The British and American Word Families Project (BAWF): A New Way to Explore English Word Families

by Clarence Green, Ph.D., University of Hong Kong

The BAWF website offers several ways to explore words, word parts, and word families

English has hundreds of thousands of words, but students don’t have to learn each one separately. Words belong to families that share meaning and structure. A student who knows act already has a foothold on action, active, activity, and reactivate. These connections become especially valuable as students meet the vocabulary of science, social studies, and other content-area texts.

Teachers and researchers have long wanted to know which word families are worth teaching and how often their members appear in what students read. To help answer those questions, I created the British and American Word Families (BAWF) Project, with assistance from Freddy Hiebert and Averil Coxhead.

BAWF is a free, searchable database of about 378,000 words from contemporary British and American English. For each word, it brings together information about its structure, its family, and how often it occurs. The database is designed for teachers, curriculum developers, and researchers.

BAWF also shows how word families are spread across English. Some families have many members that readers meet regularly. Others have only a few familiar members and many rare ones. A teacher looking at a family can see which members turn up often enough to be worth teaching and which students are unlikely to encounter. Researchers can look at these patterns across thousands of families instead of relying on a handful of examples.

How BAWF Was Built

BAWF draws on decades of vocabulary research, including Paul Nation’s BNC-COCA word-family lists, Averil Coxhead’s Academic Word List, and TextProject’s WordZones. Those resources showed educators which words students meet and how vocabulary can be organized for teaching. BAWF adds information about roots, prefixes, suffixes, and frequency, and it covers a much larger share of English.

To analyze 378,000 words, I combined major collections of British and American English with GPT-4.1., which broke each word into its parts. The GPT model was fine-tuned on analyses from established morphological databases [Morpholex]. The AI performed well in analyses of word parts, consistent with the range of agreement one finds amongst existing hand-coded morphological resources. We used the result of the AI suggested morphological analyses of words to look at the word structures children meet in books and instructional texts, including how common Greek and Latin roots are and which prefixes and suffixes appear most often.

The research behind BAWF was published in October 2026 in the open-access journal Frontiers in Psychology (Green, Hiebert, & Coxhead, 2026). The article explains how we built and evaluated the database, reports what we found about the morphology of children’s reading materials and discusses implications for vocabulary instruction. A link to the full article is at the end of this post.

What the Database Reveals

An early analysis shows how concentrated English word-building is. About 100 prefixes account for 94% of the prefixed words in the database, and about 100 suffixes account for 95% of the suffixed words. These figures count word forms in the database. They don’t tell us how many words students will understand once they learn the affixes.

Still, they suggest that teachers can focus on a fairly short list of word parts.
Common prefixes include un-, re-, non-, pre-, dis-, sub-, and anti-. Common suffixes include -er, –ion, -al, -ly, –able, -ness, -ity, and -ive. A teacher might choose two or three and keep coming back to them as they appear in different texts and subjects.

Greek and Latin roots are usually treated as secondary-school vocabulary, but children meet them much earlier. We estimate that about 20% of the distinct words in children’s picture books contain classical roots, rising to about 30% by Grade 6. Excited, incredible, perfect, and special all come from Latin.

Four Ways to Explore BAWF With Your Students

The website, at https://cgg-projects.github.io/BAWF_V1/, is built for everyday lesson planning.

1 . Search in everyday language. The AI Morphology Studio accepts requests like “Find Grade 4 words with the suffix -ity” or “Show common words with the root spect.” It turns the request into a database search and lists the matches. The first request might return ability, security, and opportunity. The second might return perspective, spectator, expect, and aspect. You then pick the words that fit what your students are reading.

2. Explore a family and its meanings. Search for generate to see its relatives. The Explain buttons open prompts in AI tools such as Gemini, Claude, or ChatGPT that ask for definitions, a breakdown of word parts, and teaching ideas. Treat the answers as drafts to check, especially for unfamiliar words or words with complicated histories.

3. Turn a word list into a display. The Word Cloud tool turns a list of related words into a graphic for discussion or classroom materials. A cloud with attract, attraction, distract, and contract gives students a chance to notice the shared spelling and talk about how the meanings are related

4. Start with familiar words. The Curated Database is a smaller set based on word familiarity ratings from Marc Brysbaert and colleagues. It sets aside rare and technical forms so you can focus on words your students are likely to need. Other tools let you build custom downloads and link to dictionaries and etymology sources.

A Work in Progress

Some word relationships are obvious, like happy and unhappy. Others are historical, and most readers today wouldn’t find it challenging to infer one word from the other: Grave and engrave, for instance, or tone and tonic which share a Latin root. We are developing morphological transparency ratings that show how easily a reader can see the link between related words. Teachers will be able to use these ratings to find the connections students are most likely to recognize.

Because an AI model produced many of the analyses, some will be wrong. The website links to dictionaries and etymology sources so you can check an analysis before teaching it, and we welcome corrections. Next, we plan to add tools that analyze the vocabulary and word parts in teachers’ own reading passages.
BAWF is open access under a Creative Commons license, so you are free to download the data, adapt it, and build on it. We will post updates as the website and database grow.
To get started, search for a word your students already know, such as act, and look at its family.