
Overview
Background
Martin Schweinberger uses big data and computational methods to explore the messy, fascinating reality of how people actually talk—including all the swear words, filler words, and informal expressions that traditional language education overlooks. As a Lecturer in Applied Linguistics at the University of Queensland, he bridges the gap between computer science and linguistics to understand how language evolves in our digital age.
Uncovering Hidden Language Patterns
Much of Martin's research focuses on the language phenomena that schools don't teach but that permeate everyday conversation. He analyzes massive datasets to study vulgarity and swearing patterns, as well as discourse markers—those ubiquitous filler words like "like," "you know," "well," and "I mean" that pepper our speech. By applying statistical methods to real-world language use, he reveals how these supposedly "incorrect" forms of expression actually follow sophisticated social and linguistic rules.
His work also tracks how language changes over time and varies between different social settings, using computational tools to identify patterns that would be impossible to detect through traditional research methods alone.
Building Australia's Language Data Future
As Director of the Language Technology and Data Analysis Laboratory (LADAL)—a free upskilling platform for language data science with hundreds of thousands of users worldwide—and a key figure in one of Australia's major research infrastructure projects, the Language Data Commons of Australia (LDaCA), Martin is helping build the digital infrastructure that will support language research across the country. LDaCA has received substantial funding to create accessible tools and resources that allow researchers to analyze text and speech data more effectively.
Championing Research Transparency
Beyond his linguistic research, Martin advocates for reproducibility and transparency in humanities and social science research. He provides guidance on how language researchers can adopt more rigorous, open research practices—addressing a growing concern about the reliability of academic findings across disciplines.
Martin's international visibility is reflected in his leadership roles: he serves as Vice-President Professional of the International Society for the Linguistics of English (ISLE) and sits on the board of The International Computer Archive of Modern and Medieval English (ICAME), one of the oldest and most reputable societies for corpus linguistics. These positions demonstrate his commitment to advancing computational language research on a global scale.
Potential topics for supervision
I would be particularly interested in supervising theses on the following topics:
Sociolinguistics / Language Variation and Change / World Englishes
- General extenders
- Terms-of-address and salutations
- Discourse particles and markers
- Vulgarity
- Adjective amplification
Learner Language / Applied Linguistics / Corpus Phonetics / Learner Corpus Research
- Vowel production among L1 speakers and learners of English
- Voice-onset-times among L1 speakers and learners of English
- Fluency and pauses in learner and L1 speech.
- Accent and intelligibility / comprehension.
Text Analytics / Digital Humanities / Corpus Linguistics
- Applied word embedding applications in the language sciences.
- Comparison of different association / keyness measures
Availability
- Dr Martin Schweinberger is:
- Available for supervision
- Media expert
Fields of research
Qualifications
- Doctor of Philosophy, Universität Hamburg
Research interests
-
Vulgarity and Swearing
I investigate how swear words and taboo language are used in everyday speech and online discourse. Contrary to popular belief, vulgar language follows systematic social and linguistic rules. My research uncovers how such expressions function in communication and what they reveal about speakers’ identities, emotions, and group memberships.
-
Discourse Markers and Filler Words
I study words like like, you know, and well—terms often dismissed as meaningless. Using computational analysis, I show how these elements structure conversations and convey nuanced meanings. My work demonstrates that such "filler" words play important roles in signaling attitudes, managing interactions, and guiding listener expectations.
-
Open Science and Research Transparency
I actively promote reproducible, open research practices in the humanities and social sciences. I provide practical training and resources to help language researchers adopt transparent workflows. My advocacy supports greater academic rigor and long-term trust in empirical research.
-
Text Analytics and Computational Linguistics
I apply computational methods—like machine learning and statistical modelling—to large corpora to uncover hidden linguistic patterns. These tools help quantify language use in a way that supports replicable, empirical research. My work is at the intersection of computer science and linguistics, making it especially relevant in the digital age.
-
Digital Infrastructure and Research Tools
As Director of LADAL and a lead in LDaCA, I am building accessible digital platforms that support large-scale language analysis. These initiatives democratize access to language data and computational tools for researchers, students, and educators alike. My infrastructure work enhances the capacity for advanced language research in Australia and beyond.
-
Language Variation and Change
I explore how language evolves over time and across different social settings. By analyzing large-scale linguistic datasets, I identify subtle patterns of variation in how people speak, particularly in informal and digital contexts. This research helps reveal how social norms and technology influence the way we communicate.
-
Learner Language and Second Language Acquisition
I analyze how learners of English produce sounds, manage fluency, and develop pronunciation over time. This includes examining features like vowel quality, voice-onset time, pauses, and accent intelligibility. By comparing learner and native speaker data, my research informs language teaching and helps improve learner outcomes.
-
Corpus Phonetics
I use corpus-based methods to investigate the phonetic characteristics of spoken language, including pronunciation patterns among both native and non-native speakers. I focus on measurable acoustic features such as vowel production and timing cues. This approach allows for the large-scale, data-driven analysis of speech in real-life settings.
Research impacts
As director and initiator of the Language Technology and Data Analysis Laboratory (LADAL) I am very proud that LADAL has emerged as one of Australia’s most prominent web-based collaborative support infrastructures for digital and computational humanities with more than 1.1 million page views of more than 500,000 active users in nearly 750,000 engaged sessions since 2021.
Works
Search Professor Martin Schweinberger’s works on UQ eSpace
2023
Conference Publication
A corpus-based acoustic analysis of vowel production by L1-Chinese learners and native speakers of English
Schweinberger, Martin and Yin, Ruihua (2023). A corpus-based acoustic analysis of vowel production by L1-Chinese learners and native speakers of English. 7th Meeting of the International Society for the Linguistics of English (ISLE7), Brisbane, QLD, Australia, 19 - 23 June 2023. Brisbane, QLD, Australia: University of Queensland.
2023
Other Outputs
Data in the humanities with text analytics
Schweinberger, Martin and Hames, Sam (2023). Data in the humanities with text analytics. Brisbane, QLD, Australia: The University of Queensland.
2023
Conference Publication
A corpus-based acoustic analysis of vowel production by L1-Chinese learners and native speakers of English
Schweinberger, Martin and Yin, Rui (2023). A corpus-based acoustic analysis of vowel production by L1-Chinese learners and native speakers of English. 44th Meeting of the International Computer Archive of Modern and Medieval English (ICAME44), Vanderbijlpark, South Africa, 17 - 21 May 2023. Brisbane, QLD, Australia: University of Queensland.
2023
Other Outputs
Text Analytics with Language Technology and Data Analysis Laboratory Resources: an introduction to free, open-source interactive resources for linguists
Schweinberger, Martin (2023). Text Analytics with Language Technology and Data Analysis Laboratory Resources: an introduction to free, open-source interactive resources for linguists. Hamburg, Germany: University of Hamburg.
2023
Other Outputs
An introduction to conditional inference trees in R
Schweinberger, Martin (2023). An introduction to conditional inference trees in R. Bonn, Germany: Rheinische Friedrich-Wilhelms-University.
2022
Conference Publication
A corpus-based computational analysis of high-front and -back vowel production of L1-Japanese learners of English and L1-English speakers
Schweinberger, Martin and Komiya, Yuki (2022). A corpus-based computational analysis of high-front and -back vowel production of L1-Japanese learners of English and L1-English speakers. Australasian International Conference on Speech Science and Technology, Canberra, ACT, Australia, 13 - 16 December 2022. Canberra, ACT, Australia: Australasian Speech Science and Technology Association.
2022
Conference Publication
Exploring powerful tools to ensure robust and reproducible results in corpus linguistics
Schweinberger, Martin, Flanaghan, Joseph and Schneider, Gerold (2022). Exploring powerful tools to ensure robust and reproducible results in corpus linguistics. 43rd Meeting of the International Computer Archive of Modern and Medieval English (ICAME43), Cambridge, United Kingdom, 18 - 21 August 2021. Cambridge, United Kingdom: TU Dortmund University.
2022
Conference Publication
Research trends in corpus linguistics: a bibliometric analysis of two decades of Scopus-indexed corpus linguistics research in arts and humanities
Crosthwaite, P., Ningrum, S. and Schweinberger, M. (2022). Research trends in corpus linguistics: a bibliometric analysis of two decades of Scopus-indexed corpus linguistics research in arts and humanities. ICAME43, Cambridge, United Kingdom, 27-30 July 2022.
2022
Other Outputs
From tables to forests – working with tables and tree-based models
Schweinberger, Martin (2022). From tables to forests – working with tables and tree-based models. Tromsø, Norway: The Arctic University of Norway.
2022
Other Outputs
Introduction to Power Analysis with R
Schweinberger, Martin (2022). Introduction to Power Analysis with R. Tromsø, Norway: The Arctic University of Norway.
2022
Other Outputs
Introduction to data visualization with R
Schweinberger, Martin (2022). Introduction to data visualization with R. Tromsø, Norway: The Arctic University of Norway.
2022
Book Chapter
Absolutely fantastic and really really good
Schweinberger, Martin (2022). Absolutely fantastic and really really good. Expanding the landscapes of Irish English research. (pp. 129-145) edited by Stephen Lucek and Carolina P. Amador-Moreno. New York, NY, United States: Routledge. doi: 10.4324/9781003025078-7
2021
Journal Article
Ongoing change in the Australian English amplifier system
Schweinberger, Martin (2021). Ongoing change in the Australian English amplifier system. Australian Journal of Linguistics, 41 (2), 166-194. doi: 10.1080/07268602.2021.1931028
2021
Journal Article
Training disciplinary genre awareness through blended learning: an exploration into EAP students’ perceptions of online annotation of genres across disciplines
Crosthwaite, Peter, Sanhueza, Alicia Gazmuri and Schweinberger, Martin (2021). Training disciplinary genre awareness through blended learning: an exploration into EAP students’ perceptions of online annotation of genres across disciplines. Journal of English for Academic Purposes, 53 101021, 1-16. doi: 10.1016/j.jeap.2021.101021
2021
Journal Article
Which word gets the nuclear stress in a turn-at-talk?
Ruhlemann, Christoph and Schweinberger, Martin (2021). Which word gets the nuclear stress in a turn-at-talk?. Journal of Pragmatics, 178, 426-439. doi: 10.1016/j.pragma.2021.04.005
2021
Journal Article
Voices from the periphery: perceptions of Indonesian primary vs secondary pre-service teacher trainees about corpora and data-driven learning in the L2 English classroom
Crosthwaite, Peter, Luciana and Schweinberger, Martin (2021). Voices from the periphery: perceptions of Indonesian primary vs secondary pre-service teacher trainees about corpora and data-driven learning in the L2 English classroom. Applied Corpus Linguistics, 1 (1) 100003, 1-13. doi: 10.1016/j.acorp.2021.100003
2021
Journal Article
Analyzing Historical Changes in the Irish English Amplifier System
Schweinberger, M. (2021). Analyzing Historical Changes in the Irish English Amplifier System. Anglistik, 32 (1), 139-158. doi: 10.33675/angl/2021/1/11
2021
Other Outputs
Fixed- and mixed-effects regression models in R
Schweinberger, Martin (2021). Fixed- and mixed-effects regression models in R. Brisbane, QLD, Australia: The University of Queensland, School of Languages and Cultures.
2021
Journal Article
Analysing discourse around COVID-19 in the Australian Twittersphere: a real-time corpus-based analysis
Schweinberger, Martin, Haugh, Michael and Hames, Sam (2021). Analysing discourse around COVID-19 in the Australian Twittersphere: a real-time corpus-based analysis. Big Data and Society, 8 (1), 205395172110214. doi: 10.1177/20539517211021437
2021
Book Chapter
Using intensifier-adjective collocations to investigate mechanisms of change
Schweinberger, Martin (2021). Using intensifier-adjective collocations to investigate mechanisms of change. Variation in time and space: observing the world through corpora. (pp. 231-255) edited by Anna Čermáková and Markéta Malá. Berlin, Germany: De Gruyter. doi: 10.1515/9783110604719-010
Supervision
Availability
- Dr Martin Schweinberger is:
- Available for supervision
Before you email them, read our advice on how to contact a supervisor.
Supervision history
Current supervision
-
Doctor Philosophy
A multifactorial study of morpho-syntactic errors across different L1 backgrounds and language proficiency levels
Principal Advisor
Other advisors: Associate Professor Peter Crosthwaite
-
Doctor Philosophy
A corpus-based analysis of conspiracy theory discourse on Reddit: Understanding conspiracy-fuelled anomie and moral panics during COVID-19
Principal Advisor
Other advisors: Professor Ryan Ko
-
Doctor Philosophy
Integrating Artificial Intelligence and Machine Learning in TESOL: A Study on Personalised Learning and Impact on Student Engagement and Motivation in A Rural Indonesian University
Associate Advisor
Other advisors: Associate Professor Peter Crosthwaite
-
Doctor Philosophy
Corpus-based investigation of three-minute thesis presentations: Register perspective
Associate Advisor
Other advisors: Associate Professor Peter Crosthwaite
-
Doctor Philosophy
The Relationship Between Writing Tasks and Second Language Writers¿ Use of Metadiscourse
Associate Advisor
Other advisors: Associate Professor Peter Crosthwaite
Completed supervision
-
2023
Doctor Philosophy
The acquisition of number marking: The case of Indonesian as a second language
Associate Advisor
Other advisors: Associate Professor Peter Crosthwaite
Media
Enquiries
Contact Dr Martin Schweinberger directly for media enquiries about their areas of expertise.
Need help?
For help with finding experts, story ideas and media enquiries, contact our Media team: