Spent almost all day standing at my new standing desk and just submitted a journal article. I might have 4 straight hours of student meetings tomorrow, but today I'm feeling good.
Assistant Professor @ University of Illinois Urbana-Champaign | computational phonology, phonetics-phonology interface, representations, logic/model theory, philosophy of science. Fan of sound in general (synthesizers, field recordings, natural reverb,…)
164 followers 148 following 123 posts
view profile on Bluesky
Spent almost all day standing at my new standing desk and just submitted a journal article. I might have 4 straight hours of student meetings tomorrow, but today I'm feeling good.
My cousin’s child (1;10) was introduced to my niece whose name is Leighton. My cousin said this is Lei[ʔn̩] and seconds later her child repeated it back as Lei[tɨn] with a full [t]. A gigantic win for abstractionists.
🤢
when you decide that yes, you're going to assign your own paper in the grad seminar
🚨Postdoctoral fellowship in corpus phonetics / data science for speech with me and Ann Bradlow. Position is open immediately. Apply now! 🚨 Details: faculty.wcas.northwestern.edu/matt-goldric...
I love reading about Gold's singular theorem.
Will spend the rest of my life considering how we were rushed into "going paperless" for the good of the environment only for the next generation of digital tools to become far worse for the environment.
This gallery of 200+ photographs of hand-painted storefront signs in the Detroit area just made my night. Reminds me of a past life driving to pick up vinyl at Archer Records pressing plant or god knows where for a random punk show. Michigan is the best. andrewandersonphoto.com/ron-signs
Even /vr/ is a valid English onset at the vegan diner
Yes, reinforcement learning is powerful, but what put all that structure in the learning data in the first place? A system with what capacities? What architecture? What evolutionary and developmental histories? That’s where AI-as-engineering ends and Cognitive Science begins!🤖🧠
I’m happy to share that my paper with Jeff Heinz on “the blueprint model of production” is finally published online in Phonology: www.cambridge.org/core/journal...
Wow! I can't wait to get home and look at this in detail!
I had a go at sketching out BMP in python jofrhwld.github.io/blog/posts/2...
We really need to find a way to colaborate on something.
Thanks for sharing!
Important to check out if you're interested in discrete vs. gradient representations and notions like 'cascading activation' in speech production
It's not exactly the same thing as your plot, but in a WCCFL proceedings paper that hopefully will be out soon I talk about the BMP as it relates to optionality. I make a point in it that there are different predictions for the variance given the specific implementation. So def on my mind!
Thanks! Weighted sum was chosen because it's easy to conceptualize, but if a different implementation captured the quantitative aspect of the phenomena better and maintained the abstract qualitative properties then I would be happy to abandon weighted sum.
From p. 12: “…phonetic evidence alone does not bear on the nature of phonological knowledge, but rather must be evaluated in tandem with a theory of how phonological knowledge is physically manifested.” Basically, the interface itself is a variable that we can and should manipulate and study.
These two phenomena were chosen specifically because Port and Leary state that IN is evidence against discrete computation (grammar) and Gahl says that ViHD is evidence against discrete representations. These claims are only true if you assume the “modular feedforward model” is a monolith.
We use type theory to talk about the phonetics-phonology interface and show how characterizing it as a higher-order function (rather than a composed function) provides a way for discrete/symbolic phonology to result in phenomena like incomplete neutralization and variation in homophone duration.
I’m happy to share that my paper with Jeff Heinz on “the blueprint model of production” is finally published online in Phonology: www.cambridge.org/core/journal...
Them: here's this algorithm; this is a Marrian computational level theory Me: ...I don't think that's Them: here are some extensional properties of the function being computed; this is a Marrian algorithimic level theory Me: ...I don't think that's Them: Marr!
Solidarity, brother
the hypothesis that the human brain can perform rule-based computations is theoretically extravagant, which is why i think it just does big ass linear algebra problems instead
New Algernon Cadwallader is hitting for me. Relatedly, my long held belief is that Philly bands best captured the midwestern Kinsella sound because of sociolinguistic similarities (/æ/ raising). algernoncadwallader.bandcamp.com/track/hawk
the concepts & principles that computational theories of cognition rely on were pretty much all developed *before electronic computers existed*
"The brain is NOT a computer!!!!!!!!" well sure maybe yours isn't
The term "corn sweat" seems to be on the rise as I had never really heard it being used before this year. Went to Google Ngram Viewer to look at the trend. It does appear to be on the upswing, but also is less frequent than "sweat corn" which I assume is just a misspelling of "sweet corn".
I can’t even begin to explain what this track makes me feel. Many days I feel as if it is the best piece of music ever written. youtu.be/skf-EBUwAuc?...
Goodbye Eugene. It was nice to walk five miles a day and remain relatively comfortable, but I must now return to the corn sweat of the Midwest. Had a great time at both SCiL and DYMOS as well. Life is good
![A picture of a slide from an academic presentation with a picture of three turtles stacked on top of one another and a tile that says insert [infinite regress cliche] here](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:c2325fiujmbcgf3wdcbszelx/bafkreifygl35rmcuozoc5iyybu2pftfmd5ot7oggm23jzd5qzhvwunn4pe@jpeg "Click to view full size")
Here are my slides: www.scott-nelson.net/Presentation... I started my talk by saying, “usually when you’re a phonologist who is interested in phonetics you get more concrete but I decided to get more abstract instead.” If that’s of interest to you then you might be interested in checking these out!
Hell yeah. A classic. My friend plays drums for them now! I go back to “The Sun” a lot and really love all of ‘Charmer’. Home After Three Months Away by E!E! is maybe my favorite emo revival release and The World Is from 2010-2015 was undefeated in my eyes. Nice choices!!
A lot of Jerry Fodor and Midwest Emo posting on my timeline this week. This pleases me.
What’s your go to midwest emo
Breaking kayfabe to admit I don't really think Fodor's account of modularity is right; most likely, modules only need 7-8 of the proposed features, not all 9
Marie Huffman told me that when she was a grad student Peter Ladefoged told her that she should figure out exactly what her figures were gonna be and then write the paper around that. It has really shaped the way I think about “the writing process”.
where do they FIND these people?!
Nice. I’ve only ever read the Lakatos paper but am looking forward to working my way through the whole thing at some point!
It’s part of a larger target-comment-response collection. Kuhn is the target author and his response is titled “Reflections on my Critics” which I semantically parse as “A Response to my Haters” which makes me giggle.
Karthik Durvasula pointed me to Imre Lakatos’ “Falsification and the Methodology of Scientific Research Programs”. It really helped me pull together my thoughts on the philosophical underpinning of my dissertation. Found a hard copy at a used bookstore in Eugene, OR today. The world is kind.
Putting a helpful disclaimer on all my papers that reads: "I JUST WROTE THIS TO GET TENURE; DO NOT, UNDER ANY CIRCUMSTANCES, READ IT (THOUGH FEEL FREE TO CITE LIBERALLY)"
From Eugenia Chang's 'The Joy of Abstraction': "It might seem that abstraction takes us further away from 'real' situations. This is superficially true, but at the same time abstraction enables us to make connections between situations that are further apart from one another."
woah, crazy revelations at the coldplay concert
Thing I am an absolute complete total reactionary about: there has not actually been invented a better model of conveying information in a learning environment than the basic structure of a traditional lecture. A speaker standing in some sort of unique focal point for the attention of listeners...
👋🏻
Upcoming LSA Presidential Research Forum September 15, 2025 Tutorial on Logical Phonology (LP) Featuring Kyle Gorman, Rim Dabbous, and Charles Reiss Join us for a deep dive into LP! This forum highlights cutting-edge theoretical insights grounded in formal rigor! www.lsadc.org/ev_calendar_...
Happy World Emoji Day🌎🥳📆!!! Let's all celebrate by reading my two published papers on the morphosyntactic combinatorics of emojis that function as "words"! Glossa paper: doi.org/10.16995/glo... JLCL paper: doi.org/10.21248/jlc... And more to come!! #worldemojiday
😅
vG and Port take a lot of liberties when talking about computationalism. The most egrigious is in footnote 9 where they write, "despite their simplicity, [Turing machines] can (in principle) perform computations just as complex as any computational system." That parenthetical is just wrong!
I will also be at SCiL this upcoming weekend and am staying in Eugene until DYMOS. So if anyone is at the LSA institute from the 19-28 and wants to meet up to talk about phonology/phonetics, computation, synthesizers, or really antything let me know.
One thing I have discovered is both Fodor in his intro to Modularity of Mind and van Gelder and Port in their intro to Mind as Motion use the phrase, "the proof of the pudding is in the eating." Filing this nonsense away to use if I ever write a book on my thoughts on cognition.
Presenting some stuff from my dissertation at this cool workshop in a couple weeks on dynamical models of speech: samkirkham.github.io/dymos/ Doing a lot of reading/prep work as I am the "symbolicist" going to talk to a bunch of "dynamacists". Should be fun! (seriously)
Has anyone ever used "there was a farmer who had a dog and Bingo was his name(-o)" to explain ambiguous pronoun resolution?
Does anyone else find it perpetually annoying that what most folks would consider neural "computation" is Marr's "algorithmic" level, while his *computational* level is really just "what's the goal of this thing anyway?" AKA function.
I like this quote from Vinny Czarnecki's paper from SCiL this year because it makes explicit what I think I only made implicit in my dissertation. openpublishing.library.umass.edu/scil/article...
Annoyed at one page abstracts because I think they ultimately promote theoretical/formal homogeneity as there is not enough space to explain a non-standard approach AND show its use case.
Descartes gave us all the tools we need to survive in an era of misinformation. Simply begin with the understanding that you are real but everything else is fairly likely to be the cruel games of a demon sent to torment you specifically.
Got mad reading a paper so here's a Partha Niyogi quote: "for mathematical models [SN: proof based] the assumptions are more questionable but the conclusions are more reliable -- for computational models [SN: simulation based], the assumptions are more believable but the conclusions more suspect."
How I explain frequentism vs Bayesianism to my students
Computers used to scream every time they connected to the Internet. They knew. They tried to warn us. We did not listen.
I played Jerome’s Dream for my 10-month-old niece this weekend and she loved it (she approximated headbanging and laughed). Will have to wait a few years to explain the virtues of symbolic-based models of intelligence to her. jeromesdream.bandcamp.com/album/comple...
Looks like the SCiL proceedings have been published and I’m really excited to dig more into this paper from Yifan Hu: openpublishing.library.umass.edu/scil/article...
When you bring up AI nowadays and everyone just assumes you are gonna talk about LLM’s is spiritually the same thing as when I say I’ve been listening to a lot of screamo bands lately and people assume I mean The Used instead of pageninetynine. Is GOFAI equivalent to skramz in this case?
I just watched a video where the person was describing music equipment and said the phrase, “this is an a hundred watt amp.” Language is cool.
Contra Anderson here, I think LabPhon and Articulatory Phonology should be taken very seriously as they provide new insights into this relationship. I show in my dissertation and related papers that the architecture Anderson wants is possible while still engaging with the type of work he dismisses.
Phonology and phonetic implementation are just different types of well-structured functions. Personally, I don't think we fully understand how the two relate which makes "the interface" itself a variable that can be manipulated for research.
This isn't a well thought out idea, but I kinda think that the grammatical (phonology) vs. non-grammatical (phonetics) divide is like the perception vs. cognition debate in psychology. It's worse, though, because "non-grammatical" usually primes "non-structured" and that is clearly not right.
OCP satellite workshop on the role of representation in computational phonology? 👀👀👀👀👀👀👀👀👀👀👀👀👀
nobody: the Bantu languages:
Left: indirect communication (gynocentric) Right: direct communication (phallocentric) I understand you aren't used to it
✨We are organising (again) an online workshop on Theoretical Modeling for PhD candidates and Master students in cognitive science and psychology. Check out the website and the schedule: Registration is open! (closes July 1, spaces are limited) 🧵1/n computationalcognitivescience.github.io/workshop/
Postscript: it’s also funny that of all the places to end up, I’m in a department that is a 15 minute walk from the American Football house. Before going the academia route I was pursuing a music career and lead an “emo” band that was at times very Kinsella inspired. open.spotify.com/track/1X067E...
I officially graduated from Stony Brook yesterday so I guess this is a nice time to share that I accepted a tenure track position at Illinois and will be staying in Champaign-Urbana for the foreseeable future. Happy to be able to continue to do phonology (and phonetics, and comp/math ling, and …)
The fascist win if you lose your compassion for other people. That's part of the war. You have to keep compassion.
I am very happy that two of my phonology 1 students accidentally discovered mutual bleeding interactions when writing final papers about languages that they speak.
Agreed!!!
Thinking once again about the “Against Cop Shit” essay: jeffreymoro.com/blog/2020-02...
😅
Who would have thought that switching the entire research community from a decentralised, distributed, network of git repos to a single point-of-failure commercial paper collaboration tool might lead to problems?
youtu.be/iu21Q34OSvQ?...
Watching them open with “The Land Between Solar Systems” at Carnegie Hall in 2018 was an all time moment for me.
Múm doing a 16 show North American tour in 2025 is shocking in the best way. But why does the Chicago show have to be when I’m going to be out of town? :(
Yes, we absolutely should!
Here you go! It's an early sketch of an idea, but I think the general idea comes across. www.scott-nelson.net/Presentation...
Will share once I get them up on my website!
Excited to be headed to Montreal this weekend for NAPhC. Program looks great. I'm giving a talk called "Phonological Knowledge, Weighted Logic, and the Competence/Performance Distinction". sites.google.com/view/naphcxi...
Say no!
Remembering that I said "pentamic iambeter" when trying to say "iambic pentameter" while teaching earlier this semester. Guess I really wanted to keep that m to avoid hiatus even if it means splitting up a morpheme? Also lets me express the b in iamb?
Today I am going to sit in the sun and listen to Built to Spill.
In Nevin's monograph on vowel harmony he writes "the no-lookahed property of the Search procedure seems to require a target-initiated, procedural search, which cannot obviously be replicated in declarative models." BMRS is a declarative model that has this property. Another insight captured. Cool.
We're reading Chandlee and Jardine's (2021) Language paper on BMRS in my seminar this week. I hadn't read it carefully in quite some time and I really like the discussion on how BMRS unites insights from RBP, OT, and FLT. HIghly recommended! muse.jhu.edu/pub/24/artic...
This is a really nice paper by Tatevik Yolyan at Rutgers that uses Boolean Monadic Recursive Schemes to provide a decision precedure for whether or not a given phonological map is weakly-deterministic. link.springer.com/article/10.1...
It is a feminine trait to use the government to give you an advantage in trade via tariffs. The alpha male relishes competition, and seeks only to dominate his enemies on an equal footing, with honor. (me desperately trying to keep the price of Nintendo Switch 2 under $500)
Out this week @cogscisociety.bsky.social! @samhforbes.bsky.social & I ”urge extreme caution in the use of LLMs in classrooms lest we further normalize: pupils losing their privacy, reducing contact between learner and educator, deskilling teachers & polluting the environment” doi.org/10.1111/cogs...
Veeeeeeery nice
Paradox looms
I like this a lot bec it reminds us that oftentimes we're making progress by being wrong. That's why language abt replication results frustrates me so much. Just bec you've reproduced a result doesn't mean you've learned something. Just bec you've failed doesn't mean there's nothing to learn there.
My sister-in-law (grew up in the Boston area but doesn’t have a strong regional accent) just informed me that my 2.5 year-old-niece has started saying [fɔk] (fork) and [stæəz] (stairs) after my SIL’s mom stayed with them for a week (mom has a very strong regional accent).