Lateralized Whorf Experiment

Lateralized Whorf Experiment Abstract The Sapir-Whorf Hypothesis (Sapir and Whorf, 1952) states that our reality is shaped by the language we speak. Gilbert et al. (2008) conducted several experiments investigating the effects of visual field (VF) and lexical-category terms on reaction time (RT). Here, we replicated Gilbert et al. (2008) Experiment 2, demonstrating that between-category pairs for lexical terms are easier to discriminate between than within-category pairs. We also find a right-visual field (RVF) advantage for lexical-category discrimination. Demonstrating the Whorf effect in the interaction between lexical and perceptual categories as well as a lateralized Whorf effects due to left hemisphere language coding.

Introduction

The Sapir-Whorf Hypothesis (also known as ‘Linguistic Relativity) proposed by Sapir and Whorf (1952) states that our reality is shaped by the language we speak. Much research has gone into studying the effects of colour terminology on perceptual discrimination of colour categories (Winawer et al, 2007; Drivinkou et al. 2006). Gilbert et al. (2008) was the first to conduct a lateralized Whorf experiment outside the domain of colour discrimination, instead conducting three experiments testing animal lexical-category effects on perception and the interaction between category-discrimination affects within the left-visual field (LVF) and (RVF). In our study, we replicated Gilbert et al. (2008) Experiment 2.

Gilbert et al. (2008) conducted the first study that investigated Whorfian effects in a domain other than colour discrimination but since then, there have been investigations into many areas including time duration and gender. Byland and Athanasopoulos (2017) found that time-metaphors in Spanish and Swedish (either quantity-time or length-time metaphors) influenced speaker perception of time duration. Segel & Boroditsky (2011) found that in 78% of cases, personified depictions in art could be predicted by grammatical gender (e.g. ‘sin’ is feminine in German and was personified as a woman).

One such study conducted by Winawer et al (2007) nicknamed the ‘Russian Blues’ study looked at the effects of colour-terms on colour discrimination in Russian and English native speakers. Russian has two colour terms for diving the blue area of the colour spectrum, ‘siniy’ (dark blue) and ‘golouby’ (light blue) whilst English only has one. Their colour-discrimination resulted in faster RTs when colours fell into different lexical categories (‘siniy’ and ‘goluboy’) as opposed to the same lexical category (‘goluboy’ and ‘goluboy’). This was further demonstrated how such affects disappear when using a verbal interference task which disrupted the recall of lexical terms for colours. Demonstrating a Whorfian affect between lexical-categories and category-perception.

Gilbert et al. (2006) conducted a study into lateralized colour discrimination. They used a stimulus of twelve colours displayed in a ring in which 11 were the same (distractors) and 1 was different (target). Participants were asked to identify which side of the VF the target appeared when the stimulus briefly flashed on a screen and the target being able to occupy any of the twelve positions. RVF appearing targets yielded a faster RT when the target and distractor colours had different names. However, LVF appearing target resulted in no RT difference for similar or dissimilar colour names. She concluded that the Whorfian effect is present but affects only the RVF.

Gilbert et al. (2008) conducted three experiments investigating the interaction between VF and lexical category-pairs. In a ‘Naming Pre-Task’, participants were prompted to identify an image as either a ‘cat’ or a ‘dog’. In a ‘Visual-Search Task’, a stimulus image of 12 dogs and cats arranged in a ring were briefly shown on a screen; 11 were the same (distractors) whilst one was different (target) and could occupy any position. Participants were asked to identify whether the target appeared in the LVF or the RVF. Within the pair-type condition there were between category trials (‘cat-amongst-dogs’ or ‘dog-amongst-cats’) and within-category trials (‘cat amongst-cats’ and ‘dog-amongst-dogs’). Experiment 1, ‘Naming Pre-Task’ and ‘Visual-Search. Experiment 2, an unprompted ‘Naming Pre-Task’ to test whether participants referred to images by basic category names (‘dog’ and not. breed). Also, the ‘Visual-Search task’ was accompanied by Verbal/Non-verbal interference conditions to occupy working memory. In Experiment 3, Callosotomy patients were used to replicate Experiment 1’s conditions.

We replicated Gilbert et al. (2008)’s Experiment 2 but excluded the interference conditions, only used 2 target-distractor-pairs (‘dog-amongst-dogs’ and ‘cat-amongst-dogs’) and included the prompted ‘Naming Pre-Task’. In Gilbert et al. (2008)’s Experiment 2, eleven right-handed (RH) native English speakers with normal to correct-to-normal vision were recruited. There were three conditions that followed after each ring-stimulus: no-interference (black screen), verbal interference (colour words), and non-verbal interference (grid). Gilbert et al. (2008) found a significant three-way interaction between category-pair type, VF, and interference condition. Their results demonstrated a between-category advantage for the RVF although also present in a lesser state in the LVF. Results also showed a diminished performance for within-category pairs in the RVF compared to the LVF. Their results did not show a significant difference between the RVF & LVF for between-category pairs in the verbal-interference condition whilst the non-verbal condition presented the same results as the no-interference condition.

We hypothesized that the RVF would yield better (faster) discrimination of category-pairs than the LVF, and that the ‘between-category’ pairs would be discriminated better than the ‘within category’ pairs.

Method

Participants

We recruited a volunteer sample of 10 participants and sent all participants a unique URL which sent them to the ‘Gorilla’ experiment page. Participants signed a consent form then completed a demographics questionnaire (see ‘Materials’ for questions). We removed participants who scored below 75% (12/16) on the ‘Pre-learning Naming Task’ as this demonstrated inability to correctly label the ‘cat’ and a ‘dog’. This left us with 7 participants who ranged in age from 19 21 years, five were native speakers of English and the remaining two were proficient L2 English speakers, the ratio of left-hand to right-hand dominancy was 2:5 , the ratio of foot dominancy was 3 left-footed: 3 right-footed: 1 saying they “don’t know”, all participants have an A Level qualification or above.

Materials

This experiment was conducted online using ‘Gorilla Experiment Building’ software which recorded demographic, score, and RT data. We used a demographic questionnaire which asked participant age, Gender, dominancy of hands and feet, most recent educational qualification, current study level, native language, knowledge of other languages and to rate their fluency on a scale of 1-10 (1 = very low – 10 = perfect), whether participants regarded themselves as bilingual (if ‘yes’, then stated strongest language). We also used different image stimuli shown below, including an image of a cat and dog ‘Pre-task’ and a ‘within-category condition’ and ‘between category condition’ image.

Procedure

The experiment was conducted online using ‘Gorilla’ and all participants were sent a unique URL link in which to access the testing page. The experiment consisted of two parts: a ‘Naming Pre-Task’ and a ‘Visual Search Task’. Participants were instructed to complete both parts as fast as possible.

Naming Pre-task Learning Phase. Participants were instructed they would see images appear briefly at the centre of the screen and they had to identify whether the image was a cat or a dog by clicking the corresponding word ‘cat’ or ‘dog’ with their mouse. Participants were not told that the images would appear in random order and that the corresponding words would switch the side of the screen they appeared.

The ‘Pre-task’ consisted of 16 images of a cat or a dog, which appeared fleetingly and in random order. We excluded data from three participants who scored below 75% (12/16), which demonstrated ‘in theory’ they could not discriminate between a ‘cat’ and a ‘dog’ and thus could not participate in the study.

Visual Search Task. In the second part of the experiment, participants were told that they would see an image with 12 pictures arranged in a circle. One of the 12 would be different (target) whilst the other 11 (distractors) would remain the same. The participants had to identify which side of the screen the target image appeared as quickly as possible. Participants were instructed to press the ‘Q’ key if the target appeared on the left and the ‘P’ key if the target appeared on the right. They were then told to hit ‘spacebar’ if they wanted to see some examples.

Participants were then shown two examples of a possible image: a ‘cat-amongst-dogs’ (between-category condition) and a ‘dog-amongst-dogs’ (within-category condition). Participants were told that the target could appear on either side of the screen in any of the 12 possible positions and were reminded that ‘Q’ corresponded to left and that ‘P’ corresponded to right. It was emphasised to participants that the image would appear very briefly.

In total, participants completed 192 trials separated with three breaks in which participants could rest before choosing when to continue. The images appeared in random order. Gorilla aggregated demographic, score, and RTs data. Any participants with a RT over 2000ms or under 100ms were excluded. No participants were excluded from this part of the experiment.

Results

We analysed RTs by conducting a 2 (VF: RVF vs. LVF) X 2 (pair type: within-category vs. between-category) ANOVA using a significance of 0.05. The ANOVA found that there was not a statistically significant interaction effect between VF x pair-type [F(1,24) = 0.140, P= .711]. ANOVA also found no statistically significant effects between the RTs of category pairs [F = .022 < F Critical = 4.25, P = 0.882] or between the LVF and RVF [F=.631 < F critical = 4.26, P=.434]. Therefore, we accept the null hypotheses that 1) there is no discrimination difference between category-pair trials, and 2) that there is no difference between discrimination between VFs. However, due to our small sample size of 7 participants it is likely that this ANOVA is unreliable thus we chose to ignore it.

The interaction between VF and category-pair can be seen on Table 1. The mean difference between ‘within-category’ and ‘between-category’ pair RTs for the LVF was 30.24ms and 84.12ms for the RVF. This demonstrates that there was greater discrimination between targets appearing in the RVF than the LVF. Mean RTs were 37.68ms slower for the ‘between-category’ pair when the target appeared on the LVF rather than the RVF, demonstrating a RVF advantage for lexical-term discrimination.

We then calculated the total mean RTs for both the ‘within-category’ and ‘between category’ pairs when the target appeared in the LVF and again for the RVF. The LVF time was 904.2ms whilst for the RVF was 893.46ms, making RTs to LVF targets 10.74ms slower than for the RVF. Therefore, we can accept our hypothesis that the RVF had a larger discrimination advantage than the LVF. Also, the mean RTs for the ‘between-category’ pairs for each VF was faster than their ‘within-category’ counterpart. Mean RT for LVF ‘within-category’ to between category’ pair was 919.317ms to 889.081ms whilst for RVF this was 935ms to 851ms. Therefore, we accept our hypothesis that there is a difference between category-pairs (and that this discrimination was better in the ‘between-category’ pair - Basic Whorf Effect).

Gilbert et al (2008) also found RT was slower for between-category targets when they appeared in the LVF. This demonstrates that ‘between-category’ pairs were easier to discriminate than ‘within-category’ pairs.

Extraneous Variables

Two extraneous variables may have influenced our results, which are the dominance of the hands and feet and native language.

Dominancy of Hands and Feet. We found a clear difference for hand and foot dominancy on RT (Table 2). Shows the total mean Left-handed to right-handed RTs differed by 63.40ms whilst mean RTs for left-footed to right-footed participants differed by 108.72ms. This demonstrates that the mean RTs for both left-handed and left-footed participants were much slower than their right dominant counterparts.

Native Language. The results demonstrate that the mean slowest RT across all categories was for English Native speakers at 954.53ms whilst the fastest occurred for the Telegu native speaker at 719.37ms. Moreover, the results show that more English native speakers scored a slower rather than a faster RT than the non-native speakers.

Discussion

In response to Whorfian experiments only exploring colour perception, Gilbert et al. (2008) conducted a study investigating the interaction between lexical-category stimuli and VF. An unexpected result of our study was that native speakers (L1) of English had slower RTs than English second language speakers (L2). A study by Alsheri (2022) found that L1 Arabic speakers who were highly proficient L2 English speakers displayed faster RTs when identifying some semantically similar words. As L2 English speakers in my study rated their English highly (Telegu = 8/10, Malayalam speaker wrote that their strongest language was English), it is possible that the L2 speakers were faster lexical-category discriminators for the words ‘cat’ and ‘dog’. Further research with a larger sample size of English L1 & L2 speakers would be needed to determine any affect within this domain.

Overall, our results demonstrates that RTs were faster for perception of ‘between-category’ pairs than ‘within-category’ pairs and that ‘between-category’ pairs were better discriminated when displayed in the RVF than the LVF. This supported Gilbert et al. (2008)’s findings as well as our hypotheses. Drivinkou et al. (2006) explains this ‘category effect’ as the result of most individuals possessing left-hemisphere language lateralisation. As our visual fields are contralaterally processed, RVF stimuli are processed in the Left-hemisphere; consequently, discrimination of lexical categories can happen faster as signals do not have to undergo transcallosal transfer (Drivinkou et al., 2006).

Despite this, the RVF heralded weaker RTs for ‘within-category’ pairs (Drivinkou et al., 2006); this is likely due to participants first recalling the basic lexical terms at which point any disadvantage of the LVF is effectively diminished by the time-taken.

Another difference between our replication and Gilbert et al. (2008)’s Experiment 2 is that our study included Left-handed participants. This was concerning as this could act as an extraneous variable on results as left-handed individuals are more likely than right-handed to have language lateralized in the Right-hemisphere (Knecht et al., 2000). However, the percentage that left handedness relates to right-hemispheric lateralisation of language is only 27% whilst Left hemisphere lateralization is 73% (Knecht et al., 2000). Therefore, it is not unexpected that our results display a RVF advantage for Left-handed and right-handed participants.

Despite the likelihood that our right-handed participants had left-hemisphere language lateralization, RTs in the LVF were not overly different from the RVF. Drivinkou et al. (2006) explains that it is possible that the right-hemisphere possesses an ability for non-lexical categorical distinction possibly possessing universal distinctions. Or that it is a result of transcallosal transfer (Gilbert et al., 2008).

Future research could investigate the interaction between category-pair discrimination within VFs and foot-dominancy in comparison to handedness. Ward and Wyler (as cited in Elias and Bryden, 1998), found that right-hemispheric language lateralization occurred more frequently with left-footedness than left-handedness.

References

Alshehri, A. A. (2022). The L1 semantic retrieval of L2 words: Evidence from advanced L2 learners’ reaction times. World Journal of English Language, 12(1), 334. doi:10.5430/wjel.v12n1p334

Bryden, M. P., & Elias, L. J. (1998). Footedness is a better predictor of language lateralisation than handedness. Laterality: Asymmetries of Body, Brain and Cognition, 3(1), 41 52. doi:10.1080/713754287

Bylund, E., & Athanasopoulos, P. (2017). The Whorfian Time Warp: Representing duration through the language hourglass. Journal of Experimental Psychology: General, 146(7), 911-916. doi:10.1037/xge0000314

Drivinkou, G. V., Kay, P., Reigeier, R., Ivry, R. B., Gilbert, A. L., Franklin, A., & Davies, I. R. L. (2006). Further evidence that Whorfian effects are stronger in the right visual field than the left. PNAS, 104(3), 1097-1102. doi: 10,1073/pnas.0610132104

Fontenot, D. J. & Benton, A. L. (1972). Perception of direction in the right and left visual fields. Neuropsychologia, 10(4), 447-452. doi: 10.1016/0028-3932(72)90007-3

Gilbert, A. L., Regier, T., Kay, P., & Ivry, R. B. (2006). Whorf hypothesis is supported in the right visual field but not the left. Proceedings of the National Academy of Sciences, 103(2), 489-494. doi:10.1073/pnas.0509868103

Gilbert, A. L., Regier, T., Kay, P., & Ivry, R. B. (2008). Support for lateralization of the Whorf effect beyond the realm of color discrimination. Brain and Language, 105, 91-98.

Knecht, S., Dräger, B., Deppe, M., Bobe, L., Lohmann, H., Flöel, A., Ringelstein, E.-B., & Henningsen, H. (2000). Handedness and hemispheric language dominance in healthy humans. Brain, 123(12), 2512-2518. doi: 10.1093/brain/123.12.2512

Segel, E., & Boroditsky, L. (2011). Grammar in art. Frontiers in Psychology, 244.

Whorf, B. L. (1952). Language, mind, and reality. ETC: A review of general semantics, 167-188.

Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R., & Boroditsky, L. (2007). Russian blues reveal effects of language on color discrimination. Proceedings of the national academy of sciences, 104(19), 7780-7785