Category: Analysis

Source: The Conversation – Global Perspectives – By Amra Lee, PhD candidate in Protection of Civilians, Australian National University

Israel partially lifted its aid blockade of Gaza this week in response to intensifying international pressure over the man-made famine in the devastated coastal strip.

The United Arab Emirates and Jordan airdropped 25 tonnes of food and humanitarian supplies on Sunday. Israel has further announced daily pauses in its military strikes on Gaza and the opening of humanitarian corridors to facilitate UN aid deliveries.

Israel reports it has permitted 70 trucks per day into the strip since May 19. This is well below the 500–600 trucks required per day, according to the United Nations.

The UN emergency relief chief, Tom Fletcher, has characterised the next few days as “make or break” for humanitarian agencies trying to reach more than two million Gazans facing “famine-like conditions”.

A third of Gazans have gone without food for several days and 90,000 women and children now require urgent care for acute malnutrition. Local health authorities have reported 147 deaths from starvation so far, 80% of whom are children.

Israeli Prime Minister Benjamin Netanyahu has claimed – without any evidence – “there is no starvation in Gaza”. This claim has been rejected by world leaders, including Netanyahu ally US President Donald Trump.

Famine expert Alex de Waal has called the famine in Gaza without precedent:

[…] there’s no case of such minutely engineered, closely monitored, precisely designed mass starvation of a population as is happening in Gaza today.

While the UN has welcomed the partial lifting of the blockade, the current aid being allowed into Gaza will not be enough to avert a wider catastrophe, due to the severity and depth of hunger in Gaza and the health needs of the people.

According to the UN World Food Programme, which has enough food stockpiled to feed all of Gaza for three months, only one thing will work:

An agreed ceasefire is the only way to reach everyone.

Airdrops a ‘distraction and a smokescreen’

Air-dropping food supplies is considered a last resort due to the undignified and unsafe manner in which the aid is delivered.

The UN has already reported civilians being injured when packages have fallen on tents.

The Global Protection Cluster, a network of non-governmental organisations and UN agencies, shared a story from a mother in Al Karama, east of Gaza City, whose home was hit by an airdropped pallet, causing the roof to collapse:

Immediately following the impact, a group of people armed with knives rushed towards the house, while the mother locked herself and her children in the remaining room to protect her family. They did not receive any assistance and are fearful for their safety.

Air-dropped pallets of food are also inefficient compared with what can be delivered by road.

One truck can carry up to 20 tonnes of supplies. Trucks can also reach Gaza quickly if they are allowed to cross at the scale required. Aid agencies have repeatedly said they have the necessary aid and personnel sitting just one hour away at the border.

Given how ineffective the air drops have been – and will continue to be – the head of the UN Relief Works Agency (UNRWA) for Palestine has called them a “distraction” and a “smokescreen”.

Malnourished women and children need specialised care

De Waal has also made clear how starvation differs from other war crimes – it takes weeks of denying aid for starvation to take hold.

For the 90,000 acutely malnourished women and children who require specialised and supplementary feeding, in addition to medical care, the type of food being air-dropped into Gaza will not help them. Malnourished children require nutritional screening and access to fortified pastes and baby food.

Gaza’s decimated health system is also not able to treat severely malnourished women and children, who are at risk of “refeeding syndrome” when they are provided with nutrients again. This can trigger a fatal metabolic response.

Gaza will take generations to heal from the long-term impacts of mass starvation. Malnourished children suffer lifelong cognitive and physical effects that can then be passed on to future generations.

What needs to happen now

The UN has characterised the limited reopening of aid deliveries to Gaza as a potential “lifeline”, if it’s upheld and expanded.

According to Ciaran Donnelly from the International Rescue Committee, what’s needed is “tragically simple”: Israel must fully open the Gaza borders to allow aid and humanitarian personnel to flood in.

Israel must also guarantee safe conditions for the dignified distribution of aid that reaches everyone, including women, children, the elderly and people with disabilities. The level of hunger and insecurity mean these groups are at high risk of exclusion.

The people of Gaza have the world’s attention – for now. They have endured increasingly dehumanising conditions – including the risk of being shot trying to access aid – under the cover of war for more than 21 months.

Two leading Israeli human rights organisations have just publicly called Israel’s war on Gaza “a genocide”. This builds on mounting evidence compiled by the UN and other experts that supports the same conclusion, triggering the duty under international law for all states to act to prevent genocide.

These obligations require more than words – states must exercise their full diplomatic leverage to pressure Israel to let aid in at the scale required to avert famine. States must also pressure Israel to extend its military pauses into the only durable solution – a permanent ceasefire.

Amra Lee does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. Air-dropping food into Gaza is a ‘smokescreen’ – this is what must be done to prevent mass starvation – https://theconversation.com/air-dropping-food-into-gaza-is-a-smokescreen-this-is-what-must-be-done-to-prevent-mass-starvation-262053

Source: The Conversation – Global Perspectives – By James Giesecke, Professor, Centre of Policy Studies and the Impact Project, Victoria University

It’s tariff season again, with the next deadline looming on Friday, August 1.

Since the beginning of July, the United States has issued another flurry of tariff announcements, revising the sweeping plan announced on April 2. Back then, the Trump administration threatened to apply so-called “reciprocal” tariffs of up to 50% against many trading partners, plus an eye-watering 125% on Chinese imports.

In April, we modelled those measures, together with retaliation by trade partners. We reported they could cut more than 2.5% from US gross domestic product (GDP), reduce US short-run employment by 2.7%, and cut US real investment by almost 7%.

In the wake of those “Liberation Day” tariffs, financial markets took fright. On April 9 the Trump administration hit pause: the “reciprocal” tariffs were deferred until July 9 and replaced by an across-the-board 10% tariff increase, with a handful of exceptions.

Even so, the Trump tariff drum kept beating. Duties on steel and aluminium were doubled to 50%, and copper was swept in with its own 50% rate. Washington announced some “trade deals” with:

• United Kingdom – dropping the UK rate to the base rate of 10%

• China – cutting the tariff to 34%

• Vietnam – reducing its “reciprocal” tariff from 46% to 20%

• Japan – a 15% levy on all imports, including motor vehicles (otherwise tariffed at 25% for other regions)

• European Union – just announced at the weekend, reducing its “reciprocal” tariff from 30% to 15%.

When the first pause expired this month, a second extension pushed the start date for the “reciprocal” tariffs to August 1. But the tariff announcements keep coming, with recent threats to apply revised tariffs on imports from many trading partners, including a 50% tariff on imports from Brazil.

What do the new tariffs mean for the economy?

To find out, we reran our global economic model with the US tariff schedule as it stood on July 28, again allowing trading partners to retaliate proportionally (excluding Australia, Japan and South Korea, which have ruled out retaliation). This table compares the April projections with the updated results.

Damage to the US economy is less severe, but still substantial. In 2025, the falls in: real (inflation-adjusted) consumption narrow from a decline of 2.4% to 1.6%; real gross domestic product (GDP) from a fall of 2.6% to 1.7%; and real investment from a slide of 6.6% to 5.1%.

For the US, lower tariffs on the EU, UK, Japan, Vietnam and especially China, mean less disruption to short-run employment and long-run capital markets, lower efficiency losses in product markets, and less punishing retaliation from abroad.

Beijing also benefits from Washington’s climbdown. Short-run losses in Chinese real consumption shrink from 0.4% to 0.1% in 2025, and the GDP loss all but disappears. Cutting the US tariff on Chinese goods to 34%, and the corresponding pullback in China’s retaliatory duties, explains most of the improvement.

Australia is still a winner – but less so

Australia remains a beneficiary, but to a lesser degree. In April we projected short-run gains of 0.6% in consumption and 0.4% in GDP. These are now more modest, with a gain of 0.3% and 0.2% respectively. Two forces lie behind the downgrade:

1. Australia’s relative tariff treatment has narrowed. In April, Australia faced a 10% base tariff while many of our trade competitors in the US market confronted much higher “reciprocal” rates. Many of those have now been cut, eroding the relative price advantage of Australian products in the US market.

2. The global investment diversion is smaller. When investment contracts in the US and in regions relatively hard-hit by US tariffs and retaliatory action, some investment is reallocated to more lightly hit economies.

Because the projected fall in investment in the US and other regions is now milder, the corresponding investment inflow to Australia weakens, with our forecast boost to Australian investment dropping from 2.9% to 2.1% in the short-run.

What should Canberra do?

For the moment, the tariff outcomes still tilt slightly in Australia’s favour. That hardly argues for rushing into concessions in a bilateral “deal of the week” with Washington, let alone making unilateral concessions outside of any bargaining framework. This is especially true when US policy continues to appear reactive, volatile and unreliable.

However, the source of Australia’s edge is fragile.

As the US pares back its own tariff threats against other countries, the relative price advantage Australia enjoys of being subject only to the 10% base rate will diminish, and so too will the investment diversion effect.

Hence, a further US retreat from high and differentiated tariffs may yet expose Australia to net economic harm. That point has not arrived, but it may be on the horizon as US tariff policy evolves.

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

– ref. As Trump has pulled back from the highest tariffs, this chart shows the economic shock has eased – https://theconversation.com/as-trump-has-pulled-back-from-the-highest-tariffs-this-chart-shows-the-economic-shock-has-eased-261846

Source: The Conversation – Global Perspectives – By Kate Fitz-Gibbon, Professor (Practice), Faculty of Business and Economics, Monash University

In Australia, there is growing recognition that children and young people are not just witnesses to domestic, family and sexual violence, but victim-survivors in their own right.

While we are getting better at understanding how coercive control operates in adult relationships – particularly where men use it against women – much less attention has been given to how children experience this kind of abuse, especially when it comes from a parent or caregiver.

New research interviewing teenage victim-survivors reveals how parents can coercively control their children under the guise of parental discipline.

What is coercive control?

Coercive control is a pattern of abusive behaviours used to instil fear, dominate or isolate someone over time. It can include:

• physical violence

• sexual abuse

• surveillance

• threats

• humiliation

• limiting access to money

• technology-facilitated abuse

• animal abuse, among many other abusive tactics.

Focusing largely on adult victim-survivors, research has found experiences of coercive control can have cumulative and long-lasting negative impacts.

Studies of children show how coercive control can erode a child’s mental health, self-esteem and sense of safety.

Fear, guilt and manipulation

For young people, within the context of the family, coercive control may be perpetrated by parents, step-parents, caregivers, siblings and other family members. The tactics used may mirror those seen in adult contexts.

But there are different circumstances at play for children. They are typically dependent on their caregivers, still mentally developing, and often have limited access to external support.

My new report, Silence and Inaction, released by the South Australian Royal Commission into Domestic, Family and Sexual Violence, draws on interviews with 53 young people aged 13–18 who have experienced different forms of domestic, family and sexual violence in that state.

In this study, young victim-survivors spoke of rules imposed by abusive adults in their family to control their friendships, communication, bodily autonomy and emotional expression. These were often enforced through fear, guilt or manipulation. One child told me:

I kicked the wall when I was eight, and my parents came in and they stripped my entire room bare, just got rid of everything […] I was either in my room or was at school […] I got water brought to me, food brought to me three times a day […] they said, “You have abused this home. It was a loving place, and you’ve abused it so when people do things wrong, they go to prison”. I was very fearful of my parents.

Several young people described experiences that reflect the dynamics of coercive control, even if they did not use that language themselves. They spoke of environments where control, surveillance and isolation were constant, and where resistance or independence was met with punishment.

Experiences of gaslighting

Several young victim-survivors interviewed described being made to feel “crazy” or “overdramatic” when they challenged the behaviour they were experiencing. Others were punished for asserting boundaries or seeking help.

A number of young people described experiences of gaslighting – being told their memories or feelings were wrong or exaggerated.

---

Read more:
Explainer: what does ‘gaslighting’ mean?

---

This was particularly apparent among young people who had tried to speak up about the violence they were experiencing. One young victim-survivor told me:

I was very much gaslighting myself, and then also was being gaslit for years prior by my father and not made to feel that I could ever tell anyone.

Some young victim-survivors described beginning to question their own perceptions or feeling responsible for the harm they experienced. One young person said:

I always have a fear in my head that everything I’ve said and done [is] just a massive lie, which is why I documented a lot of things […] I have photos and videos of things that have happened […] it kind of keeps me a little bit sane.

For the young people interviewed, the dynamics of coercive control were further compounded by their legal and financial dependence on the person using violence.

Young people described having limited avenues to escape or resist the abuse, and having little access to alternative sources of care or trusted adults for support.

Discipline or control?

Many of the young people I interviewed said the abuse they experienced was explained away by parents as “discipline”.

Reasonable parenting involves setting boundaries and enforcing rules through clear communication and respect for a child’s emotional and physical safety. What the young people in the study described went well beyond that.

The young people interviewed described being physically punished – through beatings, slaps or threats – as a way of “correcting” behaviour or “teaching respect”.

For young people, this led to confusion and self-doubt about whether what they experienced “counted” as abuse.

This mislabelling of abuse as discipline was particularly difficult for young people to challenge when it was reinforced by religious, cultural or generational norms. In some cases, violence was deeply embedded in family tradition and viewed as an expected method of parenting.

Young people interviewed expressed a strong desire for this cycle to be broken, including through education for caregivers. One young victim said:

it’s not just kids who need to learn – adults need to unlearn the stuff they were taught too.

The need for change

Several young people believed some parents may be unaware of the impacts of these forms of punishment. They called for targeted awareness campaigns and community education. One young victim-survivor suggested:

they feel that is still part of discipline, whereas they are actually going extra miles […] I think parents too need to be educated on how they treat their children.

Several young people said their experiences of abuse were often minimised or dismissed as necessary or appropriate acts of discipline by extended family, caregivers or other adults in their community.

This highlights the need to better engage families and communities to change understandings of discipline, particularly through culturally responsive, trauma-informed approaches to education.

We must develop deeper understandings of coercive and controlling behaviours as they are experienced by children and young people in families.

Without such awareness, there is a risk that controlling behaviour will continue to be minimised as “strict parenting”, or young people’s disclosures will be dismissed.

These experiences highlight the problem of the normalisation of violence in some households and the need for greater prevention and early intervention efforts, both for young people and caregivers.

Kate has received funding for research on violence against women and children from a range of federal and state government and non-government sources, including Australia’s National Research Organisation for Women’s Safety (ANROWS), South Australian Government, ACT Government, Australian Childhood Foundation, and 54 Reasons. This piece is written by Kate Fitz-Gibbon in her role at Sequre Consulting, and is wholly independent of Kate Fitz-Gibbon’s role as chair of Respect Victoria and membership on the Victorian Children’s Council.

– ref. ‘I was very fearful of my parents’: new research shows how parents can use coercive control on their children – https://theconversation.com/i-was-very-fearful-of-my-parents-new-research-shows-how-parents-can-use-coercive-control-on-their-children-261169

Source: The Conversation – Global Perspectives – By Taliah Jade Prince, Postdoctoral Research Fellow in Youth Mental Health and Neuroimaging, University of the Sunshine Coast

You’re so ugly it hurts.

Maybe if you lost some weight, someone would actually like you.

No filter can fix that face.

These are the sorts of comments teenage girls see online daily, via social media, group chats, or anonymous messages. While some may dismiss this as teasing, these comments constitute appearance-related cyberbullying.

Our previous research shows appearance-related cyberbullying is one of the most common and harmful forms of online abuse of young people. It not only hurts feelings – it changes how teens, particularly girls, see themselves.

In a new study, we’ve looked at brain images of teenage girls viewing appearance-related cyberbullying. We’ve found even just being exposed to online body shaming directed at others can activate regions of the brain linked to emotional pain and social threat.

What is appearance-related cyberbullying?

Appearance-related cyberbullying is any online behaviour that targets the way someone looks. This includes comments about their face, clothes or body. It often happens in public forums, such as comment sections or social media posts, where other people can see it, join in or share it.

The most damaging type focuses on someone’s body, such as their weight, shape or size. These messages don’t need to be long or explicit to hurt. Sometimes a single word, hashtag or even emoji is enough.

While appearance-related bullying can affect anyone, previous studies have shown teenage girls are particularly vulnerable.

During adolescence, the brain is still developing – especially the parts that shape self-esteem and help us make sense of how others see us. This means teenagers can be more affected by what people say about them.

What’s more, girls often feel strong societal pressure to look a certain way. This combination makes body shaming especially harmful.

How common is it?

In a survey of 336 teenage girls we published last year, 98% had experienced some form of cyberbullying. For 62% of them, the abuse targeted their appearance.

Most of those girls said this bullying had lasting effects on their body image and mental health, with 96% saying it made them want to change how they looked. More than 80% felt they needed to consider cosmetic procedures.

Studies from around the world have shown appearance-related cyberbullying is a strong predictor of body dissatisfaction, which is one of the biggest risk factors for eating disorders in teenage girls.

What does it do to the brain?

To understand how body-shaming content affects girls on a deeper level, we designed a brain imaging study.

First, we created a set of social media posts based on typical comments teenage girls see online. Some posts were neutral, while others included body shaming comments.

More than 400 girls rated how realistic and emotionally powerful these posts were. This helped us validate the content so it could be used in current and future studies on how young people respond to body shaming online.

We then invited 26 girls aged 14 to 18, from the Longitudinal Adolescent Brain Study – a five-year research project at our university seeking to better understand how the teenage brain develops and how this relates to mental health – to take part in a brain scan study.

We used functional MRI, a technique that shows which areas of the brain are more active during certain experiences. Alongside the scans, participants completed questionnaires about their recent experiences of cyberbullying and their body image.

When girls viewed body-shaming posts, we found certain brain regions “lit up” more than others. These included areas involved in emotional pain, self-image, and social judgement. These are regions the brain uses to interpret how others see us, and how we deal with feelings such as shame or rejection.

Girls who had recently been cyberbullied showed more activity in memory and attention regions. This suggests they were reprocessing earlier, painful experiences. Girls with more positive body image, meanwhile, showed calmer, more regulated brain responses, suggesting healthy self-image might be protective.

Girls are affected even when they’re not targeted

Notably, the girls in our study were viewing posts aimed at others – not being subjected to bullying directly. But even so, we saw changes in the way their brains reacted, and how they felt about their own bodies seemed to affect these reactions.

This tells us something important: body-shaming content doesn’t just hurt the person it targets. When appearance is constantly judged and criticised, it can change what girls think is normal or acceptable. It may also affect how their brains respond to social and emotional situations.

What needs to change?

Appearance-related cyberbullying is not just about teenage conflict. It’s a wider, societal issue. Social media platforms reward content that grabs attention, even when it causes harm.

All of this is happening during a sensitive period of brain development, where social feedback shapes how teenagers see themselves and others.

To reduce harm, we need to act on multiple levels:

• start early: while some schools offer lessons on body image and online safety, these topics are not taught consistently. Many young people say they want more support in dealing with appearance-related pressure online

• support parents and educators: adults need tools, resources and language to talk with young people about what they see online, without shame or blame

• hold platforms accountable: social media companies should strengthen reporting systems, and better moderate content that may promote appearance-related abuse such as “before-and-after” posts or other viral trends that target how someone looks

• celebrate all body types: schools, media and influencers can help by showing real people with different body types and focusing on strengths such as kindness, talent, or what bodies can do.

Adolescence is a time of major change in how teenagers think, manage emotions and build relationships. What teenagers experience during these years can shape how they see themselves and understand the world.

Online body shaming may seem like just words on a screen. But if we want the next generation to grow up confident and well, we need to take it seriously.

In Australia, if you are experiencing body image concerns, you can contact the Butterfly Foundation’s national helpline on 1800 33 4673 (or use their online chat).

Daniel Hermens receives funding from the Commonwealth government’s Prioritising Mental Health Initiative and the Queensland Mental Health Commission.

Taliah Jade Prince does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. ‘No filter can fix that face’: how online body shaming harms teenage girls – https://theconversation.com/no-filter-can-fix-that-face-how-online-body-shaming-harms-teenage-girls-261362

Source: The Conversation – Global Perspectives – By Celia Harris, Associate Professor in Cognitive Science, Western Sydney University

As school returns, parents and teachers might each be faced with the familiar chorus of “I can’t find my school jumper” and “I left my hat at home”. For parents of older kids, the stakes may be even higher: lost mobile phones or laptops left on the bus.

As parents, it can be tempting to take charge by packing schoolbags yourself, or texting older children a list of things to remember at the end of each day.

However, doing everything for your child robs them of an opportunity to learn.

What’s happening in their developing brain?

Our kids, in their busy lives, are constantly using and developing their memory skills – remembering where they put things, new conceptual knowledge, and routines required for the day-to-day.

Prospective memory – which involves remembering to do things in the future – is particularly challenging.

It’s prospective memory children draw on when they set a drink bottle down at play time and must remember to pick it up later, or get a note from their teacher and must remember to show their parent after school.

Success in prospective memory involves multiple cognitive processes going right.

Children must pay attention to what is needed in a given situation (“I can’t play outside if I don’t have a hat”), and then form and store a particular intention to act in the future (“I need to take my hat with me to school”).

Then, they must bring the intention back to mind at the crucial moment (taking the hat on the way out the door).

This “remembering to remember” requires memory to spontaneously occur at just the right time, without prompts or reminders.

These processes all require a higher-order cognitive skill known as “executive function”.

This is the ability to consciously control our attention and memory and to engage in challenging thinking tasks.

Processes that rely on executive function are hard, which is why lost drink bottles and forgotten hats are such frustratingly common experiences for parents.

Even for adults, the majority of day-to-day memory errors involve prospective memory.

Executive function develops later in childhood compared with some other skills, such as language and play.

The prefrontal cortex, which underpins executive function tasks, is not mature until early adulthood.

This means forgetfulness among children is common, and a natural part of development. Chances are you were like this too when you were a kid (you just might not remember it).

Could some kids struggle even more?

Yes.

Children (and adults) vary widely in their executive function skills.

While all children get better at executive function throughout childhood, this happens at different rates; some children may be more forgetful than others their own age.

One condition particularly related to forgetfulness is attention-deficit/hyperactivity disorder (ADHD).

Children with an ADHD-inattentive subtype may be more likely to lose things and be forgetful during everyday activities such as chores or errands.

Children with ADHD will still develop prospective memory skills over time, but may be more forgetful relative to other children their own age.

How can I help my kid?

Do build routines and stick to them. Research shows routines help children develop cognitive skills and self-regulation. Children are best able to remember a routine when it is “automatised” – practised often enough they know it without thinking.

Do promote “metacognition”: an awareness about one’s own cognitive processes. Research suggests children are over-optimistic about their likelihood of remembering successfully. Parents and teachers can help them to notice when remembering is hard and put in strategies that help.

Do model the behaviour you want to see. For example, you might set up your own lists and strategies to help you remember daily tasks. You could also have a family routine of “bags by the door” and checking them the night before. Don’t do it for them, do it together.

Do seek professional support if you’re worried. All children will forget sometimes, and some more than others. If your child is particularly absentminded or forgetful, it could be worth consulting a GP or school psychologist. Conditions such as ADHD must be observed in more than one setting (for example, home and school, or home and sport), and specific diagnostic criteria must be met. Diagnosis can be helpful in accessing supports.

What should I not do – and why?

Don’t rely on kids being able to spontaneously self-initiate memory – that’s the hardest part of prospective memory! Instead, use checklists and memory aids. For instance, if they are consistently leaving their drink bottle at school, you could put a tag on their bag that says “where is your drink bottle?” Using prompts isn’t cheating – it’s supporting success.

Don’t sweat the slip-ups – these are normal. One study with 3–5-year-old children found incentives in the form of food treats weren’t enough to improve performance. Punishing is also unlikely to help. Instead, use instances of forgetting as teachable moments – strategise about how to adjust next time.

Don’t leave things too late. Anxiety and stress can make forgetting more likely, because children can easily become overwhelmed. Pack bags the night before, practise new routines, and avoid rushing where possible.

Don’t judge. Prospective memory failures are sometimes perceived as character flaws, particularly when they affect other people (such as when forgetting to return a borrowed item).

Understanding how memory works, however, helps reveal that forgetfulness is an everyday part of development.

Celia Harris receives funding from the Australian Research Council and the Longitude Prize on Dementia.

Penny Van Bergen receives funding from the ARC, Marsden, Google, and the James Kirby Foundation.

– ref. My child is always losing and forgetting things. How can I help – without making it worse? – https://theconversation.com/my-child-is-always-losing-and-forgetting-things-how-can-i-help-without-making-it-worse-261565

Source: The Conversation – Global Perspectives – By Aditya Joshi, Senior Lecturer, School of Computer Science and Engineering, UNSW Sydney

In 2018, my Australian co-worker asked me, “Hey, how are you going?”. My response – “I am taking a bus” – was met with a smirk. I had recently moved to Australia. Despite studying English for more than 20 years, it took me a while to familiarise myself with the Australian variety of the language.

It turns out large language models powered by artificial intelligence (AI) such as ChatGPT experience a similar problem.

In new research, published in the Findings of the Association for Computational Linguistics 2025, my colleagues and I introduce a new tool for evaluating the ability of different large language models to detect sentiment and sarcasm in three varieties of English: Australian English, Indian English and British English.

The results show there is still a long way to go until the promised benefits of AI are enjoyed by all, no matter the type or variety of language they speak.

Limited English

Large language models are often reported to achieve superlative performance on several standardised sets of tasks known as benchmarks.

The majority of benchmark tests are written in Standard American English. This implies that, while large language models are being aggressively sold by commercial providers, they have predominantly been tested – and trained – only on this one type of English.

This has major consequences.

For example, in a recent survey my colleagues and I found large language models are more likely to classify a text as hateful if it is written in the African-American variety of English. They also often “default” to Standard American English – even if the input is in other varieties of English, such as Irish English and Indian English.

To build on this research, we built BESSTIE.

What is BESSTIE?

BESSTIE is the first-of-its-kind benchmark for sentiment and sarcasm classification of three varieties of English: Australian English, Indian English and British English.

For our purposes, “sentiment” is the characteristic of the emotion: positive (the Aussie “not bad!”) or negative (“I hate the movie”). Sarcasm is defined as a form of verbal irony intended to express contempt or ridicule (“I love being ignored”).

To build BESSTIE, we collected two kinds of data: reviews of places on Google Maps and Reddit posts. We carefully curated the topics and employed language variety predictors – AI models specialised in detecting the language variety of a text. We selected texts that were predicted to be greater than 95% probability of a specific language variety.

The two steps (location filtering and language variety prediction) ensured the data represents the national variety, such as Australian English.

We then used BESSTIE to evaluate nine powerful, freely usable large language models, including RoBERTa, mBERT, Mistral, Gemma and Qwen.

Inflated claims

Overall, we found the large language models we tested worked better for Australian English and British English (which are native varieties of English) than the non-native variety of Indian English.

We also found large language models are better at detecting sentiment than they are at sarcasm.

Sarcasm is particularly challenging, not only as a linguistic phenomenon but also as a challenge for AI. For example, we found the models were able to detect sarcasm in Australian English only 62% of the time. This number was lower for Indian English and British English – about 57%.

These performances are lower than those claimed by the tech companies that develop large language models. For example, GLUE is a leaderboard that tracks how well AI models perform at sentiment classification on American English text.

The highest value is 97.5% for the model Turing ULR v6 and 96.7% for RoBERTa (from our suite of models) – both higher for American English than our observations for Australian, Indian and British English.

National context matters

As more and more people around the world use large language models, researchers and practitioners are waking up to the fact that these tools need to be evaluated for a specific national context.

For example, earlier this year the University of Western Australia along with Google launched a project to improve the efficacy of large language models for Aboriginal English.

Our benchmark will help evaluate future large language model techniques for their ability to detect sentiment and sarcasm. We’re also currently working on a project for large language models in emergency departments of hospitals to help patients with varying proficiencies of English.

The research, led by Dipankar Srirag, was funded by Google’s Research Scholar grant awarded in 2024 to Aditya Joshi and Diptesh Kanojia.

– ref. ‘Are you joking, mate?’ AI doesn’t get sarcasm in non-American varieties of English – https://theconversation.com/are-you-joking-mate-ai-doesnt-get-sarcasm-in-non-american-varieties-of-english-254986