Paul Morrill | October 2026

Introduction

I have been following the renewed enthusiasm for oracy in British education with some interest, although I confess to having reservations about the direction in which the debate appears to be heading. The argument that children should develop the confidence and ability to express themselves verbally seems perfectly reasonable. Indeed, it would be difficult to argue otherwise. However, the proposal to give oracy equal standing with reading, writing and arithmetic raises questions about what we understand by intellectual competence, how we recognise it, and whose interests might ultimately be served by such a change.

My reservations are partly personal. I have never particularly enjoyed public speaking, and I suspect I am not alone in finding that the ability to formulate an idea bears little relationship to my capacity to deliver it convincingly before an audience. Given the choice, I would much rather be a playwright than an actor. The former is concerned with constructing something worth saying; the latter with bringing it to life before an audience. Both require considerable skill, but I am not convinced that excellence in one necessarily implies excellence in the other.

What concerns me is the possibility that, in our enthusiasm to cultivate articulate speakers, we may inadvertently reinforce an existing confusion between intellectual ability and the social performance of intelligence. This becomes particularly interesting when considered alongside generative artificial intelligence, which is beginning to make sophisticated written expression available to individuals who may previously have struggled to communicate their ideas within established academic and professional conventions.

My argument is not that oracy should be abandoned. Rather, I question whether its elevation within education might, under certain circumstances, reproduce existing social inequalities while introducing new barriers to intellectual participation. There is also a more speculative possibility worth examining: that as AI weakens some traditional advantages associated with linguistic proficiency, institutions may increasingly attach value to forms of communication through which established distinctions can be maintained.

Oracy and the problem of intellectual performance

The Commission on the Future of Oracy Education in England (2024) recommended that oracy should become the fourth 'R' of education, enjoying equal status with reading, writing and arithmetic. Its report, We Need to Talk, identifies spoken communication as important for educational attainment, citizenship and employment, particularly in a society increasingly influenced by artificial intelligence.

There is evidence supporting the educational value of spoken language. The Education Endowment Foundation (2025) reports that oral language interventions produce, on average, the equivalent of six additional months of academic progress. The findings are particularly encouraging for disadvantaged pupils, suggesting that well-designed interventions may help address existing educational inequalities.

However, an important distinction is sometimes lost in the wider discussion. Much of the evidence concerns structured classroom dialogue, vocabulary development, listening and reading comprehension rather than public speaking or the ability to deliver a persuasive presentation. The Education Endowment Foundation itself cautions against assuming that evidence supporting one form of oral language activity necessarily validates every practice associated with oracy.

This distinction matters because learning through conversation is not equivalent to demonstrating intellectual competence through verbal performance. A pupil who benefits from discussing a difficult concept with a teacher may not necessarily demonstrate the same understanding when required to explain it spontaneously before a group.

Nor should we assume that a confident verbal explanation provides a particularly reliable measure of the sophistication of the underlying thought. Some individuals think through conversation, while others arrive at their conclusions through prolonged reflection, writing, revision or experimentation. Neither approach is inherently superior, although educational practices may reward one more readily than the other.

The concern, therefore, is not with spoken language as a means of learning, but with the possibility of treating fluent speech as evidence of intellectual ability.

Linguistic capital and the privilege of sounding authoritative

Bourdieu's (1991) account of language and symbolic power provides a useful framework for understanding why certain forms of expression acquire authority. Language does more than communicate information. It also conveys social position, legitimacy and membership of particular cultural groups.

Within British education, the ability to speak confidently, debate persuasively and present oneself with apparent authority has long been associated with certain educational and social environments. Independent schools have historically placed considerable emphasis on public speaking, debating and leadership activities, although such opportunities are by no means exclusive to them. Nor should grammar schools and independent schools be treated as socially or educationally interchangeable.

The important issue is that confidence and linguistic authority are not distributed independently of social experience. Individuals enter educational and professional institutions with different opportunities to acquire the forms of expression those institutions recognise and reward.

Research by Lev-Ari and Keysar (2010) offers a useful illustration of how speech characteristics can influence judgements independently of the information being communicated. Their experiments found that statements delivered with a non-native accent were judged less credible than equivalent statements spoken by native speakers. The researchers attributed this effect, at least partly, to difficulties in processing accented speech.

The findings do not establish that public speaking ability generally determines perceived intelligence, nor do they demonstrate class discrimination within British education. They do, however, illustrate a broader psychological difficulty: judgements about communicated information can be influenced by characteristics of its delivery that are unrelated to its truth.

We might reasonably ask whether something similar occurs when educational institutions reward gravitas, confidence and fluency. How often do we mistake an authoritative presentation for a sound argument? How frequently does the person who speaks most readily become the person whose contribution receives the greatest attention?

I suspect most people who have spent time in universities or other large organisations will recognise the phenomenon, although personal observation is no substitute for systematic evidence.

There is a further irony. If oracy becomes an increasingly important educational and professional credential, those already familiar with the conventions of authoritative speech may be especially well placed to benefit. A reform intended to improve equality of opportunity could therefore reproduce aspects of the very inequality it seeks to address.

That outcome is not inevitable. It depends on how oracy is defined, taught and assessed.

Construct validity: What are we actually measuring?

For psychologists, the distinction between an underlying ability and its observable expression should be familiar. It is, after all, central to the problem of construct validity.

Messick (1995) argued that the validity of educational and psychological assessment concerns the interpretation and use of scores, including their broader consequences. The question is not simply whether an assessment produces a reliable measurement, but whether the inferences drawn from that measurement are justified.

Consider a student who has developed an original and theoretically sophisticated argument but struggles to explain it during an oral examination. Another student may possess a relatively superficial understanding yet deliver a fluent and confident account.

If the assessment is intended to measure public speaking, differences in verbal performance are clearly relevant. If it is intended to measure conceptual understanding, however, the relationship becomes more problematic.

The danger is that characteristics associated with verbal performance may introduce construct-irrelevant variation into assessments ostensibly concerned with knowledge, reasoning or originality.

This is not an argument against oral examinations. In some circumstances, particularly where spontaneous reasoning or professional communication forms part of the intended competence, they may be entirely appropriate. Oral questioning can also provide useful evidence about whether a student understands work submitted in another format.

Nevertheless, the legitimacy of an assessment method depends on the claims made about what it measures. We should be cautious about allowing an individual's ability to communicate knowledge in one particular manner to stand as a comprehensive measure of the knowledge itself.

The implications extend beyond assessment. Individuals who experience anxiety when speaking publicly, who require additional time to formulate responses, or whose communication differs from dominant social conventions may find themselves disadvantaged by systems that increasingly equate immediate verbal performance with intellectual competence.

It would be unfortunate if an educational movement intended to broaden participation ended up narrowing the acceptable ways in which understanding could be demonstrated.

Artificial intelligence and the changing value of linguistic capital

Generative AI introduces another dimension to this discussion.

For much of modern academic and professional life, sophisticated written expression has operated as both a means of communication and a marker of educational distinction. The ability to produce an argument in an accepted register, organise complex material and conform to disciplinary conventions has often been taken as evidence of intellectual competence.

These skills remain valuable. However, generative AI has made aspects of their production considerably more accessible.

An individual who possesses an original idea but lacks the confidence, vocabulary or familiarity with academic conventions to communicate it effectively can now obtain assistance with structure, grammar and expression. This does not guarantee the quality or originality of the resulting argument. AI may distort an author's intentions, introduce errors or produce language that appears more sophisticated than the reasoning it contains.

Nevertheless, the technology creates opportunities for people who have previously found established linguistic conventions difficult to negotiate.

UNESCO (Miao & Holmes, 2023) identifies both the opportunities and risks associated with generative AI in education, including questions of equity, inclusion and human agency. These considerations are particularly important because access to AI tools does not automatically eliminate existing inequalities. Digital literacy, resources and the ability to evaluate generated material remain unevenly distributed.

Even with these qualifications, AI potentially changes the relationship between intellectual content and its linguistic presentation.

Someone who previously struggled to write in an academic register may now be able to participate in discussions from which they felt excluded. The intellectual merit of their contribution must still be established, but the conventional barriers to presenting it may be reduced.

This raises a question about the future distribution of intellectual authority.

If sophisticated written expression becomes more widely accessible, will institutions become more willing to distinguish intellectual originality from linguistic accomplishment? Or will they increasingly privilege forms of performance that remain less readily mediated by technology?

The growing interest in oral assessment and spontaneous verbal defence may be educationally justified, particularly where institutions need to establish whether students understand AI-assisted work. However, there is a possibility that these developments will also reproduce older distinctions between those comfortable with public intellectual performance and those who are not.

Bourdieu's framework suggests a useful hypothesis: when the means of acquiring a valued form of cultural capital become more widely available, the social significance of other forms of distinction may increase.

This is an interpretation rather than an established explanation of current educational policy. The movement to promote oracy predates the widespread adoption of generative AI, and there is insufficient evidence to conclude that its present prominence represents a deliberate attempt to protect existing intellectual hierarchies.

Nevertheless, the intersection deserves investigation. The motivations behind an educational reform and its eventual social consequences need not be identical.

An argument against my own position

There is a substantial counterargument to the concerns raised here, and it should not be dismissed.

If the ability to speak confidently and persuasively has historically conferred an advantage on privileged groups, surely teaching those skills to everyone is an effective way of reducing that advantage?

This is, in essence, the argument advanced by supporters of universal oracy education. It is also consistent with evidence suggesting that structured oral language interventions can be particularly beneficial for disadvantaged pupils (Education Endowment Foundation, 2025).

The argument has considerable merit. Children should not be denied opportunities to develop their speaking and listening skills simply because such skills have historically been associated with privileged educational environments. Nor should discomfort with public speaking be treated as an immutable characteristic that education has no responsibility to address.

There is a difference, however, between providing everyone with opportunities to develop a capability and making that capability an increasingly important condition of intellectual recognition.

The former can expand participation. The latter may create new forms of exclusion if institutions fail to distinguish between communicative performance and the knowledge or reasoning they intend to assess.

There is also a danger in treating oracy as a single, easily measurable attribute. Listening, questioning, explaining, debating, persuading and presenting involve overlapping but distinguishable abilities. A person may be an excellent listener and thoughtful conversationalist while remaining deeply uncomfortable with public performance.

An inclusive approach to oracy should recognise these differences rather than establish a narrow ideal of the articulate, confident and immediately responsive individual.

AI may even assist with this objective. Private conversational practice, simulated discussion and assistance with organising arguments could help individuals develop communication skills without exposing them immediately to the social pressures of public performance.

The educational challenge is therefore not to choose between oracy and AI, but to avoid allowing either to dictate an unnecessarily narrow definition of intellectual competence.

Towards a more inclusive conception of intellectual expression

The central issue is whether education should reward particular forms of expression or the intellectual achievements those forms are intended to communicate.

Of course, the distinction cannot be absolute. Language is not simply a container into which completed thoughts are placed. Discussion, explanation and argument can contribute directly to the development of understanding. In some circumstances, the act of articulating an idea changes the idea itself.

Nevertheless, this does not establish that every individual must demonstrate understanding through the same communicative processes.

An educational system genuinely concerned with intellectual development should recognise multiple legitimate routes through which knowledge and reasoning can be expressed. Written argument, oral explanation, practical demonstration, visual representation and computational modelling may each provide valuable evidence, depending on the nature of the intellectual task.

Where spoken communication is itself an intended learning outcome, it should be taught and assessed accordingly. Where the objective is conceptual understanding, originality or critical reasoning, institutions should consider whether requiring a particular form of performance introduces unnecessary barriers.

This distinction becomes increasingly important as AI alters the means through which people can communicate.

The possibility that an individual may produce an original idea while relying on technological assistance to express it should not automatically diminish the value of that idea. Equally, the ability to produce polished language with AI should not be mistaken for evidence of independent understanding.

Both cases demand more careful judgements about intellectual contribution, not a return to familiar but potentially misleading indicators of competence.

Conclusion

I began with a personal observation: I would rather be a playwright than an actor. It is a distinction that continues to trouble me when I consider the increasing emphasis on oracy.

There are excellent reasons to help children and adults become more confident speakers. The evidence supporting purposeful oral language development is substantial, and the potential benefits for disadvantaged learners should be taken seriously.

My concern is with what happens when an educational capability becomes a measure of intellectual worth.

The ability to speak fluently, confidently and persuasively is valuable, but it should not be confused with the capacity to think originally or critically. Nor should we assume that individuals who struggle with public expression have less to contribute.

Generative AI has begun to challenge some of the linguistic conventions through which intellectual authority has traditionally been recognised. Whether this produces a more inclusive intellectual environment remains uncertain. It could broaden participation, but it could also encourage institutions to seek reassurance in other, more familiar demonstrations of competence.

The renewed enthusiasm for oracy may therefore represent an opportunity to reduce educational inequality, or it may unintentionally preserve aspects of it. Much will depend on whether we teach spoken communication as one means of intellectual participation or allow it to become a privileged measure of intellectual ability.

I would welcome an education system in which more people felt able to speak. I would be considerably less enthusiastic about one in which those who speak most confidently are assumed to have the most worth saying.

Perhaps the real challenge is not simply to give more people a voice, but to become better at recognising the value of what they have to say, however they choose to say it.

References

Bourdieu, P. (1991). Language and symbolic power (J. B. Thompson, Ed.; G. Raymond & M. Adamson, Trans.). Polity Press.

Commission on the Future of Oracy Education in England. (2024). We need to talk: The report of the Commission on the Future of Oracy Education in England. https://oracyeducationcommission.co.uk/oec-report/

Education Endowment Foundation. (2025). Oral language interventions. https://educationendowmentfoundation.org.uk/education-evidence/teaching-learning-toolkit/oral-language-interventions

Lev-Ari, S., & Keysar, B. (2010). Why don't we believe non-native speakers? The influence of accent on credibility. Journal of Experimental Social Psychology, 46(6), 1093–1096. https://doi.org/10.1016/j.jesp.2010.05.025

Messick, S. (1995). Validity of psychological assessment: Validation of inferences from persons' responses and performances as scientific inquiry into score meaning. American Psychologist, 50(9), 741–749. https://doi.org/10.1037/0003-066X.50.9.741

Miao, F., & Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO. https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research