This section is comprised of reviews of literature that specifically focused on researching linguistic relativity or research areas that could be attributed to linguistic relativity. The most significant article reviewed is “Language and Prejudice: Direct and Moderated Effects” by Katherine A. Collins and Richard Clément, which reviews various studies across disciplines itself to summarise the current understanding of studies on language and prejudice. By reviewing the following articles, we were able to find theoretical authority and arguments on the linguistic relativity thesis as it relates to other phenomena - in particular PC language and culture which has been included in the scope of various studies. Whilst through the reviews we convey a deeper understanding of what research across various fields have found on linguistic relativity, language and prejudice, and the effectiveness of PC culture, it still remains inconclusive whether or not PC language or verbal hygiene effectively achieves its aim in counteracting prejudice.
Brown begins the article by stating the Spair-Whor hypothesis of linguistic determinism: that language is a determinant of perception and thought. Although this theory has largely been denounced and dismissed by academics, giving way to the rise of linguistic relativity (specifically its soft version), Brown adopts the then prominent theory in order to explore the more subtle phenomena of how grammatical features might affect our cognition. Instead of trying to elaborately apply and prove the strong theory of linguistic determinism through this study, Brown focuses on trying to detect and illuminate the nuanced effects of our linguistic design in our everyday thought. Specifically, Brown focuses on addressing the question of how allocation of words to one or another part of speech affects our thought. By "part of speech", Brown is referring to the semantic categories in grammar in which words are grouped by their grammatical function and classification. Such categories are familiar to most of us; classifications like nouns, verbs, adjectives and so on. Brown sets out that whilst these "parts of speech" are functional and descriptive, they also have an inevitable semantic characterisation. Whilst our language can be seen as a descriptive system, those descriptive elements do take on a prescriptive element - influencing the language and our thought in feedback. Brown points out that whilst descriptivist seek to find categories with no semantic attributes (pure abstractions in a sense), in all languages, there can never be such categories or classes without "exceptionless validity".
Brown's experiment has young children as participants. The children are shown different illustrations of ambiguous and odd objects, such as a stripped shallow container filled with seemingly red confetti with a pair of hands conducting a sewing like motion over them. Brown assumes that the children do not yet have vocabulary or the knowledge to identify or describe what they see in the illustration - hence the images being foreign and new to them. He introduces three made-up words and uses them to describe the illustrations as if they were real words. For example, he comes up with the word "sib" and uses it to describe the motion that the hand is doing, explaining to the children "this is a picture of sibbing". He finds that the children are quick to adopt the made-up words and apply them into different "parts of speech", being able to incorporate them into their grammar and using them in the way the researcher present them. So "sibbing" is quickly identified as used as a verb, and other made-up nouns are well appropriated into their grammar - making a fine distinction between mass nouns and singular/plural nouns. Our grammar makes a distinction between mass nouns and nouns that can be singular and plural. Brown notes that this must have some subtle impact on the way we think about people, objects, and beings in reality. Mass nouns carry the semantic characterisation of being in-differentiable, continuous, and holistically singular. Whilst we now can appreciate that water is a complex mechanical collection of H2O molecules and other ionised particles, we still think of it as a whole and largely singular substance. We do not differentiate between bodies of water, and they are in our cognition, much of the same. Brown concludes that the findings of his study shows that young native English speakers use "part of speech" membership of new vocabulary to interpret the meaning of the reality around them. The semantic distinctiveness of a word being a noun, verb, adjective or other class of words, affects the way the conceptualise and characterise the real referent of the symbol (word).
In this psychological study, Steele & Aronson sets out to examine stereotype threat amongst Africa-American university students. Stereotype threat is defined as an individual's fear of the risk that they will confirm or fulfil a negative stereotype about a social group they belong to through their behaviour. Steele & Aronson generally hypothesised that the black students would only perform worse than white students of equal or similar SAT scores not because of a difference in disposition of intelligence or ability, but simply because of being hindered by stereotype threat.
The researchers conducted four separate experiments in order to explore stereotype threat - each involving black and white students as participants. Study and 2 focused on the stereotype vulnerability of the black participants taking a difficult verbal test by varying whether or not the test was reflective of their intellectual ability. In one condition, the researchers told the participants the test was diagnostic of ability, and the other they simply explained it as a laboratory problem-solving task which did not assess their ability. The researchers assumed that this will either trigger their racial stereotypes on their intellectual ability or not depending on the seemingly diagnostic nature of the test. The first test strictly assessed performance results, whilst the second measured the anxiety of the students immediately after the test and compared them to their performance. The black students who were told the test was diagnostic of their ability performed the worst out of all groups in all conditions. However, when the data was processed, it was concluded that the black students did not underperform to the white students in the non-diagnostic condition. Anxiety also correlated with the test performances, supporting the notion of stereotype threat having a significant effect on their performances.
Study 3 and 4 explored the stereotype threat further by requiring students to indicate their race on the tests. Study 3 found that when the test was said to be diagnostic of ability, 75% of black students did not indicate their race. This demonstrated that they wished to distinguish themselves from the stereotype. Study 4 was able to demonstrate that the mere priming about race was enough to hinder the black students performance, even when the test was said to not be reflective of their ability.
Thus the findings are that when black students are not reminded about their race, or that the test is diagnostic of ability, they perform equally to white students of similar ability and background. However, when they are primed about their race, told that the test is diagnostic of ability, or both, they significantly underperform compared to white students in the same condition. Hence stereotype threat has the debilitating effect on the intellectual performance of individuals by making them anxious about conforming to negative stereotypes about their social group through poor performance. Study 1 and 3 demonstrated to some extent that the stereotype threat can actually motivate the black students to try to perform beyond their typical range of ability in order to distinguish themselves from the stereotype or prove it wrong.
Collins and Clement begin the study by stating that in this globalised era, there is a wide-spread discrimination and prejudice caused by constant sociopolitical frictions and conflicts, and therefore a study on language and prejudice is significant. The researchers examine various empirical evidence linking language and prejudice and attempt to offer taxonomic classification for such studies. They summarise that prejudice has an implicit and explicit effect on language, and that language itself plays a critical role in all forms of prejudice. Therefore, because language and prejudice is inextricably linked, a study of prejudice without consideration of language is incomplete.
The researchers elaborate the connection between language and prejudice by discussing two fundamental links. The first is that language can be a tool for prejudice. In a study by Augoustinos & Every (2007), it was demonstrated that language is a powerful force in perpetuating prejudice. Moreover, they found that despite the increase in anti-prejudice norms such as the PC movement or verbal hygiene movement, and the decrease in overtly prejudicial expressions, prejudice continues to persist in language in more subtle forms.
Secondly, language is an embodiment of culture and can be an anti-prejudicial tool. Prejudice is a part of our culture and language embodies culture in general. Rubenfeld et al. (2007) found that second language acquisition lead to learning a second culture as well, and thus forming an identity associated with the second culture. Hence language acquisition can be an anti-prejudicial tool. In a similar vein, Wright & Bongie (2007) found that bilingual education can reduce prejudice.
The overall study by Collins and Clement aims to expand the different conceptualisations of language and prejudice, place relevant research into a coherent framework, identify current trends and directions for future research, and provide conclusions on role of language in prejudice. The ambitious research takes inspiration from Taylor & Usborne (2007) who wrote an article in the 30th anniversary edition of the Journal of Language and Social Psychology on the issue of what the study of language has contributed to the understanding of prejudice.
Language and prejudice intersect largely in three ways: 1. language is the verbal expression of prejudice, 2. different prejudicial expressions have different connotations and activate different perceptions, and 3. language and prejudice are moderated by third factors, such as social norms, within a communication context. The researchers adopt Sutton's 2010 model on the functions of language, in which he conceptualises language as a vessel, barometer, and a lens, as metaphors to understand its functions. The researchers develop on this model by slightly redefining them and adding a fourth metaphor. Language as a vessel is explained as language functioning as a tool to hold and convey our thoughts. Language can also be a barometer when it acts as a gauge to the speaker's attitudes and beliefs. Language is also a lens as it directs and distorts our cognition through its transformative power on attention, thought, and memory. Lastly, the researchers add that language can also be a sign - language itself can be a factor of prejudice, e.g. accents.
The researchers explain that explicit expressions of prejudice and its direct effects are rarely studied, and also rather irrelevant given the current social norm which largely condemn such explicit expression of prejudice. Instead they largely focus on implicit expressions of prejudice in language.
Language can reflect our implicit beliefs, called linguistic bias. Such phenomena typically involve description of stereotype consistent and stereotype inconsistent behaviour. From research on this behaviour, the Stereotype Consistency Bias ("SCB"), a tendency for more stereotype consistent information to be transmitted, was found. In addition to this, research found a Linguistic Expectancy Bias ("LEB") and Linguistic Intergroup Bias ("LIB"), which are defined through the use of linguistic abstraction.
LEB is the tendency to describe stereotype consistent behaviour more abstractly than stereotype-inconsistent behaviour. LIB is the tendency for desirable behaviour of the in-group members and undesirable behaviour by out-group members to be described more abstractly than undesirable behaviour of in-group members and desirable behaviour of out-group members. In other words, LEB is the phenomenon of generalising stereotypical behaviour, whilst isolating stereotype-inconsistent behaviour as a situational phenomenon. LIB is the general bias to deem desirable behaviour as a general trait of the in-group, whilst generalising and totalising undesirable behaviour of out-group members as a characteristic of the out-group. This is coupled with the isolation of undesirable behaviour of the out-group members to be specific to the individuals as exception to the trend, and desirable behaviour of out-group members to also be exceptions to the out-groups general characteristics. Howard & Rothbart (1980) found that stereotypes are negatively distorted for the out-group. Thus linguistic biases are used as an indicator of prejudiced beliefs held by the speaker. Beukeboom, Finkenauer, & Wigboldus (2010) found that when stereotype-inconsistent behaviour must be communicated, speakers will use subtle linguistic techniques, such as negations, to describe the behaviour. For instance, if a group member from a stereotypically unintelligent group performs a studious behaviour, it is more likely that the behaviour will be described using a negation. (e.g. "their performance wasn't bad") These linguistic biases are subtle, but have great consequences, and implies the prejudice of the speaker.
Language also acts as a lens that distorts the recipient's perceptions to be more similar to the presumed perception of the speaker. Negations for example, will lead the recipient to believe that the speaker has opposite expectations for the group being described. Due to this function of language acting as a lens, Goodman et al. (2008) argue merit in campaigns for positive and inclusive language use to reduce negative evaluation of social groups based on stereotypes and prejudice.
Nevertheless some research reviewed suggests that linguistic biases may not contribute to the maintenance of stereotypes. Geschke et al. (2007) found that concrete descriptions of stereotype-inconsistent behaviour has greater impact on the recipient's perception than abstract descriptions. This suggests that explicit evidence may be necessary to support stereotype-inconsistent behaviour. It may also suggest that concrete descriptions may have a greater short-term effect, but abstract descriptions have a more significant long-term effect. It is not clear how linguistic biases perpetuate stereotype through these mechanisms.
Through their connotations, words can distort recipients' perceptions in ways that are not necessarily similar to the perception (or intentions) of speakers. Most research in this regard has been on gender biases in language. A clear finding has been that mental representations of gender is formed through an interaction of semantic information and stereotypes. Words which are intended to be gender neutral (grammatically do not denote gender) can still activate gender stereotypes. (Cacciari & Padovanic, 2007). Generic masculine words, such as chairman or fireman, which are used conventionally and not intended to be prejudices can also be interpreted as excluding females. (Stout & Dasgupta, 2011).
Grammar is a feature of language which affects our perception. For instance, grammatical comparisons set X as a basis for Y to be compared to. This establishes X as an implicit norm even if it is unintended to be. The grammatical structure of "Y compares to X" focuses on Y, and its differences having to be articulated and explained. Bruckmüller and Abele (2010) found that the implicit norm group is perceived as more powerful and of higher status than the compared group. The typical member of the implicit norm group is perceived as more powerful, agentic, and less communal. In other words, members of the implicit norm group are seen as individuals, rather than part of the collective group, and therefore perceived as having more agency. It is also the linguistic convention to compare the smaller group to the larger group, and such linguistic practices perpetuate the status quo of intergroup power relations.
Douglass & Sutton (2010) found that speakers who use abstract language to describe positive behaviour and concrete language to describe negative behaviour were perceived as more likeable than speakers who abstracted negative behaviour and situated positive behaviour. However, Czopp (2008) found that speakers who expressed positive stereotypes were less liked and perceived as more biased, and less qualified for a diversity job than those who did not express positive stereotypes. A subtle distinction between the two studies to reconcile their findings may be that whilst abstracting positive behaviour of an individual to their disposition is well-received, abstracting it to their membership to a group has the opposite effect. In other words, describing someone as being fiscally responsible because they are disciplined and well educated in finance and economics is positively perceived, but relating the same behaviour to the stereotype of Jewish people because the person is Jewish is condemned. Research in this area collectively indicate that language cannot be a neutral vessel to imply convey our cognition, but has an unyielding transformative effect on our perception.
Collins and Clement discuss several research that deals with language as a sign, the fourth metaphor they added to their reform of the Sutton (2010) model of functions of language. They find that peripheral nonstructural aspects of language can influence perception by allowing recipients to make inferences about speakers' group membership. By peripheral nonstructural aspects, they refer to features of language that do not relate to the properties of the word of language content, such as accents. Hosoda et al. (2007) found that speakers who have different accents to that of the recipients' were evaluated more negatively than speakers who had the same accent (perceived to have no accent) to that of the recipients'. Thus, language acts as a sign of group membership of the speaker.
This feature of language as a sign of group membership may have been evolutionarily advantageous as it allows for a more accurate in-group and out-group distinction than other factors such as race or culture. Kinzeler & Spelke (2011) found that language preferences for native speakers of the recipient's in-group develops earlier than racial preferences in children. When out-groups were generally more hostile and dangerous for majority of human history, this would have been a survival mechanism to distinguish ally from foe, given that people living in close geographical proximity spoke the same or mutually intelligible languages, but differed in accents or dialect.
Due to this function of language being a divisive measure in establishing one's group membership, there is a strong interconnection between language and identity. Thus language can play a central role in definition of group boundaries. Collins and Clement suggest three contextual factors moderating the relationship of language and prejudice: group membership, communication strategy, and social norms.
Group membership of a speaker, through stereotyping, affect how the same words are interpreted by the same group of recipients. Maoz (2009) found that a proposal to compromise made by females was favoured over the same proposal by males. Maoz attributed this difference to the stereotypes of the two genders, with females seen as more compassionate, warm, and males to be more aggressive and dominant. Collins et al. (2009) found that when recipients were asked to interpret descriptions of academic records back into objective academic scores, translators decoded more favourable descriptions into weaker academic performances when the students were purported to be black. The recipients were perceiving the positive remarks to indicate that the performance of the student was good given the stereotype of their group they belonged to. Reitsma-van Rooijen et al. (2007) found that participants who were exposed to subtly biased language against them perceived more interpersonal distance with the speaker than participants who were exposed to language biased towards them. Language is not only acting as a sign of group-membership, but a tool to suggest social distance between interlocutors.
Ladegaard (2011) found that communication strategies speakers used in intergroup communication affected the language used, or the style in which it was used. Speakers from different groups pursue different impression management goals. Amongst several research reviewed, it was found that for instance, Whites tried to appear as non-racists and strive to appear moral, unbiased, and try to be liked by their recipient. Minorities on the other hand, tried to avoid being negatively evaluated, and strived to appear competent and receive support from their recipients. However, some research suggests that having a communication agenda may have the opposite effect to what is intended, or be ineffective in achieving the agenda. (e.g. Penner et al., 2010)
Social norms affect language use by triggering self-censorship in speakers when in the presence of another social group or called out for their linguistic biases. Conway et al. (2009) found that participants would avoid speaking explicitly negatively about a fraternity when asked to describe a fraternity to a group of students with a fraternity member present. However when asked to talk about the same topic with a friend, they would not self-censor their language. Mallet & Wagner (2011) found that when male participants were confronted for their sexist linguistic bias by a female confederate, they would continue the conversation with more self-censorship than male participants who were not confronted in their conversation. In this case, self-censorship lead to a more positive interaction. Hence, the research on self-censorship is not clear on whether or not it is effective in reducing prejudice.
The researchers conclude that language bridges the cognitive and social worlds and is the primary means for prejudice to be explicitly or implicitly communicated. Language reflects thought, and is intentionally used as an instrument to affect the external world by influencing other's cognition, and language itself is shaped by the external world. The researchers find that language 'may act as a vessel that conveys cognition, a barometer that reveals cognition, a lens that distorts cognition, and a sign that focuses cognition.'