Matthew Purver

• BA, MPhil, PhD
• Professor (Full) at Queen Mary University of London

Previous research has shown that political leanings correlate with various psychological factors. While surveys and experiments provide a rich source of information for political psychology, data from social networks can offer more naturalistic and robust material for analysis. This research investigates psychological differences between individual...

We present STIR (STrongly Incremental Repair detection), a system that detects speech repairs and edit terms on transcripts incrementally with minimal latency. STIR uses information-theoretic measures from n-gram models as its principal decision features in a pipeline of classifiers detecting the different stages of repairs. Results on the Switchbo...

We provide a comparative study between neural word representations and traditional vector spaces based on co-occurrence counts, in a number of compositional tasks. We use three different semantic spaces and implement seven tensor-based compositional models, which we then test (together with simpler additive and multiplicative approaches) in tasks i...

Mental illnesses such as depression and anxiety are highly prevalent, and therapy is increasingly being offered online. This new setting is a departure from face-toface therapy, and offers both a challenge and an opportunity ‐ it is not yet known what features or approaches are likely to lead to successful outcomes in such a different medium, but o...

One of the best known claims about human communication is that people’s behaviour and language use converge during conversation. It has been proposed that these patterns can be explained by automatic, cross-person priming. A key test case is structural priming: does exposure to one syntactic structure, in production or comprehension, make reuse of...

From natural language processing to vision, Scaled Dot Product Attention (SDPA) is the backbone of most modern deep learning applications. Unfortunately, its memory and computational requirements can be prohibitive in low-resource settings. In this paper, we improve its efficiency without sacrificing its versatility. We propose three attention vari...

In the domain of text-to-image generative models, biases inherent in training datasets often propagate into generated content, posing significant ethical challenges, particularly in socially sensitive contexts. We introduce FairCoT, a novel framework that enhances fairness in text-to-image models through Chain-of-Thought (CoT) reasoning within mult...

As large-scale vision-language models (VLMs) become increasingly central to modern AI applications, understanding and mitigating social biases in these systems has never been more critical.We investigate how dataset composition, model size, and multilingual training affect gender and racial bias in a popular VLM, CLIP, and its open-source variants....

Linear Text Segmentation is the task of automatically tagging text documents with topic shifts, i.e. the places in the text where the topics change. A well-established area of research in Natural Language Processing, drawing from well-understood concepts in linguistic and computational linguistic research, the field has recently seen a lot of inter...

We investigate zero-shot cross-lingual news sentiment detection, aiming to develop robust sentiment classifiers that can be deployed across multiple languages without target-language training data. We introduce novel evaluation datasets in several less-resourced languages, and experiment with a range of approaches including the use of machine trans...

We introduce ClarQ-LLM, an evaluation framework consisting of bilingual English-Chinese conversation tasks, conversational agents and evaluation metrics, designed to serve as a strong benchmark for assessing agents' ability to ask clarification questions in task-oriented dialogues. The benchmark includes 31 different task types, each with 10 unique...

Large Language Models (LLMs) have demonstrated remarkable capabilities in comprehending and analyzing lengthy sequential inputs, owing to their extensive context windows that allow processing millions of tokens in a single forward pass. However, this paper uncovers a surprising limitation: LLMs fall short when handling long input sequences. We inve...

The application of machine learning (ML) in detecting, diagnosing, and treating mental health disorders is garnering increasing attention. Traditionally, research has focused on single modalities, such as text from clinical notes, audio from speech samples, or video of interaction patterns. Recently, multimodal ML, which combines information from m...

The self is a pervasive aspect of human experience, influencing crucial areas like mental health and manifesting in the texts we produce. Previous research indicates a significant correlation between the use of self-related expressions-terms and linguistic structures individuals use to refer to themselves, such as first-person pronouns-and various...

This paper presents a split of a Slovenian news corpus based on the readers' political leaning. By combining Slovenian news data with a large survey giving data on media consumption and self-reported political orientation, we create sub-corpora of news outlets consumed by left-, centre, and right-leaning readers and use it to build a political orie...

An important concept in organisational behaviour is how hierarchy affects the voice of individuals, whereby members of a given organisation exhibit differing power relations based on their hierarchical position. Although there have been prior studies of the relationship between hierarchy and voice, they tend to focus on more qualitative small-scale...

In this paper we propose a new framework and new methods for the reference-free evaluation of topic segmentation systems directly in the embedding space. Specifically, we define a common framework for reference-free, embedding-based topic segmentation metrics, and show how this applies to an existing metric. We then define new metrics, based on a p...

Dementia affects cognitive functions of adults, including memory, language, and behaviour. Standard diagnostic biomarkers such as MRI are costly, whilst neuropsychological tests suffer from sensitivity issues in detecting dementia onset. The analysis of speech and language has emerged as a promising and non-intrusive technology to diagnose and moni...

Organizational responsibilities can give people power but also expose them to scrutiny. This tension leads to divergent predictions about the use of potentially sensitive language: power might license it, while exposure might inhibit it. Analysis of peoples' language use in a large corpus of organizational emails using standardized Linguistic Inqui...

Neural sentence encoders (NSE) are effective in many NLP tasks, including topic segmentation. However, no systematic comparison of their performance in topic segmentation has been performed. Here, we present such a comparison, using supervised and unsupervised segmentation models based on NSEs. We first compare results with baselines, showing that...

Social biases are biases toward specific social groups, often accompanied by discriminatory behavior. They are reflected and perpetuated through language and language models. In this study, we consider two language models (RoBERTa, in English; and UmBERTo, in Italian), and investigate and compare the presence of social biases in each one. Masking t...

Recent works on linear text segmentation have shown new state-of-the-art results nearly every year. Most times, however, these recent advances include a variety of different elements which makes it difficult to evaluate which individual components of the proposed methods bring about improvements for the task and,more generally, what actually works...

We present two multimodal models for topic segmentation of podcasts built on pre-trained neural text and audio embeddings. We show that results can be improved by combining different modalities; but also by combining different encoders from the same modality, especially general-purpose sentence embeddings with specifically fine-tuned ones. We also...

Linguoplotter is a distributed and chaotic architecture where an entanglement of different processes interact to generate a text describing a raw data input. This paper describes recent additions to the architecture whereby a greater degree of language comprehension is used to improve the coherence of generated text. Some examples of the architectu...

Social robots have limited social competences. This leads us to view them as depictions of social agents rather than actual social agents. However, people also have limited social competences. We argue that all social interaction involves the depiction of social roles and that they originate in, and are defined by, their function in accounting for...

We study the influence of different activation functions in the output layer of deep neural network models for soft and hard label prediction in the learning with disagreement task. In this task, the goal is to quantify the amount of disagreement via predicting soft labels. To predict the soft labels, we use BERT-based preprocessors and encoders an...

In light of unprecedented increases in the popularity of the internet and social media, comment moderation has never been a more relevant task. Semi-automated comment moderation systems greatly aid human moderators by either automatically classifying the examples or allowing the moderators to prioritize which comments to consider first. However, th...

This paper introduces Linguoplotter, a workspace-based architecture for generating short natural language descriptions. All processes within Linguoplotter are carried out by codelets, small pieces of code each responsible for making incremental changes to the program’s state, the idea of which is borrowed from Hofstadter et al. [6]. Codelets in Lin...

Nowadays in the finance world, there is a global trend for responsible investing, linked with a growing need for developing automated methods for analysing Environmental, Social and Governance (ESG) related elements in financial texts. In this work we propose a solution to the FinSim4-ESG task, consisting in classifying sentences from financial rep...

The reverse dictionary is a sequence-to-vector task in which a gloss is provided as input, and the model is trained to output a semantically matching word vector. The reverse dictionary is useful in practical applications such as solving the tip-of-the-tongue problem, helping new language learners, etc. In this paper , we evaluate the Transformer-b...

User-generated content is full of misspellings. Rather than being just random noise, we hypothesise that many misspellings contain hidden semantics that can be leveraged for language understanding tasks. This paper presents a fine-grained annotated corpus of misspelling in Thai, together with an analysis of misspelling intention and its possible se...

The Internet Engineering Task Force (IETF) has developed many of the technical standards that underpin the Internet. The standards development process followed by the IETF is open and consensus-driven, but is inherently both a social and political activity, and latent influential structures might exist within the community. Exploring and understand...

In text-to-SQL tasks -- as in much of NLP -- compositional generalization is a major challenge: neural networks struggle with compositional generalization where training and test distributions differ. However, most recent attempts to improve this are based on word-level synthetic data or specific dataset splits to generate compositional biases. In...

We present a conversational management act (CMA) annotation schema for one-to-one tutorial dialogue sessions where a tutor uses an analogy to teach a student a concept. CMAs are more fine-grained sub-utterance acts compared to traditional dialogue act markup. The schema achieves an inter-annotator agreement (IAA) Cohen Kappa score of at least 0.66...

Moderation of reader comments is a significant problem for online news platforms. Here, we experiment with models for automatic moderation, using a dataset of comments from a popular Croatian newspaper. Our analysis shows that while comments that violate the moderation rules mostly share common linguistic and thematic features, their content varies...

Addressing the mismatch between natural language descriptions and the corresponding SQL queries is a key challenge for text-to-SQL translation. To bridge this gap, we propose an SQL intermediate representation (IR) called Natural SQL (NatSQL). Specifically, NatSQL preserves the core functionalities of SQL, while it simplifies the queries as follows...

Recently, there has been significant progress in studying neural networks for translating text descriptions into SQL queries under the zero-shot cross-domain setting. Despite achieving good performance on some public benchmarks, we observe that existing text-to-SQL models do not generalize when facing domain knowledge that does not frequently appea...

Dementia is a family of neurogenerative conditions affecting memory and cognition in an increasing number of individuals in our globally aging population. Automated analysis of language, speech and paralinguistic indicators have been gaining popularity as potential indicators of cognitive decline. Here we propose a novel longitudinal multi-modal da...

Computational humour is a difficult and overlooked challenge in natural language processing, yet humour remains an integral part of human communication and a recipe for interesting discourse. While considerable progress has been made in language synthesis, the nuanced task of humour generation transcends conventional syntactic accuracy and semantic...

The current dominance of deep neural networks in natural language processing is based on contextual embeddings such as ELMo, BERT, and BERT derivatives. Most existing work focuses on English; in contrast, we present here the first multilingual empirical comparison of two ELMo and several monolingual and multilingual BERT models using 14 tasks in ni...

We present two multimodal fusion-based deep learning models that consume ASR transcribed speech and acoustic data simultaneously to classify whether a speaker in a structured diagnostic task has Alzheimer's Disease and to what degree, evaluating the ADReSSo challenge 2021 data. Our best model, a BiLSTM with highway layers using words, word probabil...

Platforms that feature user-generated content (social media, online forums, newspaper comment sections etc.) have to detect and filter offensive speech within large, fast-changing datasets. While many automatic methods have been proposed and achieve good accuracies, most of these focus on the English language, and are hard to apply directly to lang...

Alzheimer’s disease (AD) is a progressive, neurodegenerative disorder mainly characterized by memory loss with deficits in other cognitive domains, including language, visuospatial abilities, and changes in behavior. Detecting diagnostic biomarkers that are noninvasive and cost-effective is of great value not only for clinical assessments and diagn...

This paper is a submission to the Alzheimer's Dementia Recognition through Spontaneous Speech (ADReSS) challenge, which aims to develop methods that can assist in the automated prediction of severity of Alzheimer's Disease from speech data. We focus on acoustic and natural language features for cognitive impairment detection in spontaneous speech i...

Recently, there has been significant progress in studying neural networks to translate text descriptions into SQL queries. Despite achieving good performance on some public benchmarks, existing text-to-SQL models typically rely on the lexical matching between words in natural language (NL) questions and tokens in table schemas, which may render the...

Despite the incremental nature of Dynamic Syntax (DS), the semantic grounding of it remains that of predicate logic, itself grounded in set theory, so is poorly suited to expressing the rampantly context-relative nature of word meaning, and related phenomena such as incremental judgements of similarity needed for the modelling of disambiguation. He...

This collection contains a selection of recent work on people’s production of figurative language (metaphoric, ironic, metonymic, hyperbolic, ...) and similarly of figurative expression in visual media and artefact design. The articles illuminate issues such as why and under what circumstances people produce figurative expression and how it is moul...

This paper presents the Graded Word Similarity in Context (GWSC) task which asked participants to predict the effects of context on human perception of similarity in English, Croatian, Slovene and Finnish. We received 15 submissions and 11 system description papers. A new dataset (CoSimLex) was created for evaluation in this task: it contains pairs...

This article describes initial work into the automatic classification of user-generated content in news media to support human moderators. We work with real-world data-comments posted by readers under online news articles-in two less-resourced European languages, Croatian and Estonian. We describe our dataset, and experiments into automatic classif...

In everyday conversation, no notion of “complete sentence” is required for syntactic licensing. However, so-called “fragmentary”, “incomplete”, and abandoned utterances are problematic for standard formalisms. When contextualised, such data show that (a) non-sentential utterances are adequate to underpin agent coordination, while (b) all linguistic...

We describe a set of experiments for building a temporal mental health dynamics system. We utilise a pre-existing methodology for distant-supervision of mental health data mining from social media platforms and deploy the system during the global COVID-19 pandemic as a case study. Despite the challenging nature of the task, we produce encouraging r...

In this paper, we explore the idea that independently developed Dynamic Syntax accounts of dialogue and interaction fit well within the general approach of radical embodied and enac-tive accounts of cognition (REEC). This approach enables a rethinking of the grounding of linguistic universal constraints, specifically tree structure restrictions, as...

We introduce an annotation scheme and corpus study to investigate the use of base and target components of analogies in tutorial dialogues. We present the development of the scheme and test its final form on a corpus of one-to-one tutorial dialogues on computer science , for which we achieve over 0.77 multi-rater inter-annotator agreement. We then...

We study the influence of context on sentence acceptability. First we compare the acceptability ratings of sentences judged in isolation, with a relevant context, and with an irrelevant context. Our results show that context induces a cognitive load for humans, which compresses the distribution of ratings. Moreover, in relevant contexts we observe...

State of the art natural language processing tools are built on context-dependent word embeddings, but no direct method for evaluating these representations currently exists. Standard tasks and datasets for intrinsic evaluation of embeddings are based on judgements of similarity, but ignore context; standard tasks for word sense disambiguation take...

We study the influence of context on sentence acceptability. First we compare the acceptability ratings of sentences judged in isolation, with a relevant context, and with an irrelevant context. Our results show that context induces a cognitive load for humans, which compresses the distribution of ratings. Moreover, in relevant contexts we observe...

State of the art natural language processing tools are built on context-dependent word embeddings, but no direct method for evaluating these representations currently exists. Standard tasks and datasets for intrinsic evaluation of embeddings are based on judgements of similarity, but ignore context; standard tasks for word sense disambiguation take...

Research on language and gender has a long tradition, and large electronic text corpora and novel computational methods for representing word meaning have recently opened new directions. We explain how gender can be analysed using word embeddings: vector representations of words computationally derived from lexical context in large corpora and capt...

This paper describes an initial corpus study of question-answer pairs in the Carolina Conversations Collection corpus of conversational interviews with older people. Our aim is to compare the behaviour of patients with and without Alzheimer's Disease (AD) on the basis of types of question asked and their responses in dialogue. It has been suggested...

While Voice User Interfaces (VUI) are becoming increasingly embedded into everyday life, their ability to tailor their output to individual users is limited. Research in VUIs has explored the use of static user models to encode general preferences; and, separately, dynamic models of dialogue context or short-term common ground have been used to inf...

In this paper, we present a novel context-dependent approach to modeling word meaning, and apply it to the modeling of metaphor. In distributional semantic approaches, words are represented as points in a high dimensional space generated from co-occurrence statistics; the distances between points may then be used to quantifying semantic relationshi...

Computational creativity seeks to understand computational mechanisms that can be characterized as creative. The creation of new concepts is a central challenge for any creative system. In this article, we outline different approaches to computational concept creation and then review conceptual representations relevant to concept creation, and ther...

We examine the benefit of a variety of discourse and semantic features for the identification of summary-worthy content in narrative stories. Using logistic regression models, we find that the most informative features are those that relate to the narrative structure of a text. We show that automatic methods for feature extraction perform significa...

This book constitutes the proceedings of the 7th International Conference on Statistical Language and Speech Processing, SLSP 2019, held in Ljubljana, Slovenia, in October 2019. The 25 full papers presented together with one invited paper in this volume were carefully reviewed and selected from 48 submissions. They were organized in topical section...

One of the fundamental requirements for models of semantic processing in dialogue is incrementality: a model must reflect how people interpret and generate language at least on a word-by-word basis, and handle phenomena such as fragments, incomplete and jointly-produced utterances. We show that the incremental word-by-word parsing process of Dynami...