Word Classes

Nouns & Pronouns

The words that name things and the words that stand in for them. Three coarse categories — NOUN, PROPN, PRON — and the nine Penn tags the parser pairs with them. The fine tags add exactly two things the coarse scale throws away: whether the word is singular or plural, and whether it is a name.

What you’ll learn

Tell a common noun from a proper noun on both scales
Choose between NN and NNS, and between NNP and NNPS
Read the five pronoun tags, including the two that are not pronouns on the coarse scale
Recognise the nouns and pronouns this model tags in a way you might not expect

How this reference is laid out

Each page takes a few coarse categories and, under each one, the fine tags the parser actually pairs with it. That order is deliberate: the POS exercise asks the coarse question first, and the Tags exercise then asks the finer one, so reading coarse-to-fine is reading the way the drills are built.

Every example sentence on these pages was run through the app’s own parser, and the chips show exactly what came back — the ringed word is the one the entry is about, but the rest are real too. Where the parser does something surprising, the page says so rather than quietly picking a sentence that hides it, because the Practice Lab grades against the same parser and would otherwise contradict the lesson.

Three categories, nine tags

NOUNNoun

A word that names a person, place, thing, substance or idea — kettle, stove, courage. The category is not about concreteness: an abstract idea like practice is exactly as much a noun as a kettle is. What makes a word a noun is the job it does — it can be the subject of a verb, the object of a verb or a preposition, and it can take a determiner in front of it (the kettle, some courage).

How spaCy decides: mostly from the word’s neighbours. A word after the, a or an adjective, and before a verb or at the end of a prepositional phrase, is a noun. Capitalisation pushes a word towards PROPN instead; a pronoun-like spelling pushes it towards PRON.

The^DETkettle^NOUNwhistled^VERBon^ADPthe^DETstove^NOUN.

Courage^NOUNgrows^VERBwith^ADPpractice^NOUN.

NNNoun, singular or mass

A common noun that is either singular (lighthouse, ship) or mass — uncountable, with no plural in ordinary use (happiness, sand, flour). The two are one tag because the grammar treats them alike: both take a singular verb and neither takes a plural ending.

How spaCy decides: the word does not end in a plural marker and is not capitalised as a name. This is by far the most common tag in the corpus — 112 of its tokens carry it.

The^DTlighthouse^NNguides^VBZevery^DTship^NNhome^NN.

Happiness^NNis^VBZhard^JJto^TOmeasure^VB.

NNSNoun, plural

A common noun that is plural. Regular plurals in -s/-es (wolves, hills, boxes) and irregular ones (mice, children) all take the same tag — the tag records the number, not how the plural was formed.

How spaCy decides: the ending, checked against what it knows about the word. That is why a plural that also spells a verb form can go wrong — see the warning under VBZ on Verbs & Auxiliaries.

Wolves^NNShowled^VBDat^INthe^DTdistant^JJhills^NNS.

Two^CDmice^NNShid^VBDbehind^INthe^DTboxes^NNS.

PROPNProper noun

A name for one particular person, place, organisation or time — Marisol, Nairobi, March. A proper noun refers directly rather than describing: city tells you what kind of thing something is, Nairobi tells you which one. In English they are capitalised, and that capital letter is the parser’s strongest clue.

How spaCy decides: capitalisation away from the start of a sentence, plus whether it has seen the word used as a name. Months and days are proper nouns here; the plural Tuesdays, though, drops to NOUN/NNS — see the second NNP example.

Marisol^PROPNflew^VERBto^ADPNairobi^PROPNin^ADPMarch^PROPN.

Three names in one sentence — a person, a place and a day of the week:

Marta^PROPNflew^VERBto^ADPLima^PROPNon^ADPTuesday^PROPN.

NNPProper noun, singular

A singular proper noun. Titles and their abbreviations count too: Dr. is an NNP in its own right, sitting beside the name it introduces rather than fusing with it.

How spaCy decides: a capitalised word that is not a sentence-initial common word. A name it has never seen is still usually NNP — an invented surname in a made-up sentence parses correctly, which is more than can be said for an invented common noun.

Clara^NNPmoved^VBDto^INToronto^NNPlast^JJspring^NN.

Dr.^NNPOkafor^NNPteaches^VBZon^INTuesdays^NNS.

NNPSProper noun, plural

A plural proper noun — a people, a family, a band: the Romans, the Smiths. This is the rarest of the four noun tags: there is not a single one in the app’s corpus, so if the Tags exercise ever offers it you are looking at a newly added sentence.

How spaCy decides: a capitalised word ending in -s that it reads as a group. The test is not just the ending — a plural-looking name for a single thing stays singular, and the parser applies that rule more widely than a grammar book would.

The^DTRomans^NNPSbuilt^VBDlong^JJstraight^JJroads^NNS.

The^DTSmiths^NNPSmoved^VBDnext^JJdoor^NNin^INJune^NNP.

The^DTAlps^NNPdivide^VBPItaly^NNPfrom^INSwitzerland^NNP.

PRONPronoun

A word that stands in for a noun already known or about to be asked for — she, it, them, everyone, who. Because it replaces a whole noun phrase, a pronoun normally takes no determiner: you cannot say the she. On the coarse scale this is one category; on the fine scale it splits five ways, by what kind of pronoun it is.

How spaCy decides: pronouns are a closed set, so this is mostly a lookup. The interesting cases are the indefinite pronouns — everyone, something, nobody — which the coarse scale files as PRON while the fine scale gives them a plain NN. Both are right: one scale is describing the job, the other the form.

She^PRONhanded^VERBit^PRONto^ADPthem^PRONwithout^ADPa^DETword^NOUN.

Everyone^PRONbrought^VERBsomething^PRONto^PARTshare^VERB.

PRPPersonal pronoun

A personal pronoun in any case: subject forms (they, we, I), object forms (us, them, me), and the reflexives (herself, themselves). The tag does not record which case the word is in — the dependency label does that, by making it nsubj or dobj.

How spaCy decides: a closed list. Mine, yours and the other independent possessives are also here rather than under PRP$, though not entirely consistently — yours comes back PRP$ in a question like Which coat is yours?

They^PRPinvited^VBDus^PRP,and^CCwe^PRPaccepted^VBD.

She^PRPtaught^VBDherself^PRPto^TOjuggle^VB.

PRP$Possessive pronoun

A possessive pronoun used before a noun — her brother, his keys, our neighbours, their fence. Many grammars call these possessive determiners, and in a Reed-Kellogg diagram they sit on a slant under the noun exactly like the does. The parser keeps them in the pronoun family on both scales: PRON coarse, PRP$ fine.

How spaCy decides: a closed list — my, your, his, her, its, our, their. Note that his and her can also be plain PRP (I saw her), and only the position before a noun tells the two apart.

Her^PRP$brother^NNlost^VBDhis^PRP$keys^NNSagain^RB.

Our^PRP$neighbours^NNSpainted^VBDtheir^PRP$fence^NNgreen^JJ.

WPWh-pronoun

A wh-pronoun: who, whom, what, and which when it stands alone. It either asks a question (Who left these boots?) or introduces a relative clause (the pig, who was little wiser). Same tag both ways — the fine scale does not distinguish interrogative from relative use.

How spaCy decides: the word is one of the wh-pronouns and is doing a noun’s job. When the same spelling introduces a noun instead (which road) it becomes WDT, on Determiners & Function Words.

Who^WPleft^VBDthese^DTmuddy^JJboots^NNShere^RB?

Tell^VBme^PRPwhom^WPyou^PRPmet^VBDat^INthe^DTfair^NN.

WP$Possessive wh-pronoun

The possessive wh-pronoun, and there is only one: whose. It does for a question or a relative clause what PRP$ does for a statement — whose dog is built like her dog.

How spaCy decides: the spelling. Worth knowing before the POS exercise: on the coarse scale whose is DET, not PRON — the parser treats the possessive job as a determiner’s job, whereas for her and their it does not. The corpus has exactly one WP$, and it is DET.

The^DTneighbour^NNwhose^WP$dog^NNescaped^VBDapologised^VBD.

Whose^WP$scarf^NNis^VBZthis^DT?

EXExistential 'there'

Existential there — the dummy subject of there is and there were. It fills the subject slot without referring to anything: the sentence is not about a place, it is announcing that something exists. Contrast the adverb there in put it there, which is RB and points at a location.

How spaCy decides: there directly before a form of be. Two things follow that the drills will test you on. On the coarse scale this there is PRON. And the be after it is the main verb of the sentence — VERB, not AUX — because nothing else is there for it to help. In the dependency tree it carries the label expl.

There^EXwere^VBDthree^CDeggs^NNSleft^VBNin^INthe^DTcarton^NN.

There^EXis^VBZa^DTspider^NNin^INthe^DTbath^NN.

Key terms

Common noun: A noun that names a kind of thing rather than one particular thing — river, courage. NN when singular or mass, NNS when plural. The river flooded. Courage matters.
Proper noun: A name. Capitalised in English, and the capital is the parser’s main clue. NNP singular, NNPS plural. Marta flies for Avianca. The Alps were snowy.
Mass noun: An uncountable noun — sand, happiness. Tagged NN like a singular, because it behaves like one. Sand filled her shoes. She wished us happiness.
Personal pronoun: I, you, he, she, it, we, they and their object and reflexive forms. All PRP; the case is recorded by the dependency label, not the tag. She called him. We found it ourselves.
Existential there: The there of there is. A placeholder subject, tagged EX, with PRON as its coarse category. There is a bird in the yard. There were no seats left.
See: Function Word Relations

Practice on a real sentence

This sentence carries four of the tags above at once — NN, NNS, PRP$ and a relative WP. Try the coarse drill first: every his is a PRON, and who is one too.

“The first pig built his house of straw, and the second pig, who was little wiser, built his house of sticks.”

Parts of speech

Give every word its coarse Universal Dependencies category.

Words 0/21Attempts 0Correct 0Time 0:00

Sentence — select a part of speech below, then click the word(s) it applies to

The first pig built his house of straw, and the second pig, who was little wiser, built his house of sticks.