Skip to main content

Finding noun chunks from grammar links

Compare spaCy phrases with small dependency chunks

Learn how noun chunks group nouns with attached words, using spaCy and a local UDPipe dependency parse.

After counting adjacent phrases, the Riverton Workforce Lab has a practical choice to make: should basic spreadsheet skills become one review item or three separate words?

A grammar-aware method offers one convention for making that choice. It can group candidate phrases for review, but it does not create a single correct inventory of English phrases.

Note

The Riverton Workforce Lab, its job board, and its training flyer are fictional and were created for teaching.

TipWhat you will learn

Use this lesson to practice how to:

• define a noun chunk in plain language;
• extract noun chunks with a trained spaCy pipeline;
• build simple chunks from a UDPipe dependency parse;
• explain why chunk definitions differ across tools;
• separate agreement from accuracy; and
• explain why parser errors affect chunk rules.

Ask a trained pipeline for noun chunks

This setup chunk reads the sentence file, runs spaCy, and later supports a small UDPipe-based rule. A noun chunk is a short, flat phrase built around one noun: the noun plus the determiners, numbers, compounds, and describing words directly in front of it. Chunks do not nest and do not reach past the noun, so evening classes in October becomes evening classes and October, not one long phrase. Where a chunk stops is a convention, and tools draw the line in different places.

library(readr)
library(dplyr)
library(tibble)
library(purrr)
library(stringr)
library(tokenizers)
library(udpipe)
library(spacyr)
source("R/use-spacy.R")

sentences <- read_csv(
  "data/workforce/workforce_sentences.csv",
  na = character(),
  col_types = cols(
    sentence_id = col_character(),
    document_id = col_character(),
    source_line = col_character(),
    text = col_character(),
    reference_label = col_character(),
    uncertainty = col_character(),
    annotator_id = col_character(),
    rationale = col_character(),
    codebook_version = col_character(),
    codebook_hash = col_character(),
    derived = col_character(),
    transformation = col_character()
  )
)

pipeline <- use_project_spacy()
pipeline_info <- spacy_pipeline_version()
parsed_spacy <- spacy_parse(
  setNames(sentences$text, sentences$sentence_id),
  pos = TRUE,
  lemma = TRUE,
  entity = TRUE,
  dependency = TRUE,
  nounphrase = TRUE
)
spacy_phrases <- nounphrase_extract(parsed_spacy) |>
  as_tibble() |>
  transmute(doc_id, chunk = str_replace_all(nounphrase, "_", " "))

example_text <- "Riverton Skills Centre will run evening classes in October."
example_parsed <- spacy_parse(
  c(example = example_text),
  pos = TRUE,
  lemma = TRUE,
  entity = TRUE,
  dependency = TRUE,
  nounphrase = TRUE
)
example_phrases <- nounphrase_extract(example_parsed) |>
  as_tibble() |>
  transmute(chunk = str_replace_all(nounphrase, "_", " "))

spacy_finalize()

pipeline_table <- tibble(
  field = names(pipeline_info),
  value = unlist(pipeline_info, use.names = FALSE)
)

spaCy pipeline used for noun chunks

Pipeline field	Value
name	core_web_sm
version	3.8.0
lang	en
license	MIT
spacy	3.8.7

First spaCy noun chunks from the Riverton sentences

Sentence ID	spaCy noun chunk
s001	Paid 12-week training
s002	No prior data experience
s003	Evening schedules
s004	Applicants
s004	basic spreadsheet skills
s005	A high school diploma
s006	The employer
s006	certification training
s007	Rotating night shifts
s007	part
s007	the job
s008	Workers

The released spaCy pipeline returns 40 noun chunks for the 28 sentences. For the check sentence, it returns three flat chunks: Riverton Skills Centre, evening classes, and October.

Build chunks from dependency links

The UDPipe model used here was trained for this site on 500 sentences. On 13 held-out political weblog posts from 2003 to 2006, its unlabelled attachment score was 0.7235 and its labelled attachment score was 0.6561. spaCy’s pipeline is a released production model trained on far more text. This comparison is a teaching device, not a contest between equals.

The local UDPipe model has no tokenizer, so the lesson gives it one token per line. In the code, \(tokens) is R shorthand for a small function that handles one sentence’s tokens at a time.

This classroom rule starts from each token tagged as NOUN or PROPN. It then collects direct dependents whose relation is det, amod, compound, or nummod: a determiner such as a or the, a describing word, a noun working with another noun, or a number modifying a noun.

parser <- udpipe_load_model("data/treebank/en_ewt-500-parser.udpipe")

vertical <- sentences$text |>
  tokenize_words(lowercase = FALSE, strip_punct = FALSE) |>
  map_chr(\(tokens) paste(tokens, collapse = "\n"))

annotated <- udpipe_annotate(
  parser,
  x = vertical,
  doc_id = sentences$sentence_id,
  tokenizer = "vertical",
  tagger = "default",
  parser = "default"
) |>
  as.data.frame() |>
  as_tibble()

token_count_gap <- parsed_spacy |>
  count(doc_id, name = "spacy_tokens") |>
  inner_join(
    annotated |>
      count(doc_id, name = "udpipe_tokens"),
    by = "doc_id"
  ) |>
  filter(spacy_tokens != udpipe_tokens)

chunk_relations <- c("det", "amod", "compound", "nummod")

collect_chunk <- function(current_doc_id, current_token_id) {
  annotated |>
    filter(
      doc_id == current_doc_id,
      token_id == current_token_id |
        (head_token_id == current_token_id & dep_rel %in% chunk_relations)
    ) |>
    arrange(as.integer(token_id)) |>
    pull(token) |>
    str_c(collapse = " ")
}

udpipe_chunks <- annotated |>
  filter(upos %in% c("NOUN", "PROPN")) |>
  transmute(
    doc_id,
    token_id,
    chunk = map2_chr(doc_id, token_id, collect_chunk)
  ) |>
  select(doc_id, chunk)

The one token-count difference between the two parses

Sentence ID	spaCy tokens	UDPipe tokens
s017	8	7

First noun chunks built from dependency links

Sentence ID	UDPipe-built chunk
s001	Paid
s001	Paid week
s001	week training
s002	No prior data experience
s003	Evening schedules
s004	spreadsheet
s004	need basic spreadsheet skills
s005	A high school
s006	The employer pays
s006	certification
s006	certification training
s007	night

The two systems report different row counts because they split words differently. spaCy separates driver's in s017 into driver and 's; the tokenizer feeding UDPipe keeps it as one token, which accounts for the 179 versus 178 row counts. The printed spaCy chunk strings still keep Paid 12-week training and A valid driver's license; the gap is in token rows, not those displayed chunk strings.

This rule emits one chunk for every noun-like tag, so chunks can overlap. Paid, Paid week, and week training all come from the same four words in s001. The rule also pastes a head to its dependents without checking that the words sit next to each other, so Paid week skips over 12- and is not a string in the sentence. Paid appears because the small tagger labelled it PROPN, which is wrong here: it describes the training. A chunk rule inherits tagging mistakes before it makes its own.

Compare the two lists

Exact agreement is strict: the sentence ID and the chunk text must both match. That makes agreement a string match, not accuracy. Neither list is ground truth, and spaCy chunks are non-overlapping while the UDPipe-built list can overlap.

agreed_chunks <- inner_join(
  distinct(spacy_phrases),
  distinct(udpipe_chunks),
  by = c("doc_id", "chunk")
)

comparison_summary <- tibble(
  measure = c("spaCy chunks", "UDPipe-built chunks", "exact agreements"),
  count = c(nrow(spacy_phrases), nrow(udpipe_chunks), nrow(agreed_chunks))
)

different_examples <- tibble(
  method = c("spaCy", "UDPipe-built"),
  sentence_id = c("s005", "s005"),
  chunk = c("A high school diploma", "A high school"),
  explanation = c(
    "spaCy keeps the full education phrase.",
    "The dependency rule attaches words to school and leaves diploma outside this chunk."
  )
)

spreadsheet_examples <- bind_rows(
  spacy_phrases |>
    filter(doc_id == "s004", str_detect(chunk, "spreadsheet")) |>
    mutate(method = "spaCy"),
  udpipe_chunks |>
    filter(doc_id == "s004", str_detect(chunk, "spreadsheet")) |>
    mutate(method = "UDPipe-built")
) |>
  select(method, doc_id, chunk)

String matches across two chunking conventions

Measure	Count
spaCy chunks	40
UDPipe-built chunks	56
exact agreements	25

One sentence where the two chunk methods differ

Method	Sentence ID	Chunk	Explanation
spaCy	s005	A high school diploma	spaCy keeps the full education phrase.
UDPipe-built	s005	A high school	The dependency rule attaches words to school and leaves diploma outside this chunk.

How each method handles the spreadsheet-skills phrase

Method	Sentence ID	Chunk
spaCy	s004	basic spreadsheet skills
UDPipe-built	s004	spreadsheet
UDPipe-built	s004	need basic spreadsheet skills

The two lists have 25 exact string matches. That is agreement, not accuracy, and the counts 40 and 56 are not two measurements of the same thing. For the opening question, spaCy groups basic spreadsheet skills, while the UDPipe-built rule also emits spreadsheet and overreaches to need basic spreadsheet skills. A bigram takes neighbouring words even across phrase boundaries; a chunk groups words the parser links together. Whether that helps depends on whether the parser is right.

What to remember

• Noun chunks are flat, non-overlapping spans under a tool-specific convention.
• spaCy returns 40 noun chunks for these 28 sentences.
• The UDPipe-built rule returns 56 overlapping chunks from a deliberately weak teaching model.
• The 25 exact matches are agreement, not accuracy.
• Paid week shows how a rule can manufacture text by pasting non-adjacent tokens.
• Tokenization alone explains the 179 versus 178 row-count gap.

Review chunk lists as prompts for human inspection. When the two tools disagree, check the sentence, the tokenization, and the parse before drawing any conclusion about the phrase.

Sources

• spaCy documentation for linguistic features
• spacyr package documentation
• UDPipe R package reference
• Universal Dependencies relation documentation