Verbalist

a verb-building robot — by John and Muriel Higgins, 1992–2003, recreated as a website, 2026

Launch Verbalist

VERBALIST is a verb-building robot. So what, you may ask — as English speakers we already know how to build verbs, so why get a machine to do it for us? The answer is that if you succeed in teaching a machine how to do something, you come to understand better how you do it yourself. VERBALIST tries to capture, in a program, everything an English speaker knows about forming verbs and verb phrases. Play with it. See if you can catch it out. In the process you'll bring to the surface all the subconscious knowledge you already have about how this part of the language works. If you are an EFL learner, check out some of the more complicated forms by hunting for them on Google or using a concordance.

Pick a verb, then flip through tense, modal, perfect, continuous, "going to", passive, negative, question and tag-question — each combination rebuilds the phrase live, and whenever a spelling or grammar rule applies, a note explains what just happened and why.

I walk. → He hasn't been walking. → Will they have been walked? → She isn't walking, is she?

What's in this version

Where it came from

VERBALIST began with a student who wanted a program that would test learners on English verb tenses and mark them right or wrong. John Higgins argued that the machine couldn't be trusted to grade reliably — we don't understand the grammar of tense and aspect well enough to write foolproof rules for it. So he proposed the opposite: a program that would generate and correctly inflect every possible verb phrase for any headword you gave it, and leave it to the learner to work out which ones sounded right and search for examples in a corpus of text.

It belongs to a small family of "exploratory" programs — a term coined by Higgins and Tim Johns in 1984 — that includes S-ENDING and A/AN, written by Johns for an unexpanded 1K Sinclair ZX81 (A/AN could tell a uniformed man from an uninformed man using nothing but spelling rules, no dictionary), and LOAN, SORRY and THANKS, which generated polite requests, apologies and thanks from a handful of inputs.

It's often said that the best way to learn a subject is to teach it — and that's what you're doing here: teaching the machine by making it create phrases and judging whether they're right. It's one working answer to a question Arthur Luehrmann asked back in 1972: should the computer teach the student, or vice versa?

Launch Verbalist