Language & Thought
Week 2 – Background Reading – Chapter 2 – Linguistic Principles
1) Basic Grammatical Concepts
-languages differ in a variety of concepts -> e.g. emphasis on word order or affixes to convey meaning
=> HOWEVER: certain properties are owned by every human language:
1) duality of patterning -> at one level, there is a large number of meaningful elements (words)
=> at another level, there is a relatively small number of meaningless elements that are combined to form the words
all languages create words from small meaningless elements BUT phones, phonemes & distinctive features vary
across languages
*a phone is any distinct speech sound or gesture, regardless of whether the exact sound is critical to the meanings of
words -> In contrast, a phoneme is a speech sound in a given language that, if swapped with another phoneme,
could change one word to another (e.g. dig vs big) => THUS: phonemes are a special category of phones
*distinctive feature = characteristic of speech sounds whose presence/absence distinguishes the sound from others
=> distinctive feature theory: independent units (distinctive features) are combined to form phonemes & contrasts
between sounds are binary with the presence of each feature indicated by + and its absence by -
example units: +/- voicing; +/- bilabial (sound is articulated at the lips); +/- stop (airflow from the lungs is
completely stopped during production) OVERALL: distinctive features help in formulating linguistic rules (e.g. a
word cannot begin with two stop consonants)
2) morphology = smallest meaningful unit in a language -> can be free morphisms (e.g. bed & room in the word
“bedroom”) or bound/grammatical morphisms (“s” to indicate plural in rooms) => all languages possess a
morphological system BUT they differ from each other
3) phrase structure = division of sentence into groups of words/constituents -> “(The young swimmer) (accepted the
silver medal).” can be divided into a noun sentence (NS) and a verb sentence (VS) => each again consists of multiple
individual constituents (e.g. NS -> det. + adj. + N.) = generating phrase-structure rules (can become more specific)
OVERALL: PS-rules can rule out the problem of phrase-structure ambiguity (e.g. They are eating apples.), in which
a sentence can be grouped in more than 1 way phrase structure exists in all languages BUT differs across them
4) linguistic productivity/creativity = ability to create & comprehend novel utterances -> people can create an
infinite number of sentences, even though they can’t store an infinite sentence number
=> proposed explanation: we simply store linguistic rules for sentence production one of the PS-rules, which
allows for infinite sentence creation: recursion (embedding sentences into another)
2) Transformational Grammar
-grammar = description of a person’s linguistic knowledge (according to linguistic theory)
-> more specifically: grammar is a formal device to generate the infinite number of sentences making up a language
=> THUS: grammars are theories of language, composed of more specific hypotheses about the
structure/organization of some part of the language
-Chomsky’s 3 criteria for the evaluation of grammar as theories:
1) observational adequacy -> grammar must specify what is & what is not an acceptable sequence in the language at
several levels of language (e.g. syntax level & phonological level)
2) descriptive adequacy -> grammar must specify the relationships between various sequences in the language
=> example: grammar should explain why two sentences are similar in meaning but differ in syntax
3) explanatory adequacy -> grammar should explain the role of linguistic universals in language acquisition
=> idea: a child’s innate language mechanisms should help to identify the correct grammar when learning a language
,-theory of transformational grammar (Chomsky) -> sentences have a deep structure & a surface structure => the
deep structure is derived by a series of phrase-structure rules, and the surface structure is derived from the deep
structure by a series of transformational rules overall: transformational grammar can explain certain aspects of
language, such as deep-structure ambiguity, that cannot be accounted for entirely by phrase-structure rules
*distinction bt. surface structure & deep structure shows that sentences have more than 1 level of structure:
1) surface structure -> superficial arrangement of constituents & reflects order in which the words are pronounced
2) deep structure -> underlying structure of a sentence that conveys the meaning of a sentence
=> distinction is useful due to problem of deep-structure ambiguity (a single surface structure of a sentence can be
derived from different deep structures (i.e. can have different meanings))
surface & deep structure (2 levels) are required to adequately describe the relationships between sentences
(descriptive adequacy) 2 levels require a new set of rules: transformational rules
*derivation of a sentence in transformational grammar = 2-step process:
1) PS rules are used to generate the underlying tree structure (referred to as the deep structure)
2) transformational rules are applied to the deep structure & the intermediate structures (those between the deep
& surface structure), ultimately generating the surface structure
=> transformational rules apply to multiple constituents at once (transformation of constituents by adding, deleting
or moving them) 2 exemplary transformations:
1) particle-movement transformation -> particles can occur in different positions within a sentence without
changing its meaning => problem: PS rules would differ & therefore fail to adequately describe the similarity of the
sentences solution: apply the following transformation to provide descriptive adequacy:
V + particle + NP = V + NP + particle John phoned up the women. = John phoned the woman up.
2) passive transformation -> simplified: NP 1 + V + NP 2 = NP 2 + be + V + -en + by + NP 1
=> Clara plays the piano. = The piano is played by Clara.
NOTE: these transformational rules don’t apply under all circumstances -> e.g. P-M transformation doesn’t work with
pronouns => example: John phoned her up. (works) John phoned up her. (doesn’t work)
Week 2 – Background Reading – Chapter 14 – Language, Culture & Cognition
1) Whorf Hypothesis
-Whorf(-Sapir) hypothesis -> language unconsciously shapes thought patterns (worldviews) => consists of 2 parts:
1) linguistic determinism = language determines non-linguistic cognitive processes (changes the way people think)
2) linguistic relativity = the resulting cognitive thought processes vary from language to language
=> Whorf tried to prove his hypothesis by pointing out 2 types of linguistic examples:
*lexical examples: importance of the concept of differentiation (number of words in a given domain, e.g. colour)
-> languages differ in the domains that are most differentiated (e.g. Russian more differentiated in the colour domain
than English) => indicates cultural significance of that domain regular exposure to these words can then change
our thought patterns (linguistic determinism)
*grammatical examples: e.g. languages vary in whether using morphology/word order to signal meaning
-> these grammatical distinctions also influence our thought patterns
=> criticism to Whorf’s theory:
1) it didn’t conform to behaviourist perspective of cognition, which was common at that time
2) Whorf should have assessed cognition & linguistic features independently to prove that linguistic differences
cause cognitive alternations -> he mostly talked about linguistic features BUT didn’t explain cognitive consequences
2) Lexical Influences on Cognition
-any study that tests the Whorf hypothesis needs to define 3 key terms:
,1) define differences in language -> can be absence vs presence of a linguistic marker OR simply different ways of the
same linguistic marker acting in the 2 languages
2) define differences in thinking -> focus should lay on habitual thought (routine thinking)
3) What does language “determines” cognition mean? -> 2 ways of interpretation:
*strong version: language categories make up cognitive categories => distinguishing bird & airplane verbally allows
telling them apart cognitively criticized for assigning language too much importance
*weak version: presence of linguistic categories influences the ease (through availability) with which various
cognitive operations are performed => more accepted by researchers
COLOUR TERMS
*codability (length of a verbal expression for an event/object -> can be just 1 word or multiple words combined) is
negatively correlated with frequency of usage of that verbal expression => showed to apply to colours
**a hierarchy of basic colour terms was established, which was found in all languages
-> these colours were not restricted to a small number of objects & were not contained within other colours
=> black/white -> red -> yellow/green -> blue -> brown -> pink/orange/purple/grey
e.g. language with 4 colours would have black, white, red & either yellow or green
***subsequent study: memory for focal colours (most representative example of a basic colour -> e.g. most bluish
blue) is better than for non-focal colours => support against the Whorf hypothesis
reason: cognitive factors led to linguistic category formation
=> HOWEVER: study is heavily criticized for not controlling for colour discrimination & other studies indicate the
colour perception (cognition) depends on the terms we use to refer to these colours (language)
conclusion: there are universal constraints on colour categories (study 1), but linguistic differences within those
constraints affect colour cognition & perception (later studies) support for the weak Whorf hypothesis
NUMBER TERMS
-Asian languages use an easier system for number terms between 10 & 99 than the English language
-> Asian: 18 = ten eight; 35 = three ten five => most evidence reveals a mathematical advantage in that range of
numbers for Asian children over English-speaking children ADDITIONALLY: Asians seem to approach these
mathematical tasks slightly differently
OBJECT TERMS
-object discrimination is hypothesized to be related to language acquisition -> conceptual categories related to
object names are constructed at the time when we learn a language => THEREFORE: we should see different kinds of
early object terms in children acquiring different languages turns out to be true
*study: English stresses the use of nouns, whereas Asian languages use verbs more frequently
-> English children were found to be superior in categorization tasks & inferior in means-end abilities
=> conclusion: language contributes to the timing of some cognitive achievements (e.g. object discrimination)
SPATIAL TERMS
-Bowerman & Choi propose interactionist view: children’s early word meanings result from the interaction of
existing cognitive development & the semantic categories of the input language
*entire body of studies: language influences cognitive, spatial conceptualization of the world
-> nature of that effect: language experience appears to be to strengthen or weaken conceptual categories (weak
version), BUT not to create them (strong version)
**another body of studies: language also influences non-linguistic cognition
-> example: based on categorization of frames of references: absolute terms (north, west, south, east), relative
terms (to the speaker -> left, right…), intrinsic terms (relative to another object -> upon the hill) => Dutch people
(relative) organize objects from left to right & Tzeltal people (absolute) from north to south in a memory task
, 3) Grammatical Influences on Cognition
SUBJUNCTIVE
-English uses the subjunctive to express counterfactual reasoning (If John had come earlier, the group would have
been on time.) -> a similar grammatical form to the subjunctive is missing in the Chinese language
=> THEREFORE: Chinese people should struggle more with counterfactual reasoning
*evidence: contradicting evidence does not allow for clear conclusion
-> ADDITIONALLY: counterfactual reasoning is more specialized than habitual thought anyways
GRAMMATICAL MARKING OF FORM
-in Navaho, the form of the verb for handling an object varies with the form or shape of the object
-> example: verb varies if the object is a long flexible object (piece of string) versus a long rigid object (stick)
=> in English, the form/shape of the object doesn’t influence the verb to be used
THUS: Navaho-speaking children would learn to discriminate the forms of objects at an earlier age (cognition)
than their English-speaking peers study: turned out to be true => support for weak version of Whorf hypothesis
*NOTE: other environmental factors (e.g. sub-urban environment) led to the same effect in English children
-> language isn’t the only factor playing a role
GRAMMATICAL MARKING OF OBJECTS AND SUBSTANCES
-children can already distinguish between substance (mass noun) & object (count noun), before being capable of
performing the grammatical distinction -> HOWEVER: children are influenced by this distinction before actually
learning it => this influence of language then partially affects cognition
NOTE: another body of studies showed that the distinction bt. substance & object in adults affects cognition
GRAMMATICAL MARKING OF GENDER
-gender of a word in a particular language affects attributes assigned to this word by people speaking the language
-> THEREFORE: gender as part of human language influences cognition
OVERALL CONCLUSION: even though older studies on the Whorfian effect showed inconclusive/negative results,
more recent investigations mostly support the weak version of the Whorf hypothesis
Week 2 – Focus Reading – Articles
1) Russian Blues – Russian Reveals the Effects of Language on Colour Discrimination
INTRODUCTION + BASIC INFOS
-colour discrimination is differently emphasized across languages -> while English refers to a wide range of blue
tones by saying “blue”, Russian makes the obligatory differentiation between dark blue (siniy) & light blue (goluboy)
=> Whorf hypothesis: these linguistic differences should shape our cognition (colour discrimination ability)
-previous studies: cross-linguistic differences in subjective colour similarity judgments & colour confusability in
memory were found -> example: if 2 colours are called by the same name in a language, speakers of that language
will judge the 2 colours to be more similar and will be more likely to confuse them in memory (compared with
people whose language assigns different names to the 2 colours)
CONSIDERABLY: cross-linguistic differences in similarity judgments & recognition memory were found to be
eliminated by direct verbal interference or by indirectly preventing subjects from using their normal naming
strategies -> THEREFORE: linguistic representations are involved online in these kinds of colour judgments
-current problems: cross-linguistic comparisons have relied on memory procedures (1) or subjective judgment (2):
1) possibly, when subjects make perceptual discriminations among stimuli that can all be viewed at the same time,
language may have no influence
2) possibly, any language-congruent bias results from a conscious, strategic decision on the part of the subject
,=> dealt with in this study by measuring colour discrimination performance in 2 language groups in a simple,
objective, perceptual task
METHOD
-procedure: subjects were simultaneously shown three colour squares arranged in a triad & were asked to say which
of the bottom 2 colour squares was perceptually identical to the square on top
-> advantage: objective task with a clear correct answer (because unambiguous task) & minimal memory demand
(because the entire triad was visible until the subject answered)
-types of trials:
*for English speakers: all trials were considered “blue”
*for Russian speakers: in some cases, the match & the distractor square fell into the same Russian category of blue
(within-category discriminations), while in other cases, they differed (cross-category discriminations)
=> hypothesis 1: Russian speakers should make faster cross-category than within-category discriminations + for
English speakers the category distinction should not matter
ADDITIONALLY: experiment was performed in 3 conditions:
1) normal viewing, no-interference condition (no dual task)
2) subjects silently rehearsed digit strings while completing the colour discrimination trials (verbal task)
3) subjects maintained a spatial pattern in memory while completing colour discrimination trials (non-linguistic task)
=> hypothesis 2: if linguistic processes play an active, online role in perceptual tasks, then a verbal dual task, but not
a non-linguistic dual task, should diminish the goluboy/siniy category advantage found in Russian speakers
MOREOVER: another distinction was integrated by exposing subjects to 2 types of trials:
1) easy discriminations (colours are more dissimilar) & 2) tough discriminations
=> hypothesis 3: based on previous findings, a stronger effect on linguistic differences was expected to occur for
tougher discriminations than for easy discriminations
RESULTS
-actual colour boundaries between “light” blue & “dark” blue were almost identical for both languages
-Russian speakers showed a category advantage when tested without interference, whereas English speakers did not
-> this category advantage found for Russian speakers was disrupted by verbal, BUT NOT by spatial, interference
=> NOTEWORTHY: verbal interference even led to a disadvantage of cross-category, near-colour discriminations
-effects of language were only significant for difficult discriminations (near-colour comparisons) in Russian language
-English speakers did not show a category advantage in any condition
confirms all 3 hypotheses ADDITIONALLY: neither speed/accuracy trade-offs nor overall differences between
the languages in reaction time could explain the results (can be ruled out as confounders)
DISCUSSION
-conclusions:
1) categories in language can affect performance of basic perceptual colour discrimination tasks
-> colour discrimination performance differs across languages as a function of what perceptual distinctions are
habitually (linguistically forced) made in specific languages => THUS: Whorfian effect relates to simple cognitive tasks
2) the effect of language is online, because it is disrupted by verbal interference (but not by spatial interference)
-proposed mechanism for this Whorfian effect:
-> in the process of arriving at perceptual decisions, language-specific distortions in perceptual perform. arise as a
function of interaction of lower-level perceptual processing & higher-level knowledge systems (e.g. language) online
=> language-specific representations seem to be brought online spontaneously during even rather simple
perceptual discriminations, BUT access to them is denied through any verbal interference task
, 2) Colour Naming Across Languages Reflects Colour Use
INTRODUCTION + BASIC INFOS
-question whether colour categories are universal OR culturally relativistic remains unclear
-> distinction between warm & cool colours (reflects 1st stage of basic colour hierarchy: black & white)
=> study goal: question whether they are universal categories will be dealt with in this article
METHOD
-colour-labelling data from 2 extreme versions of the world colour survey (WCS) was obtained:
1) participants had completely free choice about how they wanted to name a specific colour (chip)
2) participants had a fixed choice from the most common words used in the “free choice” version
-information theory: colour naming can be better understood by considering informativeness
-> THUS: study can be seen as a communication game => procedure:
1) speaker uses word (w) (depending on WCS version) to describe the specific colour chip c, which he/she has in mind
2) the listener, then, has to correctly guess the colour chip based on knowing the word => he/she can chose a set of
colour chips & determines the number of colours included in the set each guess is followed by a “yes” OR “no”
3) the listener keeps on guessing until the right chip is identified
-whole experiment was conducted in with 3 different languages:
1) American English speakers; 2) Bolivian Spanish speakers; 3) Bolivian speakers of Tsimane (rare language)
-measures:
1) average number of guesses a listener would take to identify the exact colour chip = listener’s average surprisal
-> indicates communication efficiency
2) sum of the scores for all words w that might have been used to label c = surprisal score for each colour chip c
=> information-theoretic measure (average surprisal) allows us to rank the colours (c) for their relative
communicative efficiency within a language
3) mean of the average surprisal scores across all colour chips within a language = colour communication efficiency
of a specific language -> allows to compare the languages’ colour communication efficiency
RESULTS (IN EXAM -> one of the figures/tables will be shown & interpretation is asked for)
-Tsimane speakers showed greater colour-naming variability (compared to the other 2 languages) = greater
variability in what colour terms were used for all colour chips -> exception: red + white & black
supported by the fact: Tsimane speakers showed more uncertainty in colour naming of familiar characteristics
associated with low-consensus colours (e.g. yellow, blue, green…) BUT this RT difference to the other 2 languages
disappeared for the high-consensus colours across all languages (white, black & red)
-nevertheless: all 3 languages have a comparable colour space => gives validity to the experiment
-greater colour-naming variability of Tsimane resulted in higher average surprisal for all colour chips (compared to
the other 2 languages) -> indicates lower informativeness of Tsimane => NOTE: results were very similar in both
versions of the WCS: similar informativeness of each of the 3 languages, even though average surprisal was lower in
the fixed version (unsurprisingly) contradicts widespread criticism about the methodology of the WCS
-Figure 3A: there are more words in English than in Tsimane BUT the overall agreement on these colours is higher
(indicated by higher communication efficiency)
-only red & yellow focal colours (best representative colour chips of a particular colour term) had low average
surprisal scores in all 3 languages -> blue & green did not => only red & yellow can be considered as “unique hues”
(irreducible primary colours, which cannot be described using any more primitive colours)
, supports growing body of evidence: warm colours (red & yellow) are associated with higher communicative
efficiency than cool colours (blue & green) overall results: despite overall gross differences in the communication
efficiency across languages, among chromatic chips, warm colours are always the easiest to communicate precisely
REMARKABLY: this relationship was found to be true across the entire WCS of 110 languages
DISCUSSION
-2 early conclusions:
1) all languages, even those with very few consensus colour terms, have a comprehensive colour lexicon
2) all languages, even those with a very sophisticated colour language, prioritize the same set of (warm) colours
-> i.e. they associate higher communicate efficiency with these (warm) colours
=> shows that there are some culturally lexical constraints on colour naming, BUT culture doesn’t affect
communicative efficiency of colour terms
-hypothesized reason for warm colours being universally associated with higher communication efficiency:
-> relevant, salient objects in nature & society have warm colours more often & background features typically
possess cool colours => confirmed by additional investigations supports communication-efficiency hypothesis:
usefulness is the reason why languages acquire a particular colour name CONSEQUENTLY: relatively low
communicative efficiency of colour naming among Tsimane’ suggests that colour is not that useful for them
IMPORTANTLY: this is evidence against the Whorf hypothesis -> we build colour terms around their use
-underlying idea: exposure to artificially coloured objects promotes the usefulness of colour for object identification
-> fact: Tsimane live in a natural, non-industrialized environment so that colours don’t play too much of a role
-example for the communication efficiency hypothesis: there simply are not that many natural blue objects, which
explains why many languages acquire the term “blue” relatively late & don’t distinguish between many blue tones
3) Language Can Boost Otherwise Unseen Objects into Visual Awareness
INTRODUCTION & BASIC INFOS
-modularity thesis -> language & vision are independent modules
=> processes in one domain do not influence processes in the other
HOWEVER: growing body of evidence: nonvisual factors can affect the content of visual perception (what one sees)
-> this study: shows the effect of auditory linguistic labels on visual perception
=> example: an otherwise invisible kangaroo can be boosted into visual awareness by language
-unconfirmed idea: “higher” - level processes (e.g. word-recognition) can influence “lower” - level processes
-> consistent with interactive-activation accounts: neural processing is intrinsically interactive
=> hypothesis: perception is modulated by top-down cues more broadly & language more specifically
-this study: top-down influences on visual awareness are investigated by suppressing explicit visual awareness &
semantic processing (bottom-up processes) -> concrete solution: creating interocular rivalry
=> mechanism: by presenting different stimuli to the 2 eyes, the stimuli compete & this retinal input can give rise to
2 different conscious percepts, which are proven to not have received semantic processing
method is called continuous flash suppression (CFS)
METHOD
-CFS was used to create interocular rivalry -> an object was placed in interocular competition with high-contrast
noise patterns that alternate at∼10 Hz => suppresses the stimulus from awareness for extended durations
3 experiments were conducted:
*first 2 experiments tested whether hearing a verbal cue can make an otherwise invisible image (suppressed through
CFS) visible => visual detection was measured (using explicit self-reports indicating awareness)
, hypothesis: if processing verbal labels activates perceptual representations that contribute to awareness (top-
down process), then hearing a label before viewing a picture made invisible through CFS should change the
likelihood of actually seeing the picture MORE SPECIFICALLY:
1) valid labels (“kangaroo” preceding a picture of a kangaroo) should increase the likelihood of seeing the picture
2) invalid labels (“kangaroo” preceding a picture of a car) should decrease this likelihood
*3rd experiment examined whether the effectiveness of labels depends on the degree of correspondence between
the shape denoted by the label & visual shape presented to the participants
hypothesis: labels would progressively improve visual detection as the correspondence between the label and the
target’s shape increased, and decrease detection rates as the label progressively mismatched the target
RESULTS
EXPERIMENT 1
-effects of cue type (valid, invalid, absent) on detection & recognition of suppressed visual objects were analysed by
including another IV (object-presence: absent, present) & by investigating multiple dependent variables: signal
detection sensitivity (d -> required detection accuracy as hit rate & false alarms) & correct-trial RTs
=> NOTE: both IVs are within-subjects factor
-results:
1) verbal cues didn’t affect false alarms
2) verbal cues affected hit rate -> valid cues had significantly higher hit rate than both invalid & no cues
=> HOWEVER: invalid cues had a slightly lower hit rate than no cues BUT this difference was insignificant
3) verbal cues affected signal detection sensitivity -> highest to lowest from valid to invalid cues BUT only this
particular difference was significant
4) verbal cues affected median RTs -> lower RT for valid cues compared to both invalid & no cues
=> HOWEVER: invalid cues had a slightly longer RT than no cues BUT this difference was insignificant
ADDITIONALLY: object recognition (NOT ONLY detection) was also boosted by verbal cues
-> participants were significantly more likely to answer “yes” to the question whether they had seen a zebra when
having heard “zebra” verbally before (compared to invalid & no cue trials)
=> explanation: bias to interpret an otherwise indeterminate image as matching a previously heard label
-problem with study 1: CFS allows for practice effects (leading to increased accuracy on later trials) especially if the
same objects are used over and over again -> was the case in study 1 (only 8 different images & 8 labels)
EXPERIMENT 2
-dealt with study 1’s problem by using more complex & diverse images
-> results: very similar to those for experiment 1 except for considerably higher overall performance (presumably
due to longer display time (6 s vs. 1.5 s)) => rules out that cueing effect results from practice
-another finding: valid cues reduced the dependence of detection on physical properties such as size/density of the
image, which, without valid cues, are highly predictive of suppression efficacy
=> OVERALL: experiments 1 & 2 showed that verbal labels affect detection of suppressed visual stimuli
EXPERIMENT 3
-goal: testing whether the detection effect occurred because of labels activating visual features with which they are
associated -> procedure: we parametrically varied the overlap between 1) the shapes associated with the label &
2) the shape actually shown, using the words “square” and “circle” + using visual stimuli that varied on a square-to-
circle continuum => results:
1) verbal cues led to significantly higher hit rates overall (confirms study 1 & 2)
2) hearing the word “square” tended to increase hit rates for the more squared stimuli, whereas hearing “circle”
tended to increase hit rates for the more circular shapes
,3) as the intermediate stimuli possess features from both circles & squares, hearing either “circle” or “square”
improved object detection (compared to no verbal cues)
4) very similar results were found for signal detection sensitivity
=> OVERALL: suggests that the effect size of object recognition depends strongly on the match between the shape of
the seen image & the associated shape of the verbal clue
DISCUSSION
-conclusion: language-based activation of visual representations can act as a top-down “boost” to perception that,
under some circumstances, can bring an otherwise invisible image into awareness -> on the contrary, activation of
“wrong/invalid” visual representation can hinder object detection => NOTEWORTHY: the effect’s strength is
positively correlated with the degree of match between 1) the shape of the seen image & 2) the associated shape of
the verbal clue
-proposed top-down mechanism -> an object name initiates feedback activity to object-selective regions of cortex
(e.g. inferotemporal cortex), producing a predictive signal or “head start” to the visual system
=> weak bottom-up signal combined with this boosted top-down signal produced by the label may be sufficient to
bring the percept into awareness (interactive-activation accounts)
4) Turning the Tables: Language & Spatial Reasoning
INTRODUCTION
-possibility: spatial linguistic categories are simply put into our head to map the natural conceptual space
-> i.e. humans invented linguistic terms that labelled their concepts => can’t account for language differences
contrary idea: Sapir-Whorf hypothesis: linguistic differences cause cultural differences in cognition (e.g. spatial
reasoning) NOTE: strong version of this hypothesis has been disproven early on
-more recent studies: there are universal linguistic-spatial categories BUT they leave room for differences
-> support for weak version of Sapir-Whorf hypothesis: linguistic differences (among other factors) cause
differences in cognition (e.g. spatial reasoning)
=> example: arrays are differently remembered by languages with varying frame of references
-fact so far: linguistic differences correlate with spatial reasoning strategies
-> goal of this paper: clarifying directionality by providing new evidence
SUMMARY OF PREVIOUS FINDINGS
-spatial orientations can be expressed in different frames of references:
-> absolute/allocentric terms (externally referenced), relative/egocentric terms (to the speaker -> left, right…)
=> NOTE: allocentric classification is subdivided into systems that refer to the intrinsic properties of external objects
(the front of the house) and those that refer to local (“near the tool shed”) or more global landmarks and regions
(“in Cleveland”, “east”) NOTE: languages vary in what frame of reference they primarily use
-Levinson et al. found that languages vary in what frame of reference they tend to use (allocentric, egocentric, mix)
-> conducted another experiment afterwards: asked native speakers of 3 different languages to memorize, recall,
and reconstruct spatial arrays after being spatially reoriented => animals in a row test
*description: panel 1 (memorizing situation); panel 2
(recall situation); panel 3 (2 possible outcomes depending
on which spatial reasoning strategy had been used)
a) subject used allocentric frame of reference & ordered
them from north to south -> THUS: 180-degree rotation
did not impact the final order
b) subject used egocentric frame of reference & ordered
them from left to right -> makes a difference then
, => results: language-prominent frame of reference was used to complete the task e.g. the Dutch used relative
frame of reference (b), as commonly done in Dutch
*problem: these findings still do not allow for causation, because the effect could still go either way
SPATIAL REASONING IN VARYING FRAMES OF REFERENCE: AN EXPERIMENTAL REVIEW
-2 experimental versions were conducted -> idea: manipulating the environment in which one group (English-
speakers) solves a spatial rotation problem & see whether this affects the spatial-reasoning strategy to be used
*rationale: if the same SR-strategies are used in both environments, then language affects spatial reasoning
HOWEVER: if different environments would make the English subjects adapt their SR-strategies to those of other
languages (environment of these languages is enforced in experiments), then local environment seems to affect SR
-initial study (experiment 1) with only 8 English participants was conducted & confirmed that English speakers
understand to speak in absolute terms BUT heavily prefer using relative SR-strategies
-in experiment 2, the participants had to perform the animals in a row task 5 times (with changing 3 (out of 5)
animals & order) => context in which task was performed varied between experiment 2 A & B by adding implicit
landmark or bearing cues of various kinds, both on and outside the experimental table tops
EXPERIMENT 2 A -> landmarks in the reference world beyond the table top (Blinds-Down/Blinds-Up; Outdoors)
-method -> 40 English speakers performed the animals in a row task in 3 conditions (between-subjects factor):
1) 10 indoor (without window view) => tested in a small, featureless room except for a window (placed in the same
way relative to the table as the Maya subjects sat relative to a house in the Levinson study) => blinds were pulled
down so that they could not see what lay behind the window (resembles Dutch condition from Levinson study)
2) 10 indoor (window view) => same room BUT with blinds up so that students could see their university library
3) 20 outdoor => tested outdoors in a grassy area on campus
-results:
1) blinds-down group behaved just like Dutch group in Levinson study (mostly only relative SR-strategy)
2) blind-up & outdoor group displayed a U-shape (about 40% each for only relative & only absolute + 20% mix)
-> pattern lays somewhere in between the Maya & Dutch language from the Levinson study
3) differences: outdoor group displayed significantly differed SR-approach than blind-down group & the difference
between the 2 indoor groups was almost significant (0.056 with a very small sample)
4) additional finding: blinds-down group asked significantly less for clarification as to how to organize the objects
compared to the blinds-up group
-discussion:
1) Dutch-resembling condition (blinds-down) replicated the results from the Levinson study -> mostly relative SR
=> further supported by subjects’ lower frequency of commenting on the ambiguity of the task
2) more landmark cues (in other 2 conditions) led to more ambiguity & participants engaged in both SR-approaches
=> presence of environmental cues weakened the bias of the English subjects toward their own language’s SR-
approach, as compared with that bias for subjects who were tested in the featureless condition
arising question: “Can landmark information if it is salient enough more completely determine the degree to
which a single population solves spatial problems from an egocentric versus allocentric perspective?” exp. 2b
EXPERIMENT 2 B -> landmark cues in table top space (Absolute/Relative Ducks)
-goal: examining the effect of absolutely vs relatively placed landmarks
-method -> 20 new subjects each per condition (2) performed animals in a row task (in a blinds-up design) with a
little toy being placed on the right on the stimulus table (extra environmental cue):
1) absolute bias -> little toy was still placed on the right after rotation
2) relative bias -> little toy was placed on the left after rotation
