🌐 xcomp: open clausal complement

Universal Dependencies · 6 key papers · all concepts

Part of: 🗨️ Natural language processing · 🌐 Universal Dependencies

xcomp (open clausal complement) is the Universal Dependencies relation that links a predicate to a complement clause or predicative phrase. That complement has no subject of its own; its understood subject is fixed by an argument of the higher clause, as in "She wants to leave."

Also: open clausal complement, xcomp, predicative complement (in part)

What it is

In Universal Dependencies, xcomp names a complement of a verb or adjective that is "open". The complement lacks an overt subject, and its missing subject is identified with the subject or object of the governing predicate. This is the classic syntactic pattern of control and raising, which goes back to Bresnan (1982) in Lexical-Functional Grammar.

The head of the relation is the higher predicate, such as want, seem, make or ready. The dependent is the head of the complement, which is usually a verb but can be an adjective or a noun. Because UD makes content words heads and attaches function words such as to below them (Marneffe et al. 2021), the arc runs from want to leave, not to to.

The relation comes from the universal Stanford scheme (Marneffe et al. 2014). That scheme fed into UD's general split between phrasal and clausal dependents (Nivre et al. 2016). Its main neighbours are these:

Unlike obj, xcomp is predicative rather than referential. Unlike acl, it depends on a predicate, not on a nominal. Language-specific subtypes are allowed by the guidelines (Nivre et al. 2020) but are rare for this relation. Where they appear, they typically single out secondary predicates (xcomp:pred).

Examples

• She wants to leave. gives xcomp(wants, leave). The subject of leave is she.
• They made him angry. gives xcomp(made, angry). The object him controls angry.
• The book is easy to read. gives xcomp(easy, read). The head here is an adjective.
• He seems tired. gives xcomp(seems, tired). This is raising: he is a grammatical argument of seems only.

Compare She said that he left. There the complement has its own subject, so the relation is ccomp(said, left).

The basic tree leaves the controlled subject implicit. The enhanced English representation adds an edge from the embedded predicate to its controller, labelled nsubj:xsubj, as in nsubj:xsubj(leave, She) (Schuster & Manning 2016). More generally, the enhanced graph adds implicit relations for control and raising on top of the tree (Marneffe et al. 2021).

Across languages

The relation is universal, but its share and its typical dependents vary.

In English and French treebanks, the dominant pattern by far is verb governing verb: infinitival control and raising.

• English: the next most frequent pair is verb governing adjective (become rich, keep quiet).
• French: verb governing noun comes before verb governing adjective, reflecting predicative nominals after verbs like nommer or devenir.
• Both languages: adjectives heading verbal complements (easy to read, facile à lire) form a smaller but steady class.

See the English and French pages for the treebanks themselves.

Languages without infinitives, such as Modern Greek or Bulgarian, express control with finite subjunctive clauses. This raises the question of whether such clauses are xcomp or ccomp. Treebanks have answered it in different ways.

Where secondary predication is marked by case, as with instrumental predicatives in Slavic languages, xcomp often covers depictives that English would treat as adjuncts. How clausal complements fit UD's core-argument system has been examined separately (Zeman 2017). Recent work recasts such choices as "constructions and strategies" that should be annotated uniformly across languages (Nivre 2025).

In treebanks and research

xcomp is one of the 37 universal relation types of UD v2 (Nivre et al. 2020). It is annotated in treebanks across the project.

Its boundaries with ccomp, advcl and obl are a known source of inconsistency. Annotation consistency across UD corpora has been audited systematically (Marneffe et al. 2017), and tutorials on UD treat the open/closed distinction as a core principle (Zeman 2021).

In dependency parsing, xcomp is scored like any other label in the multilingual evaluations built on UD, including the CoNLL shared tasks of 2017 and 2018 on raw-text parsing (Marneffe et al. 2021; Nivre et al. 2020). Errors on it commonly involve:

• confusing it with ccomp after verbs that take both complement types;
• confusing it with advcl for purpose infinitives;
• attaching it to to or a copula instead of the content head.

For downstream use, xcomp matters mainly through the subject it hides. Information extraction and semantic role labeling systems need the link between leave and she. That is why the enhanced representation restores this link explicitly (Schuster & Manning 2016). More recent UD work on nonverbal and nonprototypical predication revisits where predicative complements end and copular clauses begin (Nivre et al. 2026).

Key papers

The works this page cites, the most cited first; citations as counted by OpenAlex plus the papers of this library that cite the work, read 2026-10-24.

Related concepts

• Universal Dependencies — Universal Dependencies is a framework and an open collection of treebanks that annotates the grammar of many languages with one shared…
• Dependency parsing — Dependency parsing is the task of recovering a sentence's syntactic structure as directed, labelled links between its words. Each word is…

More papers

Topics: 🌐 Universal Dependencies · 🌳 Dependency parsing

Shelves: 🗣️ Language and Linguistics

This page in Markdown → https://md.research.elamur.ai/free/en/concepts/ud-xcomp/