75 has shown that
that the capacity to represent false beliefs may depend, categories learned with labels are more resistant to
at least in part, on people’s ability to represent interference from novel stimuli and more flexible in
embedded propositional structures, as when people their ability to incorporate new members.

258  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

THINKING AFTER LANGUAGE from the biological gender of the proper noun’s
referent. They also demonstrated that attention to
The effects of linguistic meddling and augmenting different aspects of an object could be manipulated
occur when thought and language work in experimentally by having English speakers learn a
tandem—that is, when thinking recruits linguistic novel language with grammatical gender and that the
representations online. Yet another major way in
effects of grammatical gender extended to people’s
which language might affect thought is as an influence
judgments about the similarity of unlabeled pictures.
after the use of language. The long-term use of a
Interestingly, these effects persisted even when people
language may direct habitual attention to specific
were engaged in a verbal interference task, suggesting
properties of the world, even in nonlinguistic contexts.
that the results were not due to the online recruitment
At a more general level, language use may also induce
of language, but rather to attentional biases acquired
a given mode of processing, which may persist even as
through the frequent, habitual use of language.
people engage in other nonlinguistic tasks. Unlike the
effects described in the previous section, these effects
of ‘thinking after language’ should be less attenuated Spatial Frames of Reference
by verbal interference tasks, since they occur after Another way in which language might affect how
language is no longer in use, rather than involving the people attend to the world is to promote a particular
recruitment of linguistic codes during processing. framework for conceptualizing space. Much research
supports the proposal that representations of space
utilize one of three possible frames of reference.5,78–84
Language as Spotlight An absolute (or geocentric) frame of reference involves
After exposure to words and constructions that a coordinate system in which the main axes are placed
highlight specific properties, attention may linger on within the larger environment (e.g., a house facing
those properties. In effect, language may act as a east). An intrinsic (or object-centric) frame of reference
spotlight, making certain aspects of the world more places the axes in objects (e.g., the front of a car).
salient than others. Finally, a relative (or egocentric) frame of reference
defines the axes with respect to the viewer’s own body
Grammatical Gender (e.g., the comb to my left). Findings from several
Work by Boroditsky and colleagues demonstrates how sources indicate that all three frames of reference are
such effects might occur as a result of exposure available to humans across cultures.78–81 Levinson
to grammatical gender.76 Grammatical gender is a and his colleagues have argued, however, that there
feature of nouns in many languages (English being a are cross-cultural differences in people’s preference
notable exception), whereby all nouns are assigned a and proficiency with these frames of reference,
gender. In both German and Italian, for example, the and further, that these biases stem from linguistic
words for ‘hammer’, ‘spoon’, and ‘screwdriver’ are differences.5,78,79,82 As documented by Majid et al.,5
masculine, while the words for ‘fork’, ‘bottle’, and languages vary in the frequency with which they
‘scissors’ are feminine.77 Languages often conflict in encode the three frames of reference. In English, the
their assignment of grammatical gender. For example, dominant spatial frames are the relative and intrinsic,
the word for ‘key’ is masculine in German and whereas in Tzeltal, a Mayan language spoken in
feminine in Spanish, while the word for ‘bridge’ is Mexico, the dominant frames are the absolute and
feminine in German and masculine in Spanish.76 This intrinsic. Levinson and his colleagues speculate that,
cross-linguistic variability suggests that grammatical over development, these arguably innate concepts
gender is not determined by the correlational structure become progressively re-represented to match the
of the world, but rather, in large part, by factors structures used in the learner’s language. Levinson82
that are specific to particular languages. Given conducted one of the earliest tests of this hypothesis
that the categories of masculine and feminine are with speakers of Dutch and Tzeltal. Participants were
language-specific, it can then be asked whether shown three objects organized in a row; then they
these language-imposed categories have consequences were rotated 180◦ and instructed to ‘remake the
for the kinds of properties people attend to when array just as it was’. As predicted, the Dutch speakers
thinking about objects. Boroditsky and her colleagues arranged the items according to an egocentric frame
found support for this possibility. In particular, they of reference, while the Tzeltal speakers appeared to
found that Spanish and German speakers’ ability use an absolute frame (see also Ref 79).
to learn associations between proper and common These findings are consistent with the possibility
nouns (e.g., Tom and apple) was disrupted when the that the regular use of language can lead people to
grammatical gender of the common noun differed prefer one construal of the world over others. It should

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 259

 Advanced Review wires.wiley.com/cogsci

be emphasized that Levinson et al.’s findings do not the world into discrete objects (books, flowers) and
imply that people are unable to use frames of reference unbounded continuous masses (rice, sand). In English,
that are not regularly encoded in their language. names for objects typically imply individuation. For
Indeed, as demonstrated by Li et al.,84 Tzeltal speakers example, when referring to multiple chairs in a room,
can, in fact, use their non-dominant frame of reference, we must use the plural marker. In languages like
the egocentric frame, without any apparent difficulty. Japanese and Yucatec Maya, such markers are usually
Still to be determined is whether, at the time of not needed; it is as if the noun for chair means ‘chair
encoding, participants in Levinson et al.’s experiments stuff’.27,91,92
used language to help encode the spatial relations. If Several studies have investigated whether this
so, the results from these experiments might be better difference in how objects are linguistically individu-
classified as a type of thinking with language, in which ated might have an impact on thought. In categorizing
language acts as a meddler or augmenter. objects, speakers of languages like English might be
biased to focus on shape, since objects can be indi-
Spatial Relations viduated on the basis of shape. In contrast, speakers
The idea that language might lead people to focus of languages like Japanese and Yucatec Maya might
on certain aspects of experience at the exclusion of be biased to attend to material, since their nouns do
others has also been examined with respect to the not explicitly individuate objects. These predictions
encoding of local spatial relations. As documented were borne out in several studies in which people
by Bowerman,85 the encoding of spatial relations were asked to determine whether a particular entity
varies greatly across languages. For example, verbs (e.g., a ceramic lemon squeezer) was more similar to
of placement in Korean distinguish between tight another entity that shared its shape (e.g., a wooden
and loose fit and ignore the distinction between lemon squeezer) or its material (e.g., bits of ceramic),
containment (e.g., ‘put in’) and support (e.g., ‘put on’), as indexed by novel noun generalization or explicit
while the converse is true for English prepositions.86–89 similarity judgments. For example, Imai et al.92,93 and
Work with infants suggests an ability to distinguish Lucy and Gaskins91 found that English-speaking chil-
tight from loose fit as early as 5 months (Refs 14,87,88 dren and adults tended to choose according to shape,
but see Ref 90). However, Choi88 has found that by while Japanese and Yucatec Maya children and adults
3 years of age, sensitivity to the distinction between tended to choose according to material. Recent work
tight and loose fit diminishes greatly in English- by Li et al.94 replicated these findings with speakers of
speaking children. Similarly, McDonough et al.87 Japanese, Mandarin, and English. However, this same
found that whereas Korean-speaking adults remained research showed that such cross-linguistic differences
sensitive to the distinction between tight and loose did not prevent speakers of languages like Japanese
fit, English-speaking adults were relatively insensitive from being able to think about fixed regular-shaped
to this distinction. Importantly, the results from Choi objects as individuated entities. As in the case of spa-
and McDonough et al. involved preferential-looking tial frames of reference and spatial relations, language
paradigms, so they do not indicate that English- may lead people to spontaneously focus on certain
speaking 3-year-olds and adults are unable to perceive aspects of experience, but it does not appear to rigidly
or conceptualize the distinction between tight and prevent people from considering aspects of experience
loose fit, only that they are not biased to focus on this not encoded in their language.
dimension.14 Together, the findings on local spatial
relations suggest that early on, infants are sensitive
to a wide range of spatial distinctions, but that
over development, they may develop biases toward
LANGUAGE AS INDUCER
certain distinctions, specifically those encoded in their When language acts as a spotlight, certain aspects
language, over others. However, the findings do not of the world are highlighted, in particular, those
imply that people lose the fundamental ability to that are encoded in the meanings of specific words
perceive spatial distinctions not regularly made in and constructions. However, there is another, per-
their language. haps more general way in which language might
affect thought: specifically, language may prime a
Objects and Substances particular mode of processing that continues to
One final area in which language might bias peo- be engaged even after language is no longer in
ple to attend to particular aspects of experience is use. This possibility was supported in a recent
with respect to the distinction between objects and set of studies on the simulation of motion in
substances. Languages differ in how they partition static scenes. In a replication of Freyd et al.,95

260  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

Holmes and Wolff96 found that when an object thinking (language as augmenter), and those in which
supporting another object was suddenly removed (e.g., thinking is directed toward properties highlighted by
a pedestal beneath a potted plant disappeared), peo- language (language as spotlight) or in which language
ple appeared to simulate the effect of gravity on the engages a schematic mode of processing (language
unsupported object, as evidenced by their insensitivity as inducer). Our conclusions are based, in part, on
to downward changes in the position of the unsup- several recurring findings in the field. First, we did
ported object. Holmes and Wolff also observed that not find empirical support for the view that lan-
this mental simulation of gravity was much more guage determines the basic categories of thought
likely to occur when participants were presented with or that it ‘closes doors’. Once people are able to
schematic line drawings of a scene, as opposed to pho- make a particular conceptual distinction, this abil-
torealistic images. However, when participants were ity is retained, even if it is not explicitly encoded
instructed to write a verbal description of the photore- in one’s language. For example, English speakers
alistic images, the simulation of gravity subsequently retain the ability to distinguish tight and loose fit,
occurred for these stimuli as well. Further, there was even though this distinction is not encoded in their
a positive correlation between the proportion of rela- spatial preposition system.14 There is evidence, how-
tional terms (e.g., verbs and prepositions) participants ever, that while language may not close doors, it
used in their descriptions and the magnitude of the may fling others wide open. For example, language
simulation effect. Interestingly, this influence of lan- makes certain distinctions difficult to avoid when it
guage was found to be scene-independent: relational meddles in the process of color discrimination47 or
language promoted mental simulation even when peo- renders one way of construing space more natural
ple described a completely different scene than the than another.82 Lastly, language can sometimes build
one on which they were tested. The results suggest new doors. For example, language may underlie our
that language, and relational language in particular, ability to represent exact numbers25,53,60 and enter-
can induce people to conceptualize experience in a tain false beliefs.64 Thus, language may not replace,
relatively schematic manner, a mode of processing but instead may put in place, representational sys-
effective in facilitating mental simulation. tems that make certain kinds of thinking possible.
Although the mechanism differs from that which
CONCLUSION Whorf originally proposed, current research suggests
that language can still have a powerful influence on
Our survey of the field suggests that at least two
thought.
versions of the Whorfian hypothesis can be dismissed,
namely those based on language as language-of-
thought and linguistic determinism. On the other
hand, five other versions of the Whorfian hypothe-
NOTE
sis have garnered empirical support: those in which
a
thinking occurs before language use (thinking for It should be noted that Whorf only argued for lin-
speaking), those in which linguistic and nonlinguistic guistic determinism in a portion of his writings. In
codes compete with each other (language as med- other parts, he seemed to be arguing for the idea that
dler) or in which linguistic codes extend nonlinguistic language can act as a meddler or spotlight.

REFERENCES
1. Gentner D, Goldin-Meadow S. Whither Whorf. In: 4. Evans N, Levinson S. The myth of language universals:
Gentner D, Goldin-Meadow S, eds. Language in language diversity and its importance for cognitive
Mind: Advances in the Study of Language and science. Behav Brain Sci 2009, 32:429–492.
Thought. Cambridge, MA: MIT Press; 2003, 1–14.
5. Majid A, Bowerman M, Kita S, Haun DBM, Levinson
2. Gumperz JJ, Levinson SC. Introduction to part I. In: SC. Can language restructure cognition? The case for
Gumperz JJ, Levinson SC, eds. Rethinking Linguis- space. Trends Cogn Sci 2004, 8:108–114.
tic Relativity. Cambridge, UK: Cambridge University
Press; 1996, 21–36. 6. Pinker S. The Language Instinct. New York: Morrow;
1994.
3. Malt BC, Wolff P. Words and the Mind: How Words
Capture Human Experience. New York: Oxford Uni- 7. Plato. The Dialogues of Plato Translated into English
versity Press; 2010. with Analyses and Introductions by B. Jowett, vol. 4

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 261

 Advanced Review wires.wiley.com/cogsci

(Parmenides, Theaetetus, Sophist, Statesman, Phile- In: Malt B, Wolff P, eds. Words and the Mind: How
bus). Oxford, UK: Oxford University Press; 1892. Words Capture Human Experience. Oxford: Oxford
8. Kant I. Anthropologie in Kants Gesammelte Schriften. University Press; 2010, 199–218.
In: Butts RE, ed. The Grammar of Reason: Hamann’s 26. Lucy JA. Language structure, lexical meaning, and
Challenge to Kant. Synthese, vol. 75. 1798/1988, cognition: Whorf and Vygotsky revisited. In: Malt B,
251–28. Wolff P, eds. Words and the Mind: How Words Cap-
9. Müller FM. The Science of Thought. London: Long- ture Human Experience. Oxford: Oxford University
mans, Green & Co; 1909. Press; 2010, 266–286.
10. Watson JB. Psychology as the behaviorist views it. 27. Smith LB, Colunga E, Yoshida H. Making an ontol-
Psychol Rev 1913, 20:158–177. ogy: cross-linguistic evidence. In: Rakison DH,
Oakes LM, eds. Early Category and Concept Devel-
11. Pinker S. The Stuff of Thought: Language as a Window opment: Making Sense of the Blooming, Buzzing
into Human Nature. New York: Penguin; 2007. Confusion. New York: Oxford University Press; 2003,
12. Levinson SC. From outer to inner space: linguistic 275–302.
categories and non-linguistic thinking. In: Nuyts J, 28. Gleitman L, Papafragou A. Language and thought. In:
Pederson E, eds. Language and Conceptualization. Holyoak K, Morrison B, eds. Cambridge Handbook
Cambridge: Cambridge University Press; 1997, of Thinking and Reasoning. Cambridge: Cambridge
13–45. University Press; 2005, 633–661.
13. Bloom P, Keil F. Thinking through language. Mind 29. Papafragou A, Massey C, Gleitman L. Shake, rattle,
Lang 2001, 16:351–367. ‘n’ roll: the representation of motion in language and
14. Hespos SJ, Spelke ES. Conceptual precursors to lan- cognition. Cognition 2002, 84:189–219.
guage. Nature 2004, 430:453–456. 30. Wolff P, Malt BC. The language-thought interface: an
15. Needham A, Baillargeon R. Object segregation in introduction. In: Malt BC, Wolff P, eds. Words and
8-month-old infants. Cognition 1997, 62:121–149. the Mind: How Words Capture Human Experience.
16. Spelke S. What makes us smart? Core knowledge and Oxford: Oxford University Press; 2010, 3–15.
natural language. In: Gentner D, Goldin-Meadow S, 31. Gentner D. Why nouns are learned before verbs:
eds. Language in Mind: Advances in the Study of linguistic relativity versus natural partitioning. In:
Language and Thought. Cambridge, MA: MIT Press; Kuczaj SA, ed. Language Development: Vol. 2. Lan-
2003, 277–311. guage, Thought and Culture. Hillsdale, NJ: Lawrence
Erlbaum Associates; 1982, 301–334.
17. Cheney DL, Seyfarth RM. The representation of social
relations by monkeys. Cognition 1992, 37:167–196. 32. Gentner D, Boroditsky L. Individuation, relativity, and
early word learning. In: Bowerman M, Levinson SC,
18. Hare B, Call J, Tomasello M. Do chimpanzees know
eds. Language Acquisition and Conceptual Develop-
what conspecifics know? Animal Behav 2001,
ment. Cambridge, UK: Cambridge University Press;
61:139–151.
2001, 215–256.
19. Phillips W, Santos LR. Evidence for kind representa-
33. Malt BC, Sloman SA, Gennari S, Shi M, Wang Y.
tions in the absence of language: experiments with
Knowing versus naming: similarity and the linguis-
rhesus monkeys (Macaca mulatta). Cognition 2007,
tic categorization of artifacts. J Mem Lang 1999,
102:455–463.
40:230–262.
20. Penn DC, Holyoak KJ, Povinelli DJ. Darwin’s mis-
34. Ameel E, Storms G, Malt BC, Sloman SA. How bilin-
take: explaining the discontinuity between human and
guals solve the naming problem. J Mem Lang 2005,
nonhuman minds. Behav Brain Sci 2008, 31:109–178.
53:60–80.
21. Fodor J. The Language of Thought. Cambridge: Har-
35. Munnich E, Landau B, Dosher BA. Spatial language
vard University Press; 1975.
and spatial representation: a cross-linguistic compari-
22. Whorf BL. In: Carroll JB, ed. Language, Thought, and son. Cognition 2001, 81:171–207.
Reality: Selected Writings of Benjamin Lee Whorf. 36. Gennari S, Sloman SA, Malt BC, Fitch WT. Motion
Cambridge, MA: MIT Press; 1956. events in language and cognition. Cognition 2002,
23. Bowerman M, Levinson SC. Introduction. In: 83:49–79.
Bowerman M, Levinson SC, eds. Language Acquisi- 37. Papafragou A, Massey C, Gleitman L. When English
tion and Conceptual Development. Cambridge: Cam- proposes what Greek presupposes: the cross-
bridge University Press; 2001, 1–16. linguistic encoding of motion events. Cognition 2006,
24. Werker J, Tees R. Cross-language speech perception: 98:B75–B87.
evidence for perceptual reorganization during the first 38. Slobin DI. From ‘‘thought and language’’ to ‘‘think-
year of life. Inf Behav Develop 1984, 7:49–63. ing for speaking’’. In: Gumperz J, Levinson SC, eds.
25. Gordon P. Worlds without words: commensurabil- Rethinking Linguistic Relativity. Cambridge, MA:
ity and causality in language, culture, and cognition. Cambridge University Press; 1996, 70–96.

262  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

39. Slobin DI. Language and thought online: cognitive 54. Hegarty M. Mechanical reasoning by mental simula-
consequences of linguistic relativity. In: Gentner D, tion. Trends Cogn Sci 2004, 8:280–285.
Goldin-Meadow S, eds. Language in Mind: Advances 55. Schwartz DL, Black JB. Shuttling between depictive
in the Investigation of Language and Thought. Cam- models and abstract rules: induction and fallback.
bridge, MA: MIT Press; 2003, 157–191. Cogn Sci 1996, 20:457–497.
40. Papafragou A, Hulbert J, Trueswell J. Does language 56. Wynn K. Origins of numerical knowledge. Math Cogn
guide event perception? Evidence from eye movements. 1995, 1:35–60.
Cognition 2008, 108:155–184.
57. Dehaene S. The Number Sense. New York: Oxford
41. Winawer J, Witthoft N, Frank MC, Wu L, Wade A, University Press; 1997.
Boroditsky L. Russian blues reveal effects of language
on color discrimination. Proc Natl Acad Sci 2007, 58. Gallistel CR. The Organization of Learning. Cam-
104:7780–7785. bridge, MA: MIT Press; 1990.
42. Berlin B, Kay P. Basic Color Terms: Their Universality 59. Dehaene S, Spelke E, Pinel P, Stanescu R, Tsivkin S.
and Evolution. Berkeley, CA: University of California Sources of mathematical thinking: behavioral and
Press; 1969. brain-imaging evidence. Science 1999, 284:970–974.
43. [Rosch] Heider ER. Universals in color naming and 60. Gordon P. Numerical cognition without words: evi-
memory. J Exper Psychol 1972, 93:10–20. dence from Amazonia. Science 2004, 306:496–499.
44. Roberson D, Davies I, Davidoff J. Color categories are 61. Pinker S, Jackendoff R. The faculty of language:
not universal: replications and new evidence from what’s special about it? Cognition 2005, 95:201–236.
a stone-age culture. J Exper Psychol: Gen 2000, 62. Bloom P. Generativity within language and other cog-
129:369–398. nitive domains. Cognition 1994, 51:177–189.
45. Davidoff J, Davies I, Roberson D. Colour categories 63. de Villiers JG, de Villiers PA. Language for thought:
in a stone-age tribe. Nature 1999, 398:203–204. coming to understand false beliefs. In In: Gentner D,
46. Roberson D, Davidoff J, Davies IR, Shapiro LR. Color Goldin-Meadow S, eds. Language in Mind: Advances
categories: evidence for the cultural relativity hypoth- in the Study of Language and Thought. Cambridge,
esis. Cogn Psychol 2005, 50:378–411. MA: MIT Press; 2003, 335–386.
47. Roberson D, Hanley JR. Relatively speaking: an 64. Pyers JE, Senghas A. Language promotes false-belief
account of the relationship between language and understanding: evidence from learners of a new sign
thought in the color domain. In: Malt B, Wolff P, eds. language. Psychol Sci 2009, 20:805–812.
Words and the Mind: How Words Capture Human 65. Milligan K, Astington JW, Dack LA. Language and
Experience. Oxford: Oxford University Press; 2010, theory of mind: meta-analysis of the relation between
183–198. language ability and false-belief understanding. Child
48. Gilbert AL, Regier T, Kay P, Ivry RB. Whorf hypoth- Develop 2007, 78:622–646.
esis is supported in the right visual field but not the 66. Moses LJ, Carlson SM, Sabbagh MA. On the speci-
left. Proc Natl Acad Sci 2006, 103:489–494. ficity of the relation between executive function
49. Regier T, Kay P, Gilbert AL, Ivry RB. Language and and children’s theories of mind. In: Schneider W,
thought: which side are you on, anyway? In: Malt B, Schumann-Hengsteler R, Sodian B, eds. Young Chil-
Wolff P, eds. Words and the Mind: How Words Cap- dren’s Cognitive Development: Interrelationships
ture Human Experience. Oxford: Oxford University among Executive Functioning, Working Memory,
Press; 2010, 165–182. Verbal Ability, and Theory of Mind. Mahwah, NJ:
50. Thierry G, Athanasopoulos P, Wiggett A, Dering B, Lawrence Erlbaum Associates; 2004, 131–145.
Kuipers J-R. Unconscious effects of language-specific 67. Zelazo PD, Frye D, Rapus T. An age-related dissoci-
terminology on preattentive color perception. Proc ation between knowing rules and using them. Cogn
Natl Acad Sci 2009, 106:4567–4570. Develop 1996, 11:37–63.
51. Gentner D, Christie S. Language and cognition in 68. Kirkham NZ, Cruess L, Diamond A. Helping chil-
development. Spivey MM, McRae K, Joanisse M, eds. dren apply their knowledge to their behavior on
The Cambridge Handbook of Psycholinguistics. Cam- a dimension-switching task. Develop Sci 2003,
bridge: Cambridge University Press. 6:449–476.
52. Gentner D. Why we’re so smart. In: Gentner D, 69. Jacques S, Zelazo PD. Language and the development
Goldin-Meadow S, eds. Language in Mind: Advances of cognitive flexibility: implications for theory of mind.
in the Study of Language and Thought. Cambridge, In: Astington JW, Baird JA, eds. Why Language Mat-
MA: MIT Press; 2003, 195–235. ters for Theory of Mind. Oxford: Oxford University
53. Frank M, Everett DL, Fedorenko E, Gibson E. Num- Press; 2005, 144–162.
ber as a cognitive technology: evidence from 70. Zelazo PD, Frye D. Cognitive complexity and control:
Pirahã language and cognition. Cognition 2008, a theory of the development of deliberate reasoning
108:819–824. and intentional action. In: Stamenov M, ed. Language

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 263

 Advanced Review wires.wiley.com/cogsci

Structure, Discourse, and the Access to Consciousness. Conference of the Cognitive Science Society. Hillsdale,
Amsterdam and Philadelphia: John Benjamins; 1997, NJ: Erlbaum; 2005, 1272–1277.
113–153. 85. Bowerman M. The origins of children’s spatial seman-
71. Loewenstein J, Gentner D. Relational language and tic categories: cognitive versus linguistic determinants.
the development of relational mapping. Cogn Psychol In: Gumperz JJ, Levinson SC, eds. Rethinking Linguis-
2005, 50:315–353. tic Relativity. Cambridge: Cambridge University Press;
72. Waxman S, Markow DB. Words as invitations to form 1996, 145–175.
categories: evidence from 12- to 13-month-old infants. 86. Choi S, Bowerman M. Learning to express motion
Cogn Psychol 1995, 29:257–302. events in English and Korean: the influence of
73. Gentner D, Namy LL. Comparison in the development language-specific lexicalization patterns. Cognition
of categories. Cogn Develop 1999, 14:487–513. 1991, 41:83–121.

74. Lupyan G, Rakison DH, McClelland JL. Language is 87. McDonough L, Choi S, Mandler JM. Understanding
not just for talking: redundant labels facilitate learning spatial relations: flexible infants, lexical adults. Cogn
of novel categories. Psychol Sci 2007, 18:1077–1083. Psychol 2003, 46:229–259.
88. Choi S. Influence of language-specific input on spatial
75. Lupyan G. Labels facilitate learning of novel cate-
cognition: categories of containment. First Lang 2006,
gories. In: Cangelosi A, Smith ADM, Smith KR, eds.
26:207–232.
Proceedings of the Sixth International Conference on
the Evolution of Language. Singapore: World Scien- 89. Casasola M. Can language do the driving? The effect
tific; 2006, 190–197. of linguistic input on infants’ categorization of support
spatial relations. Develop Psychol 2005, 41:183–192.
76. Boroditsky L, Schmidt LA, Phillips W. Sex, syntax,
and semantics. In: Gentner D, Goldin-Meadow S, eds. 90. Casasola M, Cohen LB. Infant categorization of con-
Language in Mind: Advances in the Study of Lan- tainment, support, and tight-fit spatial relationships.
guage and Thought. Cambridge, MA: MIT Press; Develop Sci 2002, 5:247–264.
2003, 61–79. 91. Lucy J, Gaskins S. Grammatical categories and the
77. Vigliocco G, Vinson DP, Paganelli F, Dworzynski K. development of classification preferences: a com-
Grammatical gender effects on cognition: implications parative approach. In: Bowerman M, Levinson S,
for language learning and language use. J Exper Psy- eds. Language Acquisition and Conceptual Develop-
chol: Gen 2005, 134:501–520. ment. Cambridge: Cambridge University Press; 2001,
257–283.
78. Haun DBM, Rapold C, Call J, Janzen G,
Levinson SC. Cognitive cladistics and cultural over- 92. Imai M, Gentner D. A crosslinguistic study of early
ride in Hominid spatial cognition. Proc Natl Acad Sci word meaning: universal ontology and linguistic influ-
2006, 103:17568–17573. ence. Cognition 1997, 62:169–200.
79. Levinson SC, Kita S, Haun DBM, Rasch BH. Return- 93. Imai M, Mazuka R. Re-evaluation of linguistic rel-
ing the tables: language affects spatial reasoning. ativity: language-specific categories and the role of
Cognition 2002, 84:155–188. universal ontological knowledge in the construal of
individuation. In: Gentner D, Goldin-Meadow S, eds.
80. Gallistel CR. Language and spatial frames of reference Language in Mind: Advances in the Issues of Lan-
in mind and brain. Trends Cogn Sci 2002, 6:321–322. guage and Thought. Cambridge, MA: MIT Press;
81. Newcombe NS. Language as destiny? Or not: essay 2003, 430–464.
review of Space in Language and Cognition: Explo- 94. Li P, Dunham Y, Carey S. Of substance: the nature
rations in Cognitive Diversity by Stephen C. Levinson. of language effects on entity construal. Cogn Psychol
Human Develop 2005, 48:309–314. 2009, 58:487–524.
82. Levinson SC. Frames of reference and Molyneux’s 95. Freyd JF, Pantzer TM, Cheng JL. Representing statics
question: cross-linguistic evidence. In: Bloom P, as forces in equilibrium. J Exper Psychol: Gen 1988,
Peterson M, Nadel L, Garrett M, eds. Language and 117:395–407.
Space. Cambridge, MA: MIT Press; 1996, 109–169.
96. Holmes KJ, Wolff P. Simulation from schematics: dor-
83. Taylor HA, Tversky B. Perspective in spatial descrip- sal stream processing and the perception of implied
tions. J Mem Lang 1996, 35:371–391. motion. In: Catrambone R, Ohlsson S, eds. Proceed-
84. Li P, Abarbanell L, Papafragou A. Spatial reasoning ings of the 32nd Annual Conference of the Cognitive
skills in Tenejapan Mayans. In: Bara BG, Barsalou LW, Science Society. Austin, TX: Cognitive Science Society;
Bucciarelli M, eds. Proceedings of the 27th Annual 2010.

264  2010 Jo h n Wiley & So n s, L td. Vo lu me 2, May/Ju n e 2011

 WIREs Cognitive Science Linguistic relativity

FURTHER READING
Boroditsky L. Does language shape thought? Mandarin and English speakers’ conceptions of time. Cogn Psychol 2001,
43:1–22.
Boroditsky L. Linguistic relativity. In: Nadel L, ed. Encyclopedia of Cognitive Science. London: MacMillan Press; 2003,
917–921.
Carruthers P. The cognitive functions of language. Behav Brain Sci 2002, 25:657–674.
Casasanto D. Who’s afraid of the Big Bad Whorf? Cross-linguistic differences in temporal language and thought. Lang Learn
2008, 58:63–79.
Hermer-Vazquez L, Spelke ES, Katsnelson A. Sources of flexibility in human cognition: dual-task studies of space and
language. Cogn Psychol 1999, 39:3–36.
Lucy J, Shweder R. Whorf and his critics: linguistic and nonlinguistic influences on color memory. Am Anthropol 1979,
81:581–615.
Schooler JW, Engstler-Schooler TY. Verbal overshadowing of visual memories: some things are better left unsaid. Cogn
Psychol 1990, 22:36–71.
Wolff P, Jeon G, Yu L. Causal agents in English, Korean, and Chinese: the role of internal and external causation. Lang
Cogn 2009, 2:165–194.

Vo lu me 2, May/Ju n e 2011  2010 Jo h n Wiley & So n s, L td. 265