from corpus research to finished content - at scale
AI-POWERED CONTENT DEVELOPMENT - AT SCALE
Publishers who want to offer extensive reading face two hard problems: producing enough titles to make a program worth adopting, and leveling them accurately enough that the catalog can be trusted and combined with others.
Both matter because of what makes extensive reading work. The research-established golden rules of graded reading start with two: learners must read at a level that's genuinely easy for them, and they must read things they genuinely want to read. Most publishers can manage the first by working slowly. The second requires hundreds — ideally thousands — of stories across enough topics that every learner finds something that appeals to them.
What makes the pipeline work is that I built every layer of it myself:
The word lists. Ten corpus-based lists, including one developed specifically for graded readers.
The profiling tools. I've built several; the NGSL Profiler is now used worldwide to verify texts against the 95–98% coverage learners need.
The AI tools. Designed to generate content from real word lists rather than an LLM's guesswork about difficulty.
The editorial judgment. Thirty years of deciding what actually makes a story worth reading.
There's no part of this process I'm outsourcing or licensing from someone else.
The result: a publisher can go from a standing start to a large, properly leveled, genuinely engaging catalog in months rather than years. In the past year alone, my pipelines have produced more than 300 graded readers for major publishers.
CORPUS ANALYSIS & CUSTOM VOCABULARY LISTS
General word lists only take a product so far. Most companies eventually need to know which words matter for their learners, their domain, and their content — and that question can only be answered with a corpus built for the purpose.
That's the work I've done most often, and it's where I've delivered the most value. I build custom corpora and derive vocabulary lists tailored to a specific audience and product: gym-specific and young-learner lists for UFC Gym Japan, TOEIC vocabulary for Cooori's learning system, business English for NHK, and academic corpus analysis in 2026 for the University of Tokyo Graduate School of Science and Singapore International School.
I also analyze corpora clients already have, and help teams apply existing lists — including the NGSL family — to their own platforms, starting with a needs analysis of the learners and the learning environment rather than an assumption about what they should study.
Every project comes back with more than a list: word importance rankings, cumulative coverage figures showing exactly how many words are needed to reach the 95–98% thresholds that make reading possible, and the tables and analysis to explain what it all means. The methodology is rigorous, but it's always grounded in what actually happens in classrooms.
TESTING & ASSESSMENT
Vocabulary and reading programs need reliable ways to place learners, measure progress, and prove results — and the right instrument looks different every time.
I've built assessments across that whole range: the IRT-based computer-adaptive "V-Check" vocabulary test developed for Lexxica, on which I co-hold the patent; achievement tests for the four-level In Focus series published by Kinseido; online assessment tools for GoFluent and EnglishCentral; and the free NGSL computer-adaptive placement test, still available on the NGSL Project site.
On assessment projects I work closely with Brent Culligan, my longtime collaborator and co-creator of the NGSL, whose psychometric expertise is among the best in the field. Whether you need a quick placement tool, a criterion-referenced achievement test, or a full norm-referenced proficiency instrument, we can build it.
CURRICULUM & LEARNING CONTENT DEVELOPMENT
Some clients don't need a component — they need someone to take a program from concept to delivery. That's work I've done across just about every context English gets taught in.
I managed the content development for NTT Learning Systems' "Virtual ALT," a tablet-based learning system sold to public and private junior high schools throughout Japan: I set the pedagogical goals, hired and directed the author and editorial team, and delivered the full program on time and on budget. For UFC Gym Japan I built a fitness-and-English curriculum for both adults and young learners, along with the teacher training to run it. For NHK I developed content for the "Kiso Eigo" radio program and the corpus-based business English list used on the TV show "Eigo Shaberenaito." I've also authored or co-authored textbook series for Kinseido, Cambridge University Press, Nan'un-do, and Pearson, and advised platform teams at Telelangue, EdulinX, Slang, and CognitionPoint on pedagogy and content strategy.
Junior high classrooms, corporate learners, gyms, national broadcast, online platforms — the audiences differ, but the questions are the same: what should learners actually be able to do, what content gets them there, and how do you build it at the quality and pace the project needs?
TEACHER TRAINING
I created and continue to run the EFL teacher training program at Meiji Gakuin University, and teach at the graduate level at several major Japanese universities. For GoFluent I redesigned the teacher education system supporting more than 500 teachers, and I've delivered training for UFC Gym Japan, local boards of education across Japan, and Ministries of Education around the world as a U.S. State Department English Language Specialist.
If teacher development is part of a larger project, it's built in rather than bolted on.
KEYNOTES & WORKSHOPS
I've given more than 100 plenary and keynote addresses at conferences around the world, along with two TEDx talks — one introducing the New General Service List Project, and a 2024 talk on Zipf's Law and the project's expanded scope and impact. A full list of past talks is here.
I speak on vocabulary acquisition, extensive reading, corpus linguistics, and — increasingly — what AI can and can't do well in language learning content development. Talks are built for the audience in front of them, whether that's a research conference, a publisher's editorial team, or a room full of classroom teachers.
If you're organizing a conference or an in-house session, I'd be glad to hear what you have in mind.