You are invited to join us for our English Research Seminar on 1st October 2025 from 1.15-3pm. Please contact Prof. Caroline Magennis on c.magennis@salford.ac.uk for the venue and any further details.
Our speakers will be Aisling Logan (PhD Candidate) and Dr Jack Wilson (Lecturer in English Language). Both abstracts are below – Aisling’s paper is titled Social Representations of Irish Immigrants in British Media 1923-2016: A Psycho-Linguistic Examination and Jack’s paper is titled Are the outputs of AI meaningful? We hope you can join us for what will surely be a fascinating pair of papers!
Social Representations of Irish Immigrants in British Media 1923-2016: A Psycho-Linguistic Examination (Aisling Logan)
Ireland has seen a significant increase in immigration focused protests. Research indicates protestors are divided into two categories: far-right groups opposing immigration particularly in relation to increased demands on local services and counter-protestors who reject far-right ideologies but recognise the validity of resource related complaints. Interestingly, sociological, psychological, and historical, research demonstrates that for much of the 20th century Irish immigration to Britain was debated, by British communities, under similar conditions. However, contemporary media analysis suggests immigration in Britian is discussed through the oppositional categories of “illegal migration” and “European migration”. This indicates a change in the social understanding of Irish immigrants in Britain. The processes facilitating this change are under-explored. Diasporic populations maintain links to a homeland when residing in a secondary country. The Irish government has a keen interest in its diaspora supporting diasporic communities and promoting Irish culture abroad. However, this relationship has been criticised as one-directional with ‘Irishness’ defined by the State and the complexities of the emigrant experience ignored. In examining the changing position of Irish immigrants in Britain it is suggested that 1. A considered viewpoint is offered to contemporary discussions of immigration in Ireland through examination of the experiences of diaspora and 2. Rebalancing the relationship between the Irish State and diaspora with diaspora reframed as knowledgeable resource in relation to the current political landscape. This will be achieved by accessing the British Newspaper archives and examining the most salient discussions of Irish immigration. Data will then be used to create a corpus to guide linguistic analysis. Lastly, Social Representation Theory, a theory from social psychology exploring the social understandings individuals have about the world, will be applied to examine the changing position of Irish immigrants in Britain.
Are the outputs of AI meaningful? (Jack Wilson)
In 2022, one of the pioneers of the technologies underpinning artificial intelligence (AI) and co-founder of OpenAI, Ilya Sutskever, tweeted “it may be that today’s large neural networks are slightly conscious.” In the same year, another co-founder (and current CEO) of OpenAI, Sam Altman, tweeted “I am a stochastic parrot and so ru” suggesting that his cognitive functioning is no different from the technologies found in AI. While in 2023 Microsoft published an article called “Sparks of Artificial General Intelligence”, alluding to the notion of a spark of life. Anyone paying attention to such pronouncements and the lack of critical scrutiny from the media could be forgiven for thinking that the technologies generally referred to as AI are approaching human-like cognition. However, it is almost unanimously understood that AI technologies bear very little resemblance to the human brain. Focussing on Large Language Models (LLMs) that underpin applications like chatGPT, in this talk I will address a question that is a precondition for any serious talk of AI consciousness: are the outputs of AI applications meaningful?
There are two serious answers to this question: positive and negative. On the positive side, scholars argue that because large language models (LLMs) are trained on massive databases of human language, and because organisations like OpenAI rely on reinforcement learning from human feedback, then the outputs of AI systems should be considered as meaningful. In fact, since both AI applications and humans rely on pre-existing linguistic systems, the outputs of AI are meaningful in the same way that many aspects of human utterances are. On the negative side are those who argue that we should not consider the outputs of AI as meaningful because there is no intention behind the outputs. AI systems like ChatGPT are simply predicting the next item in a sequence and have no connection to the intentionality of utterances. If AI systems appear to produce meaningful outputs this is an illusion, and we should be cautious of any claims to the contrary.
Both approaches miss a fundamental difference between the outputs of AI and human communicative acts. Namely, they conflate or ignore the multiple timeframes over which meaningfulness has emerged in human languages. These timeframes relate to evolution, history, acquisition, and processing. When an English-speaking human being uses the word “dog” to refer to a dog, they do so because (1) humans have evolved to interact with a world that includes dogs; (2) the English language has a cultural history of referring to dogs; (3) the individual human being has learned the word “dog”; and (4) the individual speaker has an intention when producing the word “dog”. The positive arguments in favour of AI outputs being meaningful rely on the (often superficial) similarities between AI systems and human beings at levels (2) and (3), whereas negative responses rely on the lack of similarity at level (4). I will argue that to develop a critical understanding of the meaningful nature of AI outputs, we need to consider all levels (1-4), with specific emphasis on level (1).