Dirk Speelman
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Dirk Speelman has a total of 87 co-authors. Here are the top co-authors by number of shared publications:
Dirk Geeraerts
Department of Linguistics , KU Leuven , Blijde-Inkomststraat 21, PO-box 3308 , Leuven 3000 , Belgium
38 shared publications- Dirk Geeraerts
University of Leuven
12 shared publications - Eline Zenner
KU Leuven
10 shared publications - Dirk Geeraerts
Linguistics, University of Leuven
6 shared publications - Laura Rosseel
Vrije Universiteit Brussel
5 shared publications Stefan Grondelaers
5 shared publications
Latest Publications
- Open Access
- 31 July 2026
- Published by Ubiquity Press, Ltd. in Journal of Open Humanities Data
- Vol. 12 (1)
Evaluating the Evaluator: Problems With SemEval-2020 Task 1 for Lexical Semantic Change Detection
This discussion paper re-examines SemEval-2020 Task 1, the most influential shared benchmark for lexical semantic change (LSC) detection, through a three-part evaluative framework: operationalisation, data quality, and benchmark design. First, at the level of operationalisation, the benchmark models semantic change mainly as gain, loss, or redistribution of discrete senses. We argue that while practical for annotation and evaluation, this framing is too narrow to capture gradual, constructional, collocational, and discourse-level change. Also, the gold labels are outcomes of annotation decisions, clustering procedures, and threshold settings, which could potentially limit the validity of the task. Second, at the level of data quality, we show that the benchmark is affected by substantial corpus and preprocessing problems, including Optical Character Recognition (OCR) noise, malformed characters, truncated sentences, inconsistent lemmatisation, Part-of-Speech (POS) tagging errors, and missed targets. These issues can distort model behaviour, complicate linguistic analysis, and reduce reproducibility. Third, at the level of benchmark design, we argue the small curated target sets and limited language coverage reduce realism and increase statistical uncertainty. Taken together, these limitations suggest that the benchmark should be treated as a useful but partial test bed rather than a definitive measure of progress. We therefore call for future datasets and shared tasks to adopt broader theories of semantic change, document preprocessing transparently, expand cross-linguistic coverage, and use more realistic evaluation settings. Such steps are necessary for more valid, interpretable, and generalisable progress in lexical semantic change detection.
- 31 July 2026
- Published by Ubiquity Press, Ltd. in Journal of Open Humanities Data
- Vol. 12 (1)
Open AccessEvaluating the Evaluator: Problems With SemEval-2020 Task 1 for Lexical Semantic Change Detection
This discussion paper re-examines SemEval-2020 Task 1, the most influential shared benchmark for lexical semantic change (LSC) detection, through a three-part evaluative framework: operationalisation, data quality, and benchmark design. First, at the level of operationalisation, the benchmark models semantic change mainly as gain, loss, or redistribution of discrete senses. We argue that while practical for annotation and evaluation, this framing is too narrow to capture gradual, constructional, collocational, and discourse-level change. Also, the gold labels are outcomes of annotation decisions, clustering procedures, and threshold settings, which could potentially limit the validity of the task. Second, at the level of data quality, we show that the benchmark is affected by substantial corpus and preprocessing problems, including Optical Character Recognition (OCR) noise, malformed characters, truncated sentences, inconsistent lemmatisation, Part-of-Speech (POS) tagging errors, and missed targets. These issues can distort model behaviour, complicate linguistic analysis, and reduce reproducibility. Third, at the level of benchmark design, we argue the small curated target sets and limited language coverage reduce realism and increase statistical uncertainty. Taken together, these limitations suggest that the benchmark should be treated as a useful but partial test bed rather than a definitive measure of progress. We therefore call for future datasets and shared tasks to adopt broader theories of semantic change, document preprocessing transparently, expand cross-linguistic coverage, and use more realistic evaluation settings. Such steps are necessary for more valid, interpretable, and generalisable progress in lexical semantic change detection.
- Open Access
- 6 July 2026
- Published by Walter de Gruyter GmbH in Cognitive Linguistics
- Vol. 37 (3) , 353-385
Salience, relevance of social cues and the learning of sociolinguistic variation: methodological lessons from an artificial language learning experiment
Usage-based accounts of sociolinguistic learning assign a pivotal role to the frequency with which a linguistic variant and a social category co-occur in listeners’ input. However, few studies have so far compared the effects of varying frequency distributions on acquisition in a lab-based context. This paper presents a critical assessment of methodological implications and considerations with regard to investigating such questions, thereby focusing on the upcoming paradigm of artificial sociolinguistic language learning. Based on the available literature on this paradigm, we designed a semi-artificial language learning experiment that manipulates the probabilistic association between a new linguistic variable and macro-social categories between participants. In a first training phase, participants were exposed to the variable – (de)voicing of the word medial stop in disyllabic pseudowords – produced by two male and two female sounding speakers (social categories of ‘gender’). Participants’ learning of the link between the linguistic variable and perceived speaker gender was subsequently tested in a perception task which also tested the generalisation of learning to words that participants did not hear during training. Results show relatively limited learning overall, yet the experiment points to key methodological insights, particularly regarding the role of salience and relevance in lab-based learning tasks.
- 6 July 2026
- Published by Walter de Gruyter GmbH in Cognitive Linguistics
- Vol. 37 (3) , 353-385
Open AccessSalience, relevance of social cues and the learning of sociolinguistic variation: methodological lessons from an artificial language learning experiment
Usage-based accounts of sociolinguistic learning assign a pivotal role to the frequency with which a linguistic variant and a social category co-occur in listeners’ input. However, few studies have so far compared the effects of varying frequency distributions on acquisition in a lab-based context. This paper presents a critical assessment of methodological implications and considerations with regard to investigating such questions, thereby focusing on the upcoming paradigm of artificial sociolinguistic language learning. Based on the available literature on this paradigm, we designed a semi-artificial language learning experiment that manipulates the probabilistic association between a new linguistic variable and macro-social categories between participants. In a first training phase, participants were exposed to the variable – (de)voicing of the word medial stop in disyllabic pseudowords – produced by two male and two female sounding speakers (social categories of ‘gender’). Participants’ learning of the link between the linguistic variable and perceived speaker gender was subsequently tested in a perception task which also tested the generalisation of learning to words that participants did not hear during training. Results show relatively limited learning overall, yet the experiment points to key methodological insights, particularly regarding the role of salience and relevance in lab-based learning tasks.
- Open Access
- 27 May 2025
- Published by John Benjamins Publishing Company
Indexing traditional and modern professional values
Traditional professional values are complemented by more modern ones in the “New Work Order”-context, roughly marking a shift from hierarchy to (seeming) egalitarianism and from formality to informality. We investigate the pragmatic uses of T/V-pronouns in job interviews in relation to this transformation. More particularly, we zoom in on the Flemish case, which is especially rich, given that next to V-forms, there is a double-barreled T-system with standard versus colloquial T-forms. Exploring twenty-one authentic job interviews using exploratory statistical techniques, we found that V-forms, emblematic of traditional values, are nearly absent in our data. In turn, the variation attested in the use of T-forms by recruiters highlights local meaning-making, prompting questions about the alignment of these forms with evolving indexical values and workplace norms.
- 27 May 2025
- Published by John Benjamins Publishing Company
Open AccessIndexing traditional and modern professional values
Traditional professional values are complemented by more modern ones in the “New Work Order”-context, roughly marking a shift from hierarchy to (seeming) egalitarianism and from formality to informality. We investigate the pragmatic uses of T/V-pronouns in job interviews in relation to this transformation. More particularly, we zoom in on the Flemish case, which is especially rich, given that next to V-forms, there is a double-barreled T-system with standard versus colloquial T-forms. Exploring twenty-one authentic job interviews using exploratory statistical techniques, we found that V-forms, emblematic of traditional values, are nearly absent in our data. In turn, the variation attested in the use of T-forms by recruiters highlights local meaning-making, prompting questions about the alignment of these forms with evolving indexical values and workplace norms.
- Open Access
- 23 December 2024
- Published by Walter de Gruyter GmbH in Corpus Linguistics and Linguistic Theory
- Vol. 22 (1) , 139-181
Investigating lexical-semantic effects on morphosyntactic variation using elastic net regression
This article showcases elastic net regression as a means to build fairer models of morphosyntactic variation. Elastic net allows lexical items to appear on the same level as traditional, high-level predictors, enabling fuller models of variation. We apply elastic net regression to 1,296,574 Dutch verbal cluster tokens from the SoNaR corpus, analysing a morphosyntactic alternance in Dutch subordinate clauses. Our results show morphosyntactic preferences among verbs, indicating that semantic effects are indeed at play. Further analysis shows that semantic patterns for either word order exist, though it remains difficult to glean any semantic generalisations. Still, the elastic net technique shows that the inclusion of lexical items as full predictors in a model is useful, as much of the variation left unexplained by high-level predictors can be explained in lexical terms.
- 23 December 2024
- Published by Walter de Gruyter GmbH in Corpus Linguistics and Linguistic Theory
- Vol. 22 (1) , 139-181
Open AccessInvestigating lexical-semantic effects on morphosyntactic variation using elastic net regression
This article showcases elastic net regression as a means to build fairer models of morphosyntactic variation. Elastic net allows lexical items to appear on the same level as traditional, high-level predictors, enabling fuller models of variation. We apply elastic net regression to 1,296,574 Dutch verbal cluster tokens from the SoNaR corpus, analysing a morphosyntactic alternance in Dutch subordinate clauses. Our results show morphosyntactic preferences among verbs, indicating that semantic effects are indeed at play. Further analysis shows that semantic patterns for either word order exist, though it remains difficult to glean any semantic generalisations. Still, the elastic net technique shows that the inclusion of lexical items as full predictors in a model is useful, as much of the variation left unexplained by high-level predictors can be explained in lexical terms.
- 12 December 2024
- Published by Elsevier in Language Sciences
- Vol. 108
Unravelling the competing dynamics of Chinese causative markers shi 使, ling 令, jiao1 叫 and jiao2 教: A diachronic analysis
- 12 December 2024
- Published by Elsevier in Language Sciences
- Vol. 108
Unravelling the competing dynamics of Chinese causative markers shi 使, ling 令, jiao1 叫 and jiao2 教: A diachronic analysis
- 9 January 2024
- Published by Walter de Gruyter GmbH in Linguistics Vanguard
- Vol. 10 (s1) , 1-7
Introduction: what are alternations and how should we study them?
The research paradigm of alternation studies is forming an increasingly large share of the empirical foundations of usage-based linguistics. As the paradigm is essentially an amalgamation of research traditions from various subfields of linguistics, including sociolinguistics, psycholinguistics, cognitive linguistics, and construction grammar, it sports various definitions of the concept of “alternation” and various ways of conducting an alternation study, which are not always compatible. The present special issue is therefore intended to bring researchers from various backgrounds in usage-based linguistics together to see how we can deal with these issues. This introduction first presents the various ways of defining an alternation and discusses the differences between them and how these definitions determine the methodological set-up of an alternation study. Next, the contributions to the special issue are each in turn summarized and related to one another.
- 9 January 2024
- Published by Walter de Gruyter GmbH in Linguistics Vanguard
- Vol. 10 (s1) , 1-7
Introduction: what are alternations and how should we study them?
The research paradigm of alternation studies is forming an increasingly large share of the empirical foundations of usage-based linguistics. As the paradigm is essentially an amalgamation of research traditions from various subfields of linguistics, including sociolinguistics, psycholinguistics, cognitive linguistics, and construction grammar, it sports various definitions of the concept of “alternation” and various ways of conducting an alternation study, which are not always compatible. The present special issue is therefore intended to bring researchers from various backgrounds in usage-based linguistics together to see how we can deal with these issues. This introduction first presents the various ways of defining an alternation and discusses the differences between them and how these definitions determine the methodological set-up of an alternation study. Next, the contributions to the special issue are each in turn summarized and related to one another.
- Open Access
- 9 February 2023
- Published by Cambridge University Press (CUP) in Journal of Germanic Linguistics
- Vol. 35 (1) , 1-68
Natiolectal Variation in Dutch Morphosyntax: A Large-Scale, Data-Driven Perspective
In this article, we report a large-scale corpus study aimed at tackling the (controversial) question to what extent the European national varieties of Dutch, that is, Belgian and Netherlandic Dutch, exhibit morpho-syntactic differences. Instead of relying on a manual selection of cases of morphosyntactic variation, we first marshal large bilingual parallel corpora and machine translation software to identify semiautomatically, in an extensively data-driven fashion, loci of variation from various “corners” of Dutch grammar. We then gauge the distribution of con-structional alternatives in a nationally as well as stylistically stratified corpus for a representative selection of twenty alternation patterns. We find that natiolectal variation in the grammar of Dutch is far more prevalent than often assumed, especially in less edited text types, and that it shows up in inflection phenomena, lexically conditioned syntactic variation, and pure word order permutations. Another key finding is that many cases of synchronic probabilistic asymmetries reflect a diachronic difference between the two varieties: Netherlandic Dutch often tends to be ahead in cases of ongoing grammatical change, with Belgian Dutch holding on somewhat longer to obsolescent features of the grammar.*
- 9 February 2023
- Published by Cambridge University Press (CUP) in Journal of Germanic Linguistics
- Vol. 35 (1) , 1-68
Open AccessNatiolectal Variation in Dutch Morphosyntax: A Large-Scale, Data-Driven Perspective
In this article, we report a large-scale corpus study aimed at tackling the (controversial) question to what extent the European national varieties of Dutch, that is, Belgian and Netherlandic Dutch, exhibit morpho-syntactic differences. Instead of relying on a manual selection of cases of morphosyntactic variation, we first marshal large bilingual parallel corpora and machine translation software to identify semiautomatically, in an extensively data-driven fashion, loci of variation from various “corners” of Dutch grammar. We then gauge the distribution of con-structional alternatives in a nationally as well as stylistically stratified corpus for a representative selection of twenty alternation patterns. We find that natiolectal variation in the grammar of Dutch is far more prevalent than often assumed, especially in less edited text types, and that it shows up in inflection phenomena, lexically conditioned syntactic variation, and pure word order permutations. Another key finding is that many cases of synchronic probabilistic asymmetries reflect a diachronic difference between the two varieties: Netherlandic Dutch often tends to be ahead in cases of ongoing grammatical change, with Belgian Dutch holding on somewhat longer to obsolescent features of the grammar.*
- 1 January 2023
- Published by International Quantitative Linguistics Association in Glottometrics
- Vol. 55, 17-43
Applying Distributional Semantic Models to a Historical Corpus of a Highly Inflected Language: the Case of Ancient Greek
- 1 January 2023
- Published by International Quantitative Linguistics Association in Glottometrics
- Vol. 55, 17-43
Applying Distributional Semantic Models to a Historical Corpus of a Highly Inflected Language: the Case of Ancient Greek
- 10 February 2022
- Published by Cambridge University Press (CUP) in Behavioral and Brain Sciences
- Vol. 45, e34
Generalizability in mixed models: Lessons from corpus linguistics
Part of the generalizability issues that haunt controlled lab experiment designs in psychology, and more particularly in psycholinguistics, can be alleviated by adopting corpus linguistic methods. These work with natural data. This advantage comes at a cost: in corpus studies, lexemes and language users can show different kinds of skew. We discuss a number of solutions to bolster the control.
- 10 February 2022
- Published by Cambridge University Press (CUP) in Behavioral and Brain Sciences
- Vol. 45, e34
Generalizability in mixed models: Lessons from corpus linguistics
Part of the generalizability issues that haunt controlled lab experiment designs in psychology, and more particularly in psycholinguistics, can be alleviated by adopting corpus linguistic methods. These work with natural data. This advantage comes at a cost: in corpus studies, lexemes and language users can show different kinds of skew. We discuss a number of solutions to bolster the control.