What Is a Lemma?

A lemma is the base form of a word that stands for all of its inflected forms in one part of speech. Walk, walks, walked and walking are four different word forms but one verb lemma, walk. The lemma is what you find as a headword in a dictionary, and it is one of the main units linguists use to count words in a text or in a person's vocabulary.

Examples of lemmas

Note the last two rows. Because a lemma is tied to a part of speech, the verb run and the noun run (as in a morning run) are usually counted as different lemmas, even though they share a spelling. Some researchers merge them into a "flemma" (a form-based lemma) to avoid this split.

Lemma vs word form vs word family

Different counting units give different answers to the question "how many words?":

Notice that saw here is the past of see; in another sentence it could be the noun or verb saw (a cutting tool). Working out the right lemma sometimes requires context, which is why automatic lemmatizers use part-of-speech tagging.

Why lemmas matter for vocabulary learners

They explain the numbers you read

Estimates of vocabulary size depend on the unit. Brysbaert and colleagues (2016) estimated that an average 20-year-old native speaker of American English knows about 42,000 lemmas, derived from about 11,000 word families. Goulden, Nation and Read (1990) estimated around 17,000 word families for educated native speakers. Neither study is wrong; they count different things. When an app or course claims "learn 3,000 words," it is worth asking: forms, lemmas or families?

Inflections are cheap, but not free

Counting by lemma assumes that once you know go, you can handle went and gone. That is true for advanced learners and for regular forms, but beginners often need to learn irregular forms (went, mice, children, better) almost as separate words.

They make text analysis meaningful

A word frequency counter that does not lemmatize will list is, are and was separately, which can hide the fact that be dominates English text. Similarly, a type-token ratio calculated on lemmas will be lower than one calculated on raw word forms.

How to use the idea in study

• Save the lemma, note the irregular forms. If you save seek, add sought to the card.
• Look up the lemma, not the form. If ubiquities returns nothing, try ubiquity. Wordie's lookup card shows the part of speech, which helps you tell the noun lemma from the verb lemma.
• Separate lemmas with different meanings. Present (noun, a gift) and present (verb, to show) deserve separate cards.
• Grow into families. Once a lemma is secure, add its derived forms; see morpheme.

Common mistakes

• Comparing figures that use different units. "20,000 words" in lemmas is a smaller vocabulary than 20,000 word families.
• Assuming the dictionary form is the most common form. Some words appear mostly inflected: according far more than accord.
• Ignoring part of speech. Knowing the verb contract does not guarantee you know the noun, which is also stressed differently (CONtract vs conTRACT).

Sources: Brysbaert, M., Stevens, M., Mandera, P. & Keuleers, E. (2016), How many words do we know?, Frontiers in Psychology; Goulden, R., Nation, P. & Read, J. (1990), How large can a receptive vocabulary be?, Applied Linguistics.