Language Log

This post documents one small step in a larger plan for improved evaluation of prosody in reading. It compares word-level timing in a large number of recordings, from the Speech Accent Archive at GMU, of 3038 people reading the 69-word "Please call Stella" passage. 661 of these people are native speakers of English, with accents from all over the anglophone world, while the remaining 2377 readers have native languages from Afrikaans to Zulu. The reading and speaking level of those non-native readers varies a lot, and many of them have problems in decoding or pronunciation that affect their timing.

Automated analysis of such problems should be useful in foreign-language teaching. And similar analyses might help in early reading instruction for students in anglophone classrooms, whatever their native language.

Let's start with a quick comparison of word-level timing in the 661 native English speakers; the 85 native French speakers; the 99 native Korean speakers; and the 82 native russian speakers.

I calculated word-level time points for those 927 speakers, using a forced-alignment system originally developed many years ago with Jiahong Yuan — a summary of the technology and a few of its application can be found here (open-access version). Here's the output for speaker english1 — note that the segment ID sp means "silent pause".

The key conclusions:

1. Time between word onsets gives a good picture of phrase structure, despite the many other effects on timing;
2. Individual non-native readers, aside from being overall a bit slower, usually show lengthened inter-word intervals in unexpected places, due to decoding or pronunciation problems.

"Why Modern Chinese is Just ‘English with Hanzi’,Hanzi Shells, English Souls: The Europeanization of the Chinese Language", by Jingyu, Old North Whale Review (2/09/26)

Learning Chinese is widely sold as the ultimate linguistic challenge. Students are warned that they must rewire their cognitive faculties entirely to grasp an alien logic. But there is a reality that few textbooks admit: The Chinese language has been Europeanized.

Beneath the intimidating surface of the Chinese Characters (汉字, Hanzi), the operating system has been quietly swapped out. If one strips away the characters and the tones, what remains is not the mysterious, ancient syntax of the Tang Dynasty poets. It is a structure that is shockingly familiar.

That's the name of a very fine health care facility nestled in the wooded hills of Philadelphia's northwestern suburbs — Malvern, Tradyffrin, Bryn Mawr ("large hill"), Bala Cynwyd (named for towns in Wales), Haverford, Narberth, Radnor, Berwyn, Merion, and Gwynedd.

My inclination is to abbreviate the name somehow — BMRH, Bryn Mawr RH, etc. — but the people who work at Bryn Mawr Rehab Hospital tend not to do that.  They want to keep the word "rehab" in their habitual reference.

On the other hand, I think "rehab" is too casual and informal for an institution of such complexity and excellence.  By nature, "rehabilitation" is hexasyllabically cumbersome and "hospital" is trisyllabically unglamorous.

Never mind what Bryn Mawr Rehabilitation Hospital is called on a day-to-day basis, it's a thoroughly admirable place.

"'The purist jungle?'" (3/27/2026) featured Anne Abeillé's book "La Grammaire se Rebelle". As background, I should have cited "Le français va très bien, merci", Tract des linguistes 5/23/2023 (= "French is doing very well, thank you").

Apparently our haplogroup (family of Joseph Charles Mair [Pfaffenhofen, Austria] and Esther Frieda Louise Boyce Mair [Zweismmen, Switzerland]) is U8a1a.

A student who's been working on LLM-style AI transformations of symbolically-represented music recently tried mapping a (fragment of a piece) back and forth between two genres, e.g. baroque and pop. She found that after a couple of steps, the results reach a fixed point and don't change any more.

Following up on Tuesday's DP link, today we have Wyatt G. Croog, "Harvard Launches 'Strategic Authenticity Initiative' to Help Students Seems Normal", The Harvard Crimson 4/1/2026:

In an effort to address growing concerns that its students are “deeply unsettling in conversation,” Harvard University announced Monday the launch of the Strategic Authenticity Initiative, a university-wide program designed to help students convincingly simulate being regular people.

The initiative, funded by three grants and one concerned parent, will offer workshops on:

1. Making eye contact without turning it into a networking opportunity

2. Having hobbies that are not a startup

3. Telling a story that does not end up in a LinkedIn article

Penn's student newspaper, The Daily Pennsylvanian, traditionally publishes an April Fool's issue every year, generally a week or so before April 1. This year's version has not (so far) been put out as a paper version, or even in the standard online form, but only as a set of images.

This may sound hopelessly old-fashioned.  People were making the accusation more than half a century ago, but the same problems it points to persist even today.

In computer science, garbage in, garbage out (GIGO) is the concept that flawed, biased or poor quality ("garbage") information or input produces a result or output of similar ("garbage") quality. The saying points to the need to improve data quality in, for example, programming. Rubbish in, rubbish out (RIRO) is an alternate wording

The principle applies to all logical argumentation: soundness implies validity, but validity does not imply soundness. In essence, the logic or algorithm may be correct, but using flawed inputs (premises) is still an informal fallacy.

(WP)

The dangers of GIGO / RIRO have only been magnified with the advent of AI.

John Burn-Murdoch, "Social media is populist and polarising; AI may be the opposite", Financial Times 3/28/2026:

Every media revolution has transformed who distributes information, what messages are distributed and what form they take. As such, some media are fundamentally democratising and polarising, widening the pool of publishers and views beyond a narrow elite and amplifying radical and anti-establishment voices. TikTok and the printing press arrived almost 600 years apart but share these characteristics. Others push the opposite way: radio and television had high barriers to entry, creating a monopoly for the voices and views of elites and experts.

As the use of AI chatbots takes off, it’s worth pausing to ask which of these categories they fall into. There is good reason to believe it is the latter.

Anne Abeillé's recently-published book "La Grammaire se Rebelle" describes linguistic prescriptivism as "la jungle puriste" / "the purist jungle".

But wait, don't prescriptivists want to turn the natural linguistic wilderness into a well-tended formal garden? Maybe, but in fact prescriptive rules are often incoherent as well as contrary to elite as well as informal usage, as we've often observed.

There's more to say about the many metaphors for linguistic prescriptivism — for example, the parallels with socio-political authority and rebellion —  but for now, here's the avant-propos of Abeillé's book, followed by Google Translate's (very good) English version:

…and other applications of non-linear dynamics. A press release from Northwestern University — "Bell-bottoms today, miniskirts tomorrow: Math reveals fashion's 20-year cycle":

Fashion insiders and beauty magazines have long cited the "20-year-rule"—the idea that clothing trends often resurface every two decades. According to Northwestern University scientists, that observation isn't just anecdotal. It's a mathematical reality.

In a new study, the Northwestern team developed a new mathematical model showing that fashion trends tend to cycle roughly every 20 years. By analyzing roughly 37,000 images of women's clothing spanning from 1869 to today, the team found that styles rise in popularity, fall out of favor and then eventually experience renewal. Along with supporting common perceptions about the life cycles of fads, the researchers say these results could help explain how new ideas spread in society.

The study's lead author, Emma Zajdela, will present these findings on Tuesday, March 17, at the American Physical Society (APS) Global Physics Summit in Denver. Her talk, "Back in Fashion: Modeling the Cyclical Dynamics of Trends," is part of the session "Statistical Physics of Networks and Complex Society Systems."