Racism, linguistics and the law –

Your new post is loading...

Scooped by Charles Tiayon

July 9, 2020 11:22 PM

Scoop.it!

From www.oxfordstudent.com - July 9, 2020 12:30 PM

Charles Tiayon's insight:

CW: Discussion of racism and the murder of Trayvon Martin.

If you want to know what white privilege is, here’s a story that perfectly exemplifies it: the dismissal of Rachel Jeantel’s testimony in court. On February 26, 2012, Trayvon Martin was chased down and shot by George Zimmerman. Initially let free, Zimmerman was eventually put on trial after public pressure—however, the jury acquitted him of second-degree murder and manslaughter in July 2013. The acquittal was largely based on the jury simply not understanding nor believing ‘star witness’ Rachel Jeantel.

Jeantel was so-called because she was phoning Trayvon as Zimmerman chased him and until moments before his death. It was, as Rickford and King state, “akin to Trayvon’s being in the courtroom himself, testifying on his own behalf.”[1] The case blew up all over social media and the phrase Black Lives Matter was in fact coined in response to the systematic racism involved in Zimmerman’s acquittal. The reality is that Zimmerman walks free today because racism led the jury to not believe Jeantel.

That conclusion is so powerful. It’s a stark indictment of the legal system, and of society. Ever since I read the article, Language and Linguistics on Trial,[1] I haven’t been able to get it out of my head. Because what it shows is the unbelievable extent of white privilege—an extent I didn’t fully comprehend. The authors—Rickford and King—outline how one of the many basic privileges denied to people is the privilege of just being understood.

The jury, the lawyers, the media—all of them said they could not understand Jeantel. But this paper’s detailed analysis of the situation yields a blunt conclusion: the physical factors (sound production, loudness) and the linguistic factors (grammar, vocabulary, dialect) would never have undermined Jeantel’s intelligibility. Ingrained racism denies people the simple right of being understood.

What on earth do I mean by that? There has been considerable research demonstrating that factors like race, ethnicity, geography and even social status introduce biases in listeners’ attitudes.[2] Even when the speech recorded is exactly the same, subjects report lower intelligibility when shown a picture of an ethnic minority speaker. Yes, you read that right: when the same speech sample is presented to people, the simple fact that a black person said it means it’s less well understood.

One of the many basic privileges denied to people is the privilege of just being understood.

What’s so shocking is that this isn’t a case of wrongly reporting intelligibility on the basis of race. The subjects aren’t purposefully racist and hence lying. They are being genuine—they genuinely struggle to understand people purely because of the colour of their skin. Indeed, this is true of a number of other factors, including country of origin, ethnicity, and status. It is not about accent, or language, or voice quality, or vocabulary, or anything else. This is about race.

Even as a linguistics student, I still can’t get my head around this. I imagine that for many this does not come as a shock, but my privilege and ignorance, I am ashamed to admit, has made this seem almost impossible to me. For this reason, I’ve chosen to summarise the key points of the 40-ish page article in a way which is accessible to those fortunate enough not to have endured Chomsky’s page-turning Syntactic Structure:

Scientific racism was the historical ‘scientific’ ‘evidence’ (it was neither scientific nor evidential) used to justify racism, slavery and colonialism. Consider Charles White, a so-called ‘Enlightenment thinker’: White argued that white and black people were two different species— an idea called polygeny. So much for being enlightened…

Others, like Alexander Thomas and Samuell Sillen, defended slavery on the basis that black people were ideal to be enslaved because of their “primitive psychological organisation.” Biological racism was probably the most virulent and common form of scientific racism for a long time. Sadly, it continues to this day—although, thankfully, with significantly less scientific support.

So where does scientific racism enter the picture today? One of the more subtle and nefarious forms of scientific racism involves the belief that languages such as African-American Vernacular English (AAVE)—a dialect of English—are somehow inferior to English. The ‘reasoning’ for this is varied: such languages apparently ‘lack grammar’, are ‘corrupt’, ‘lack sophistication’, are ‘less complex’, etc.

However, the field of linguistics has systematically shown, time and time again, that AAVE—along with all languages spoken in African and African-American communities—is highly sophisticated and follows strict grammatical rules. Despite this, the belief that such languages are deficient is pervasive and continue to entrench racism today.

Consider the fact that, in 2007, a population-wide survey found that 70% of people in Jamaica favoured a bilingual education in Jamaican and English.[3] Yet, utterly false views about Jamaican being ‘bad’ have continued to be a hindrance to achieving this necessary education. Introducing bilingual teaching would benefit vast numbers of children and allow them the education they have a right to. But pervasive racist views stop that.

Scientific racism involves embedding pseudoscientific ideas in theories which are hard to disprove, or that oppressed communities cannot disprove, in order to justify racism.

Consider when one linguist, Le Page, went into a Jamaican school in 1968 to talk about the need for students to study the grammar of the languages wrongly portrayed as ‘bad’. The headmaster introduced the linguist by saying:

‘Ladies, there is no doubt that we must eradicate the bad talk of our children, and the answer to this problem is more and more attention to the grammar—they must learn the grammar, there is no substitute for the grammar, and Professor Le Page is here to say something to us about this.'[4]

Le Page had come to talk about grammar to discuss the legitimacy of Jamaican. The headmaster thought he was there to delegitimise it on that very basis. The look on the headmaster’s face after the talk was notably different.

Scientific racism involves embedding pseudoscientific ideas in theories which are hard to disprove, or that oppressed communities cannot disprove, in order to justify racism. Toni Morrison summed it up perfectly when she said that, “the very serious function of racism is distraction… Somebody says you have no language, so you spend twenty years proving that you do. Somebody says your head isn’t shaped properly, so you have scientists working on the fact that it is. Someone says you have no art, so you dredge that up. Somebody says you have no kingdoms, so you dredge that up. None of that is necessary. There will always be one more thing.”[5]

So back to Rachel Jeantel. Firstly, as Rickford, King and many others have shown time and time again: AAVE is a highly sophisticated dialect of English. It should be noted that there are genuine distinctions between AAVE and English, so intelligibility issues may genuinely have been introduced. But, again, it was racism that stopped this problem from being resolved.

Utterly false assumptions that Jeantel’s speech was somehow ‘bad’ or ‘unsophisticated’ led people to dismiss her, rather than organise court transcriptions or even an interpreter. Lisa Bloom, a lawyer and TV legal analyst who was present for the whole trial, reported that “Jeantel spoke an urban teenaged lingo that was an alien tongue to most.”[6] Had the ‘star witness’ been someone speaking French, there would be no doubt that an interpreter and probably even a transcript would have been provided. Structural racism is pervasive.

“When the speaker is represented… as Asian rather than Caucasian, he is rated as being much more accented and harder to comprehend, even though raters are evaluating the same native-speaker.”

The gaslighting that followed involved scapegoating factors that seemed reasonable to blame. Many people agreed that Jeantel spoke with an unintelligible lisp despite linguistic analysis finding that this would have had an insignificant effect on understanding. There was a strong feeling that the acoustics in the room were awful and Jeantel was a quiet speaker. However, acoustic analysis disavowed this reasoning.

The reality is that everyone has been fed slight and subtly racist tropes and assumptions throughout our lives. This means that we have hard-to-prove racist habits. For example, one researcher conducted an experiment perfectly summarised by the conclusion; “when the speaker is represented… as Asian rather than Caucasian, he is rated as being much more accented and harder to comprehend, even though raters are evaluating the same native-speaker accent in each case.”[7][8]

What we must understand is that the subjects of these experiments aren’t ‘the bad ones’ or ‘the racist ones’. They are all of us. In 1973, one researcher found that “teachers rated the same Standard English audio track as sounding much more “nonethnic-standard” and somewhat more “confident-eager” when associated with a video image of a White child than with a video image of either a Black or Mexican American child.”[9][10] These attitudes are common in our education system, in our legal system, in all of society. That’s what structural racism means.

“teachers rated the same Standard English audio track as sounding much more “nonethnic-standard” and somewhat more “confident-eager” when associated with a video image of a White child than with a video image of either a Black or Mexican American child.”

Jeantel’s case was without a doubt worsened by the fact that she was a woman who was ‘dark-skinned, and overweight.’ Racism is one of the worst manifestations of this linguistics problem, but there are so many other social prejudices where it reveals itself; it is an intersectional problem. One linguistics study concluded that “the suspect was rated as significantly more guilty when he employed a Birmingham rather than a standard accent, and that attributions of guilt were significantly associated with the suspects’ superiority and social attractiveness.”[11]

One final study, just to hammer the point home. (If you want to read more about the evidence as a whole, read pages 30-32 of Rickford and King, below; it’s staggering.) Frumkin concluded on the basis of evidence from *undergraduates* that all other things being equal, “the eyewitness who stated that she was German was perceived as more credible than the Mexican eyewitness, who in turn was perceived as more credible than was the Lebanese eyewitness.”[12] This was in spite of Frumkin eliminating any accent distinction. This is a race issue.

The reality is that, as Bloom summarises, “Jeantel’s speech patterns because they are associated with poor African Americans, were perceived by many, including the people who mattered most, the jurors, as unintelligent, and worse, evidence that she was not credible.”[13]

What’s so problematic is that, in individual circumstances, when someone says ‘I’m not being listened to because of systematic discrimination’, such testimony is so often dismissed. The accusation is waved away as a figment of someone’s imagination, or as a result of some supposedly scientific reason: a lisp, voice quality, sound level, acoustics, accent… The list goes on and on.

There will always be another scapegoat for racism, and we cannot expect individuals to have the resources to unearth scientific reports—or worse, to conduct these experiments themselves.

We must take accusations of racism and other forms of discrimination seriously, because we cannot place the burden on an individual to produce that evidence. Fortunately, scientific evidence has unambiguously refuted biological and linguistic racism. But as Toni Morrison summarised, “there will always be one more thing.” There will always be another scapegoat for racism, and we cannot expect individuals to have the resources to unearth scientific reports—or worse, to conduct these experiments themselves – what a failure our legal system would then be. It is our moral responsibility to take these claims seriously.

Moreover, we must take on collective responsibility for this sort of structural, ingrained racism. There are a lot of people who are theoretically on board with tackling racism but see no problem with what they perceive as mere slights. They mock ‘microaggressions’, decolonising curricula and taking down statues. The effect of these small things is huge: the black community isn’t taken as seriously, listened to, or understood—even in court.

As Lippi-Green points out, these negative stereotypes are perpetuated by news and entertainment media—the mocking crows in Dumbo, the mean hyenas in The Lion King— instilling an association between dialect and “trifling, bullying, unsavoury characters.”[14][15] No one wants to destroy children’s imaginations, but we need to think carefully about the effect of feeding them the racist associations promoted by Disney films. It’s one of the reasons blackface is so nefarious.

Jeantel internalised what is a societal problem as a personal flaw.

I think one of the saddest things about the entire case is that, when asked by Piers Morgan why the public mocked her, Jeantel said “I have an underbite… The words I say, I can’t, it can’t come out right.” Jeantel blamed her own lisp as the reason Zimmerman went free. Ultimately, though, specialists have concluded that the underbite was not the cause. This shows the severe gaslighting black communities face in all areas of life. Because Jeantel was not able to rightly point the blame at prevalent racist attitudes. She couldn’t blame the structural discrimination that led to the jury not taking her seriously, not believing her, and not understanding her. Instead, she internalised what is a societal problem as a personal flaw. I find that to be particularly poignant.

Image Credit: Samilustrando @ pixabay

Sources

[1] This article is informed greatly by the fantastic article by Rickford and King which you can access here: https://muse.jhu.edu/article/641206/pdf?casa_token=QfhVLGaxMEYAAAAA:2puO5jgabS8qoejQmGDd7IEwGKXzE9wcn_OBBOG-7mrbT-ewkoumXlzZ0P6ND7VSljjB-H2rbGg

[2] Lindemann & Subtirelu. 2013. Reliably biased: The role of listener expectation in the perception of second language speech. Language Learning.

[3] Devonish, Hubert & Carpenter. 2007. Towards full bilingualism in education: The Jamaican bilingual primary education project. Social and Economic Studies. p301.

[4] Le Page. 1968. Problems to be faced in the use of English as the medium of education in four West Indian territories. Language problems of developing nations. pp437–38.

[5] Morrison, Toni. 1975. A humanist view. Paper presented at a conference on Black Studies at Portland State University. [You can listen to this here: https://soundcloud.com/portland-state-library/portland-state-black-studies-1]

[6] Bloom. 2014. Suspicion nation: The inside story of the Trayvon Martin injustice and why we continue to repeat it. p132.

[7] Rubin, D.L. 1992. Nonlanguage factors affecting undergraduates’ judgments of nonnative English-speaking teaching assistants. Research in Higher Education. p518.

[8] Quote from Rickford and King.

[9] Williams. 1973. Some research notes on dialect attitudes and stereotypes. Language attitudes: Current trends and prospects.

[10] Quote is a summary from Rickford et al. 2013, p291.

[11] Dixon, Mahoney & Cocks. 2002. Accents of guilt? Effects of regional accent, race, and crime type on attributions of guilt. Journal of Language and Social Psychology. p162.

[12] Frumkin. 2007. Influences of accent and ethnic background on perceptions of eyewitness testimony. Psychology, Crime & Law. p327.

[13] Bloom. 2014. Suspicion nation: The inside story of the Trayvon Martin injustice and why we continue to repeat it. p133.

[14] Lippi-Green 2012. English with an accent: Language, ideology and discrimination in the US. pp101-129.

[15] Quote from Rickford and King.

Liked reading this article? Sign up to our weekly mailing list to receive a summary of our best articles each week – click here to register

Want to contribute? Join our contributors group here or email us – click here for contact details

No comment yet.

As AI giants duel, the Global South builds its own brainpower - Features

Researchers across Africa, Asia and the Middle East are building their own language models designed for local tongues, cultural nuance and digital independence

"In a high-stakes artificial intelligence race between the United States and China, an equally transformative movement is taking shape elsewhere. From Cape Town to Bangalore, from Cairo to Riyadh, researchers, engineers and public institutions are building homegrown AI systems, models that speak not just in local languages, but with regional insight and cultural depth.

The dominant narrative in AI, particularly since the early 2020s, has focused on a handful of US-based companies like OpenAI with GPT, Google with Gemini, Meta’s LLaMa, Anthropic’s Claude. They vie to build ever larger and more capable models. Earlier in 2025, China’s DeepSeek, a Hangzhou-based startup, added a new twist by releasing large language models (LLMs) that rival their American counterparts, with a smaller computational demand. But increasingly, researchers across the Global South are challenging the notion that technological leadership in AI is the exclusive domain of these two superpowers.

Instead, scientists and institutions in countries like India, South Africa, Egypt and Saudi Arabia are rethinking the very premise of generative AI. Their focus is not on scaling up, but on scaling right, building models that work for local users, in their languages, and within their social and economic realities.

“How do we make sure that the entire planet benefits from AI?” asks Benjamin Rosman, a professor at the University of the Witwatersrand and a lead developer of InkubaLM, a generative model trained on five African languages. “I want more and more voices to be in the conversation”.

Beyond English, beyond Silicon Valley

Large language models work by training on massive troves of online text. While the latest versions of GPT, Gemini or LLaMa boast multilingual capabilities, the overwhelming presence of English-language material and Western cultural contexts in these datasets skews their outputs. For speakers of Hindi, Arabic, Swahili, Xhosa and countless other languages, that means AI systems may not only stumble over grammar and syntax, they can also miss the point entirely.

“In Indian languages, large models trained on English data just don’t perform well,” says Janki Nawale, a linguist at AI4Bharat, a lab at the Indian Institute of Technology Madras. “There are cultural nuances, dialectal variations, and even non-standard scripts that make translation and understanding difficult.” Nawale’s team builds supervised datasets and evaluation benchmarks for what specialists call “low resource” languages, those that lack robust digital corpora for machine learning.

It’s not just a question of grammar or vocabulary. “The meaning often lies in the implication,” says Vukosi Marivate, a professor of computer science at the University of Pretoria, in South Africa. “In isiXhosa, the words are one thing but what’s being implied is what really matters.” Marivate co-leads Masakhane NLP, a pan-African collective of AI researchers that recently developed AFROBENCH, a rigorous benchmark for evaluating how well large language models perform on 64 African languages across 15 tasks. The results, published in a preprint in March, revealed major gaps in performance between English and nearly all African languages, especially with open-source models.

Similar concerns arise in the Arabic-speaking world. “If English dominates the training process, the answers will be filtered through a Western lens rather than an Arab one,” says Mekki Habib, a robotics professor at the American University in Cairo. A 2024 preprint from the Tunisian AI firm Clusterlab finds that many multilingual models fail to capture Arabic’s syntactic complexity or cultural frames of reference, particularly in dialect-rich contexts.

Governments step in

For many countries in the Global South, the stakes are geopolitical as well as linguistic. Dependence on Western or Chinese AI infrastructure could mean diminished sovereignty over information, technology, and even national narratives. In response, governments are pouring resources into creating their own models.

Saudi Arabia’s national AI authority, SDAIA, has built ‘ALLaM,’ an Arabic-first model based on Meta’s LLaMa-2, enriched with more than 540 billion Arabic tokens. The United Arab Emirates has backed several initiatives, including ‘Jais,’ an open-source Arabic-English model built by MBZUAI in collaboration with US chipmaker Cerebras Systems and the Abu Dhabi firm Inception. Another UAE-backed project, Noor, focuses on educational and Islamic applications.

In Qatar, researchers at Hamad Bin Khalifa University, and the Qatar Computing Research Institute, have developed the Fanar platform and its LLMs Fanar Star and Fanar Prime. Trained on a trillion tokens of Arabic, English, and code, Fanar’s tokenization approach is specifically engineered to reflect Arabic’s rich morphology and syntax.

India has emerged as a major hub for AI localization. In 2024, the government launched BharatGen, a public-private initiative funded with 235 crore (€26 million) initiative aimed at building foundation models attuned to India’s vast linguistic and cultural diversity. The project is led by the Indian Institute of Technology in Bombay and also involves its sister organizations in Hyderabad, Mandi, Kanpur, Indore, and Madras. The programme’s first product, e-vikrAI, can generate product descriptions and pricing suggestions from images in various Indic languages. Startups like Ola-backed Krutrim and CoRover’s BharatGPT have jumped in, while Google’s Indian lab unveiled MuRIL, a language model trained exclusively on Indian languages. The Indian governments’ AI Mission has received more than180 proposals from local researchers and startups to build national-scale AI infrastructure and large language models, and the Bengaluru-based company, AI Sarvam, has been selected to build India’s first ‘sovereign’ LLM, expected to be fluent in various Indian languages.

In Africa, much of the energy comes from the ground up. Masakhane NLP and Deep Learning Indaba, a pan-African academic movement, have created a decentralized research culture across the continent. One notable offshoot, Johannesburg-based Lelapa AI, launched InkubaLM in September 2024. It’s a ‘small language model’ (SLM) focused on five African languages with broad reach: Swahili, Hausa, Yoruba, isiZulu and isiXhosa.

“With only 0.4 billion parameters, it performs comparably to much larger models,” says Rosman. The model’s compact size and efficiency are designed to meet Africa’s infrastructure constraints while serving real-world applications. Another African model is UlizaLlama, a 7-billion parameter model developed by the Kenyan foundation Jacaranda Health, to support new and expectant mothers with AI-driven support in Swahili, Hausa, Yoruba, Xhosa, and Zulu.

India’s research scene is similarly vibrant. The AI4Bharat laboratory at IIT Madras has just released IndicTrans2, that supports translation across all 22 scheduled Indian languages. Sarvam AI, another startup, released its first LLM last year to support 10 major Indian languages. And KissanAI, co-founded by Pratik Desai, develops generative AI tools to deliver agricultural advice to farmers in their native languages.

The data dilemma

Yet building LLMs for underrepresented languages poses enormous challenges. Chief among them is data scarcity. “Even Hindi datasets are tiny compared to English,” says Tapas Kumar Mishra, a professor at the National Institute of Technology, Rourkela in eastern India. “So, training models from scratch is unlikely to match English-based models in performance.”

Rosman agrees. “The big-data paradigm doesn’t work for African languages. We simply don’t have the volume.” His team is pioneering alternative approaches like the Esethu Framework, a protocol for ethically collecting speech datasets from native speakers and redistributing revenue back to further development of AI tools for under-resourced languages. The project’s pilot used read speech from isiXhosa speakers, complete with metadata, to build voice-based applications.

In Arab nations, similar work is underway. Clusterlab’s 101 Billion Arabic Words Dataset is the largest of its kind, meticulously extracted and cleaned from the web to support Arabic-first model training.

The cost of staying local

But for all the innovation, practical obstacles remain. “The return on investment is low,” says KissanAI’s Desai. “The market for regional language models is big, but those with purchasing power still work in English.” And while Western tech companies attract the best minds globally, including many Indian and African scientists, researchers at home often face limited funding, patchy computing infrastructure, and unclear legal frameworks around data and privacy.

“There’s still a lack of sustainable funding, a shortage of specialists, and insufficient integration with educational or public systems,” warns Habib, the Cairo-based professor. “All of this has to change.”

A different vision for AI

Despite the hurdles, what’s emerging is a distinct vision for AI in the Global South – one that favours practical impact over prestige, and community ownership over corporate secrecy.

“There’s more emphasis here on solving real problems for real people,” says Nawale of AI4Bharat. Rather than chasing benchmark scores, researchers are aiming for relevance: tools for farmers, students, and small business owners.

And openness matters. “Some companies claim to be open-source, but they only release the model weights, not the data,” Marivate says. “With InkubaLM, we release both. We want others to build on what we’ve done, to do it better.”

In a global contest often measured in teraflops and tokens, these efforts may seem modest. But for the billions who speak the world’s less-resourced languages, they represent a future in which AI doesn’t just speak to them, but with them."

Sibusiso Biyela, Amr Rageh and Shakoor Rather

20 May 2025

https://www.natureasia.com/en/nmiddleeast/article/10.1038/nmiddleeast.2025.65

#metaglossia_mundus

From www.natureasia.com - May 20, 2025 10:49 PM

Scoop.it!