<?xml version="1.0" encoding="UTF-8"?>
<rss  xmlns:atom="http://www.w3.org/2005/Atom" 
      xmlns:media="http://search.yahoo.com/mrss/" 
      xmlns:content="http://purl.org/rss/1.0/modules/content/" 
      xmlns:dc="http://purl.org/dc/elements/1.1/" 
      version="2.0">
<channel>
<title>Matt Coler</title>
<link>https://mattcoler.github.io/news.html</link>
<atom:link href="https://mattcoler.github.io/news.xml" rel="self" type="application/rss+xml"/>
<description>Speech technology, AI ethics and language diversity. Associate Professor, University of Groningen.</description>
<generator>quarto-1.11.9</generator>
<lastBuildDate>Wed, 04 Nov 2026 00:00:00 GMT</lastBuildDate>
<item>
  <title>Standing to Be Heard — Keynote at Ege University</title>
  <link>https://mattcoler.github.io/posts/2026-11-04-standing-to-be-heard-ege-university/</link>
  <description><![CDATA[ 





<p>On November 4 2026, I gave the online keynote “Standing to Be Heard: Speech Technology and the Community of Speakers” at Human in the Age of Algorithms, a hybrid conference hosted by Ege University in İzmir.</p>
<p>The talk picks up where <a href="../../posts/2026-06-09-marginalized-knowledge-systems-warsaw/index.html">Who Governs the Spoken Word?</a> left off in Warsaw. There I mapped the governance gaps around speech tech for minoritized languages. Here I propose a way to close one of them.</p>
<blockquote class="blockquote">
<p>Speech technology decides who gets heard in public agencies, classrooms and courts. The law governing it was written earlier, for other problems, and assumes the claimant is an individual with a grievance. But when speech technology fails, it fails everyone who speaks a given way. The law recognizes the harm but not the claimant.</p>
<p>The talk tests that gap against two cases. West Frisian is absent from the speech recognition systems in widest use, and its speakers fall between two instruments. The European Charter for Regional or Minority Languages recognizes them but offers no judicial remedy. The EU AI Act is enforceable but gives no group a right to complain; the European Parliament’s 2023 draft included that right, and negotiations removed it. Türkiye’s data protection law, built on the individual data subject, reproduces the same gap. In the second case, Wikimedia refused a Tasmanian Aboriginal request over palawa kani vocabulary on the ground that a language cannot be owned.</p>
<p>I argue for a category the law does not have: the community of speakers as a rights-holder, entitled to consultation before deployment, consent over the corpus, and standing to be heard once a system runs.</p>
</blockquote>
<p>Thanks to the organizers for the invitation. <img src="https://i.postimg.cc/7ZFL3CMF/Standing-To-Be-Heard-Keynote.jpg" class="img-fluid" alt="Standing to Be Heard keynote at Ege University"></p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/T1Z9SFJM/Screenshot-2026-11-04-at-13-07-17.png" class="img-fluid figure-img"></p>
<figcaption>certificate</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Keynote</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-11-04-standing-to-be-heard-ege-university/</guid>
  <pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/7ZFL3CMF/Standing-To-Be-Heard-Keynote.jpg" medium="image" type="image/jpeg"/>
</item>
<item>
  <title>Middendorf’s Aymara Grammar as a Dated Record — talk at the Goethe-Institut in La Paz</title>
  <link>https://mattcoler.github.io/posts/2026-10-30-middendorf-aymara-grammar-la-paz/</link>
  <description><![CDATA[ 





<p>On 30 October 2026, I took part in a round table at the Goethe-Institut La Paz, Bolivia, on Ernst W. Middendorf’s <em>Die Aimará-Sprache</em> (Leipzig, 1891). The occasion was the book’s Spanish translation by Boris Chamani Velasco and was attended even by representatives of the German consulate.</p>
<p>Middendorf compiled his grammar in La Paz in 1887–88. He worked mainly with one bilingual consultant and checked what he heard, form by form, against Ludovico Bertonio’s grammar of 1612. My argument was that this makes the book a dated record of La Paz Aymara, sitting between the colonial sources and the language as it’s spoken today.</p>
<p>I looked at three examples: his treatment of the distal past <em>-tayna</em>, the vowel-deletion patterns he records, and his observation that Spanish was already shaping the use of the illative for human objects.</p>
<p>I closed with some proposals for the edition. Brief notes in plain Spanish for Aymara readers. Middendorf’s spelling kept alongside the modern alphabet. And the Introduction published in full, with a critical preliminary study.</p>
<p>It was a pleasure to see this book find new readers in the city where it was written, 135 years on.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/ZRVFQZ77/855503771-1263294772653879-6538914759306802905-n.jpg" class="img-fluid figure-img"></p>
<figcaption>Aymara_talk</figcaption>
</figure>
</div>



 ]]></description>
  <category>Aymara</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-10-30-middendorf-aymara-grammar-la-paz/</guid>
  <pubDate>Fri, 30 Oct 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/ZRVFQZ77/855503771-1263294772653879-6538914759306802905-n.jpg" medium="image" type="image/jpeg"/>
</item>
<item>
  <title>PLURILINGMEDIA ad hoc group for AI in Valletta, Malta</title>
  <link>https://mattcoler.github.io/posts/2026-10-06-plurilingmedia-ai-committee-valletta/</link>
  <description><![CDATA[ 





<p>On 5–6 October 2026, I joined the AI Ad Hoc Committee of <a href="https://plurilingmedia.eu/">PLURILINGMEDIA</a> in Valletta for its working meeting, hosted by Claudia Borg at the University of Malta.</p>
<p>The meeting picked up where my September lecture, <a href="../../posts/2026-09-14-plurilingmedia-ai-committee/index.html">The Governance Gap in AI for Minoritized Languages</a>, left off. That talk argued that when AI fails a minoritized language, hardly anyone can act on it. In Valletta the committee started turning that problem into something it can recommend.</p>
<p>Each of us brought one slide on the AI development most affecting media in our language, and we compared notes with colleagues from NPLD and MIDAS. The second day went to drafting shared principles for AI in minority-language media and planning the committee’s final report.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/fknrxr84/Screenshot-2026-10-08-at-12-13-35.png" class="img-fluid figure-img"></p>
<figcaption>PLURILINGMEDIA AI Ad Hoc Committee in Valletta</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-10-06-plurilingmedia-ai-committee-valletta/</guid>
  <pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/fknrxr84/Screenshot-2026-10-08-at-12-13-35.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Two years as Interspeech Ethics Chair</title>
  <link>https://mattcoler.github.io/posts/2026-10-01-interspeech-2026-ethics-2026/</link>
  <description><![CDATA[ 





<p>Interspeech 2026 is wrapping up in Sydney. I followed it from Amsterdam.</p>
<p>A friend in the audience sent me this photo from the committee acknowledgements: the organizing team lined up on stage, each role taking its turn on the big screen. Ethics got its turn too, even with its chair half a world away.</p>
<p>I’ve served as Ethics Chair for Interspeech 2025 and 2026. Most of that work happens out of sight: reviewing papers with ethics concerns, drafting policy, writing the occasional delicate email. It’s good to see it recognized alongside everything else that makes a conference this size run.</p>
<p>Thanks to the Sydney team for a great conference, and to everyone who answered ethics queries with patience over these two years.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><a href="https://postimg.cc/gxNMMSVc"><img src="https://i.postimg.cc/BvGdnRtF/Matt-IS26.jpg" class="img-fluid figure-img"></a></p>
<figcaption>The Interspeech 2026 organizing committee on stage in Sydney during the committee acknowledgements</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <guid>https://mattcoler.github.io/posts/2026-10-01-interspeech-2026-ethics-2026/</guid>
  <pubDate>Thu, 01 Oct 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/BvGdnRtF/Matt-IS26.jpg" medium="image" type="image/jpeg"/>
</item>
<item>
  <title>What Makes Synthetic Speech Sound Sarcastic? (Interspeech 2026)</title>
  <link>https://mattcoler.github.io/posts/2026-09-30-what-makes-synthetic-speech-sound-sarcastic-interspeech-2026/</link>
  <description><![CDATA[ 





<p>Today at Interspeech 2026 in Sydney, we presented <em>What Makes Synthetic Speech Sound Sarcastic? A Prosody-Controlled Perception Study</em>, with Zhu Li and Shekhar Nayak.</p>
<p>Prosody is a big part of how we hear sarcasm. But in natural speech the cues move together: a speaker slows down, gets louder and shifts their pitch all at once, so it’s hard to say which one listeners are actually picking up on. We got around that by using neural TTS to vary speech rate, pitch variation and loudness independently, then asked listeners how sarcastic and how natural each clip sounded.</p>
<p>For people, loudness did most of the work. We ran the same clips through an audio-capable foundation model, and it leaned on speech rate instead. The model reaches similar judgments by a different route than humans do, which is worth knowing before anyone trusts these systems to read tone of voice.</p>
<p>The paper is on <a href="https://arxiv.org/abs/2606.09717">arXiv</a>. It’s part of our wider work on <a href="../../research.html">sarcasm and multimodal voice AI</a>.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/3JpcJXD3/Screenshot-2026-09-30-at-16-48-52.png" class="img-fluid figure-img"></p>
<figcaption>Dr Shekhar Nayak presenting What Makes Synthetic Speech Sound Sarcastic? at Interspeech 2026 in Sydney</figcaption>
</figure>
</div>



 ]]></description>
  <category>Multimodal AI</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-09-30-what-makes-synthetic-speech-sound-sarcastic-interspeech-2026/</guid>
  <pubDate>Wed, 30 Sep 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/3JpcJXD3/Screenshot-2026-09-30-at-16-48-52.png" medium="image" type="image/png"/>
</item>
<item>
  <title>The Governance Gap in AI for Minoritized Languages</title>
  <link>https://mattcoler.github.io/posts/2026-09-14-plurilingmedia-ai-committee/</link>
  <description><![CDATA[ 





<p>On 14 September 2026, I gave a guest lecture, “The Governance Gap in AI for Minoritized Languages,” opening a working meeting of the AI Ad Hoc Committee of <a href="https://plurilingmedia.eu/">PLURILINGMEDIA</a>, a COST Action on media in Europe’s regional and minority languages.</p>
<p>When an AI system performs badly in a minoritized language, hardly anyone can do anything about it: not the individual speaker, not the language community, not the state. Two EU instruments come up short. The AI Act is binding and enforceable but says nothing about language. The European Charter for Regional or Minority Languages names language as a protected interest but has no enforcement behind it.</p>
<p>I wrapped up with three recommendations: one turns on a regulator’s interpretive choice, one on the AI Act review already scheduled for 2028, and one on a registration mechanism that has never been applied to language.</p>
<p>The committee moved straight from the lecture into planning its October meeting, so the aim was to leave them with something to draft.</p>
<p>This talk stems from another one I gave in June in Warsaw, entitled <a href="../../posts/2026-06-09-marginalized-knowledge-systems-warsaw/index.html">Who Governs the Spoken Word?</a></p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/prMdd280/Screenshot-2026-09-14-at-09-29-31.png" class="img-fluid figure-img"></p>
<figcaption>Guest lecture to the PLURILINGMEDIA AI Ad Hoc Committee</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-09-14-plurilingmedia-ai-committee/</guid>
  <pubDate>Mon, 14 Sep 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/prMdd280/Screenshot-2026-09-14-at-09-29-31.png" medium="image" type="image/png"/>
</item>
<item>
  <title>From Play to Practice: Open Education Award</title>
  <link>https://mattcoler.github.io/posts/2026-06-26-from-play-to-practice-open-education-award/</link>
  <description><![CDATA[ 





<p>My project “From Play to Practice” has been awarded the <a href="https://www.rug.nl/about-ug/organization/service-departments/teaching-academy-groningen/grants-support/open-education-incubator-fund?lang=en">Open Education Incubator Fund by the Teaching Academy Groningen</a>.</p>
<p>The premise: you can learn serious research skills by playing games. My MSc Speech Technology Research Design course already runs on game-based activities — a board game about research values, a LEGO reproducibility challenge, card-based ethics dilemmas — and the award lets me turn them into standalone, openly licensed resources any educator can pick up and run.</p>
<p>Students help build the materials, proposing new dilemmas and scenarios that carry forward to future cohorts, and everything ships under CC BY-SA, free to reuse and adapt. Thanks to the Open Education Community of Practice and my co-applicant Angelos Konstantinidis.</p>
<p>Read the Open Education vision <a href="https://drive.google.com/file/d/1XWWVPmlfLpUD-k20l96815vvwcbtQcdz/view">here</a></p>



 ]]></description>
  <category>Funding</category>
  <category>Teaching</category>
  <guid>https://mattcoler.github.io/posts/2026-06-26-from-play-to-practice-open-education-award/</guid>
  <pubDate>Fri, 26 Jun 2026 00:00:00 GMT</pubDate>
</item>
<item>
  <title>Ethical Implications of AI in Research Assessment</title>
  <link>https://mattcoler.github.io/posts/2026-06-12-coara-spotlight-ai-research-assessment/</link>
  <description><![CDATA[ 





<p>On 12 June 2026, I gave a talk, <strong>“Ethical Implications of AI in Research Assessment,”</strong> as part of the <a href="https://www.coara.org/news/project-news/coara-spotlight-series-spring-2026/">CoARA Spotlight</a> series, speaking from my work with the <a href="https://www.coara.org/working-groups/wg-erip/">Working Group Ethics and Research Integrity Policy in Responsible Research Assessment for Data and Artificial Intelligence (WG ERIP)</a>.</p>
<p>The talk built on the <a href="https://zenodo.org/records/17433296">ERIP essay</a>, co-authored with Susanne van den Hooff, and the <a href="https://zenodo.org/records/17435545">October 2025 ERIP whitepaper on AI-driven science futures</a>. From that starting point, I covered three things that look different from a 2026 vantage point: the new empirical evidence on LLM-assisted peer review coming out of AAAI and NeurIPS, what the EU AI Act’s 2 August 2026 high-risk deadline means for institutions, and the infrastructure CoARA-ERIP is shipping in response.</p>
<p>A Q&amp;A followed, turning to what the CoARA community should do about all of this in the year ahead.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/2SyzX0t2/Screenshot-2026-08-13-at-15-57-20.png" class="img-fluid figure-img"></p>
<figcaption>spotlight</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Research integrity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-06-12-coara-spotlight-ai-research-assessment/</guid>
  <pubDate>Fri, 12 Jun 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/2SyzX0t2/Screenshot-2026-08-13-at-15-57-20.png" medium="image" type="image/png"/>
</item>
<item>
  <title>AI Understands You? Yeah, Right — Featured in IBM Think</title>
  <link>https://mattcoler.github.io/posts/2026-06-11-ai-understands-you-yeah-right-ibm-think/</link>
  <description><![CDATA[ 





<p>Our sarcasm research was featured in IBM Think, in a piece on why AI still struggles to catch sarcasm — and why that’s a problem for AI in customer-facing roles.</p>
<p>The piece draws on our work with Xiyuan Gao, now a newly minted PhD graduate, and Shekhar Nayak: our survey of the field, and the augmented multimodal datasets we’ve built for sarcasm detection.</p>
<p>Read it here: <a href="https://www.ibm.com/think/news/ai-understands-you-yeah-right">AI understands you? Yeah, right</a></p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/8z0d9PpJ/Screenshot-2026-08-13-at-17-05-38.png" class="img-fluid figure-img"></p>
<figcaption>image</figcaption>
</figure>
</div>



 ]]></description>
  <category>Media</category>
  <category>Multimodal AI</category>
  <guid>https://mattcoler.github.io/posts/2026-06-11-ai-understands-you-yeah-right-ibm-think/</guid>
  <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/8z0d9PpJ/Screenshot-2026-08-13-at-17-05-38.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Who Governs the Spoken Word? (U Warsaw)</title>
  <link>https://mattcoler.github.io/posts/2026-06-09-marginalized-knowledge-systems-warsaw/</link>
  <description><![CDATA[ 





<p>On 9 June 2026, I gave a talk, “Who Governs the Spoken Word? Speech Tech, Minoritized Languages, and the Limits of AI Law,” at the <a href="https://filozofia.uw.edu.pl/2026/06/ai-marginalized-knowledge-systems-challenges-for-research-and-society/">AI &amp; Marginalized Knowledge Systems: Challenges for Research and Society workshop</a> at the University of Warsaw.</p>
<p>The workshop was organized as part of the seminar series of the Center for Research and Practice in Cultural Continuity, where I serve as a board member. It brought together researchers working on the relationship between AI and marginalized knowledge systems: minority languages, cultural heritage, and the ethical and social questions raised by new technologies.</p>
<p>My talk examined the governance gaps that open up when speech technology extends into minoritized languages: who has standing to build, deploy, and regulate these systems, and where existing frameworks for language rights and AI law fail to anticipate the vulnerabilities of small speech communities.</p>
<p><img src="https://i.postimg.cc/wvB9K7th/wawa1.png" class="img-fluid" alt="plan"> <img src="https://i.postimg.cc/cHJdqrKR/Screenshot-2026-08-13-at-15-08-54.png" class="img-fluid" alt="slide"></p>



 ]]></description>
  <category>AI ethics</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-06-09-marginalized-knowledge-systems-warsaw/</guid>
  <pubDate>Tue, 09 Jun 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/wvB9K7th/wawa1.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Voice AI Knows What You Say But Not What You Mean</title>
  <link>https://mattcoler.github.io/posts/2026-05-26-voice-ai-amsterdam-meetup/</link>
  <description><![CDATA[ 





<p>Voice AI Space is a global community of builders, founders, and researchers exploring how machines listen, speak, and interact with people; Amsterdam followed earlier meetups in Barcelona, London, Dubai, Paris, and New York. The evening mixed networking with short talks and live demos from people building across the voice AI stack.</p>
<p>My talk addressed one of voice AI’s harder open problems: getting systems to understand not just what people say, but what they mean. Sarcasm is a useful test case for this kind of work, since resolving it means weighing tone, timing, and context against the words themselves as opposed to the words “alone”; our research approaches this by combining speech, text, and vision, since sarcasm often shows up in one channel and not the others.</p>
<p>It’s a different kind of room than an academic conference: builders and researchers comparing notes, in real time, on what actually holds up when these systems meet real users.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/x8774dHc/ai.png" class="img-fluid figure-img"></p>
<figcaption>symbol</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2026-05-26-voice-ai-amsterdam-meetup/</guid>
  <pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/x8774dHc/ai.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Board Games Against the Machine</title>
  <link>https://mattcoler.github.io/posts/2026-02-13-unexpected-values-research-design/</link>
  <description><![CDATA[ 





<p>In February 2026 I ran a session of <a href="https://www.rug.nl/jantina-tammes-school/education/game-unexpected-values/">UnExpected Values: From Dystopia to Utopia</a> in Research Design, the thesis preparation course for the MSc Speech Technology. The Jantina Tammes School has since written it up as a <a href="https://www.rug.nl/jantina-tammes-school/education/game-unexpected-values/case-studies/making-ethical-tensions-tangible-in-speech-technology-education">case study</a>.</p>
<p>The reason for the change was partly pedagogical and partly defensive. Ethics in a thesis course tends to arrive as a written reflection, and a written reflection on ethical trade-offs is precisely the kind of text a language model produces fluently, at length, and without anyone having thought anything. You get competent paragraphs about fairness and consent that could have been written about any project at all. What a model cannot do for you is have an ethical debate in the classroom.</p>
<p>So that is what we did. We used the <a href="https://www.rug.nl/jantina-tammes-school/education/game-unexpected-values/dilemmas-speech-tech/">speech technology dilemma cards</a>, which is what makes this version of the game land with these students: the dilemmas are theirs. The Synthetic Shortcut, Revenge of the Clones, Accent Bias Fix, The Invisible Recorder — not distant hypotheticals for people about to write a thesis on synthesis or ASR. Each student took a card, read it out, staked a position, and defended it. QR codes on the cards link to the real cases behind them, so nobody could retreat into the abstract.</p>
<p>The interesting part was watching confident positions soften once someone pushed back. Ethical compromise in research rarely arrives as a temptation to do something wrong. It arrives as a deadline, a dataset that is available rather than appropriate, a metric that is easy to report. You see that faster when a classmate names it than when you write it down alone.</p>
<p>The game is free to <a href="https://drive.google.com/drive/folders/1s8lGKkdVUaD_pSZ8eFjfpIFH9-C9w-LS">download, print, and play</a> if you want to run something similar.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/MHRFXQLd/Screenshot-2026-09-08-at-15-24-53.png" class="img-fluid figure-img"></p>
<figcaption>poster</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Teaching</category>
  <guid>https://mattcoler.github.io/posts/2026-02-13-unexpected-values-research-design/</guid>
  <pubDate>Fri, 13 Feb 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/MHRFXQLd/Screenshot-2026-09-08-at-15-24-53.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Sensory Experiences featured in ISA TG07 Newsletter</title>
  <link>https://mattcoler.github.io/posts/2026-02-01-sensory-experiences-isa-tg07-review/</link>
  <description><![CDATA[ 





<p><em><a href="https://benjamins.com/catalog/celcr.24">Sensory Experiences: Exploring Meaning and the Senses</a></em> — the volume I co-authored with Danièle Dubois, Caroline Cance, Arthur Paté, and Catherine Guastavino — was featured in Issue #30 of the <a href="https://www.isa-sociology.org/uploads/imgen/2440-tg07newsletter-february-2026.pdf">ISA Research/Thematic Group 07 “Senses and Society” Newsletter</a>, themed this issue around senses and cognition.</p>
<p>The write-up situates the book within a broader discussion of how sensing and cognition are entangled in the production of social reality, alongside work by Karen Cerulo, Tia DeNora, and a review of <em>Senses, Cognition, and Ritual Experience in the Roman World</em>.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/L6jndbPV/IMG-7703.jpg" class="img-fluid figure-img"></p>
<figcaption>Review of Sensory Experiences in ISA TG07 Newsletter, Issue #30, February 2026</figcaption>
</figure>
</div>



 ]]></description>
  <category>Media</category>
  <category>Publication</category>
  <guid>https://mattcoler.github.io/posts/2026-02-01-sensory-experiences-isa-tg07-review/</guid>
  <pubDate>Sun, 01 Feb 2026 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/L6jndbPV/IMG-7703.jpg" medium="image" type="image/jpeg"/>
</item>
<item>
  <title>Speech Technology as Interdisciplinary Science — Centre Q Lecture at the University of Warsaw</title>
  <link>https://mattcoler.github.io/posts/2025-12-16-speech-technology-as-interdisciplinary-science-centre-q-university-of-warsaw/</link>
  <description><![CDATA[ 





<p>On 16 December 2025, I gave a lecture entitled <strong>“Speech Technology as Interdisciplinary Science”</strong> at Centre Q — the Centre for Research on Culture, Language and Mind at the University of Warsaw.</p>
<p>Artificial intelligence has revolutionized speech synthesis, yet these breakthroughs remain largely concentrated on a small number of major world languages. In this lecture, I explored the history of speech technology through an interactive journey from Wolfgang von Kempelen’s 18th-century speaking machine to contemporary neural architectures, examining how each era has reimagined what speech fundamentally is: physical resonance, encoded information, or learnable patterns.</p>
<p>Through examples from current research on low-resource languages, I discussed how modern speech technology can extend beyond major languages and support linguistic diversity. Live demonstrations showed how languages with limited digital resources can achieve increasingly natural speech synthesis with innovative machine learning approaches.</p>
<p>The lecture also explored deeper questions about the relationship between human communication and technology:</p>
<blockquote class="blockquote">
<p>What transforms when voice moves from embodied cultural practice to decontextualized acoustic reproduction? When lived stimulation becomes technological stimulus?</p>
</blockquote>
<p>These questions highlight why speech technology increasingly requires contributions from the social sciences and humanities—not as auxiliary perspectives, but as essential ways of understanding what technological progress and breakthrough actually mean.</p>
<p>More information about this event: <a href="https://centrumq.uw.edu.pl/en/2025/12/lecture-by-matt-coler/">Centre for Research on Culture, Language and Mind, University of Warsaw</a></p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/286Xp0sN/Screenshot-2026-08-07-at-23-11-20.png" class="img-fluid figure-img"></p>
<figcaption>Lecture on Speech Technology as Interdisciplinary Science at the University of Warsaw</figcaption>
</figure>
</div>



 ]]></description>
  <category>AI ethics</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2025-12-16-speech-technology-as-interdisciplinary-science-centre-q-university-of-warsaw/</guid>
  <pubDate>Tue, 16 Dec 2025 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/286Xp0sN/Screenshot-2026-08-07-at-23-11-20.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Billion-Dollar Voices and Minority Languages — Keynote at the University of Warsaw Winter Innovation School</title>
  <link>https://mattcoler.github.io/posts/2025-12-15-billion-dollar-voices-and-minority-languages-keynote-at-the-university-of-warsaw-winter-innovation-school/</link>
  <description><![CDATA[ 





<p>On 15 December 2025, I gave the keynote <strong>“Billion-Dollar Voices and Minority Languages: The Promise and Challenge of Speech Technology”</strong> at the University of Warsaw Winter Innovation School, <em>At the Heart of Artificial Intelligence Research</em>.</p>
<p>The Winter Innovation School brought together doctoral researchers to explore emerging approaches to innovation and artificial intelligence. My keynote focused on a central challenge in the development of modern speech technology: while recent advances have enabled remarkable progress for major world languages, thousands of languages remain largely absent from the digital ecosystem.</p>
<p>Companies developing cutting-edge voice AI systems have demonstrated that highly natural synthetic voices and conversational AI are now possible at an unprecedented scale. The challenge is ensuring that these innovations do not widen existing digital divides, but instead contribute to linguistic diversity and equal access to technology.</p>
<p>In this talk, I explored how advances in machine learning, combined with linguistic knowledge and community-driven approaches, can enable speech technology for languages with limited digital resources. Using examples from low-resource speech synthesis and recognition, I discussed how interdisciplinary research can transform technical challenges into opportunities for innovation and cultural preservation.</p>
<p>The future of voice AI will not only be measured by how well it performs for the world’s largest languages, but also by whether it can serve the many communities whose languages have historically been overlooked.</p>
<p><img src="https://i.postimg.cc/655f2TcL/Screenshot-2026-08-07-at-22-53-26.png" class="img-fluid" alt="Keynote at the University of Warsaw Winter Innovation School"> <img src="https://i.postimg.cc/4x0PTZzr/Screenshot-2026-08-07-at-22-51-58.png" class="img-fluid" alt="photo2"> <img src="https://i.postimg.cc/YSm3rV3s/Screenshot-2026-08-07-at-22-51-01.png" class="img-fluid" alt="photo3"> <img src="https://i.postimg.cc/vZKhdMWJ/Screenshot-2026-08-07-at-22-51-11.png" class="img-fluid" alt="photo4"></p>



 ]]></description>
  <category>Keynote</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2025-12-15-billion-dollar-voices-and-minority-languages-keynote-at-the-university-of-warsaw-winter-innovation-school/</guid>
  <pubDate>Mon, 15 Dec 2025 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/655f2TcL/Screenshot-2026-08-07-at-22-53-26.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Speech Technology and Minority Language Preservation — Brussels</title>
  <link>https://mattcoler.github.io/posts/2025-12-05-speech-technology-and-minority-language-preservation-brussels/</link>
  <description><![CDATA[ 





<p>On 5 December 2025, I gave a talk on <strong>“Speech Technology and Minority Language Preservation”</strong> as part of the event <em>Regional and Minority Languages: Game Changer in the Digital World?</em> in Brussels, Belgium.</p>
<p>The event brought together researchers, policymakers, and stakeholders from across Europe to discuss how regional and minority languages can shape the future of digital technologies. My contribution focused on the role of speech technology in supporting linguistic diversity and ensuring that advances in AI benefit not only major world languages, but also the many languages that remain underrepresented in digital systems.</p>
<p>Developing speech technologies for minority languages presents technical and societal challenges, from limited training data to appropriate evaluation methods and sustainable language resources. At the same time, these technologies offer new opportunities for language preservation, revitalisation, and participation in the digital world.</p>
<p>The discussions highlighted the importance of collaboration between technology developers, language communities, researchers, and policymakers to build a more inclusive digital future.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/mr2sKYy8/2512.png" class="img-fluid figure-img"></p>
<figcaption>Speaking about speech technology and minority language preservation in Brussels</figcaption>
</figure>
</div>



 ]]></description>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2025-12-05-speech-technology-and-minority-language-preservation-brussels/</guid>
  <pubDate>Fri, 05 Dec 2025 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/mr2sKYy8/2512.png" medium="image" type="image/png"/>
</item>
<item>
  <title>Speaking Machines and Living Languages — Keynote at the University of Warsaw</title>
  <link>https://mattcoler.github.io/posts/2025-10-17-speaking-machines-and-living-languages-keynote-at-the-university-of-warsaw/</link>
  <description><![CDATA[ 





<p>On 17 October 2025, I gave the keynote “Speaking Machines and Living Languages” at <a href="https://inicjatywadoskonalosci.uw.edu.pl/en/breakthroughs-2025/">Breakthroughs: Cutting-edge research in the humanities and social sciences</a> at the University of Warsaw.</p>
<p>The conference brought together researchers working across the humanities and social sciences as part of the university’s Excellence Initiative – Research University programme. My keynote explored the relationship between speech technology, linguistic diversity, and the living languages that these technologies increasingly mediate.</p>
<p>It was a pleasure to contribute to a programme showcasing the breadth of interdisciplinary research taking place at the University of Warsaw.</p>
<div class="quarto-figure quarto-figure-center">
<figure class="figure">
<p><img src="https://i.postimg.cc/4NKWdmBz/251017.png" class="img-fluid figure-img"></p>
<figcaption>Speaking Machines and Living Languages keynote at University of Warsaw</figcaption>
</figure>
</div>



 ]]></description>
  <category>Keynote</category>
  <category>Language diversity</category>
  <category>Talk</category>
  <guid>https://mattcoler.github.io/posts/2025-10-17-speaking-machines-and-living-languages-keynote-at-the-university-of-warsaw/</guid>
  <pubDate>Fri, 17 Oct 2025 00:00:00 GMT</pubDate>
  <media:content url="https://i.postimg.cc/4NKWdmBz/251017.png" medium="image" type="image/png"/>
</item>
</channel>
</rss>
