---
title: "case: case marking — definition and key papers"
description: "case (case marking) is the Universal Dependencies relation that attaches a case-marking function word, such as a preposition, postposition…"
canonical: "https://research.elamur.ai/free/en/concepts/ud-case/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/concepts/ud-case/)

[home](https://md.research.elamur.ai/free/en/) / [concepts](https://md.research.elamur.ai/free/en/concepts/) / case: case marking

# 🌐 case: case marking

Universal Dependencies · 6 key papers · [all concepts](https://md.research.elamur.ai/free/en/concepts/)

Part of: [🗨️ Natural language processing](https://md.research.elamur.ai/free/en/concepts/natural-language-processing/) · [🌐 Universal Dependencies](https://md.research.elamur.ai/free/en/concepts/universal-dependencies/)

case (case marking) is the Universal Dependencies relation that attaches a case-marking function word, such as a preposition, postposition, clitic or possessive marker, to the nominal whose grammatical or semantic role it signals.

Also: case marker, adposition relation, prep and pobj (older Stanford-style labels)

## What it is

In [Universal Dependencies](https://md.research.elamur.ai/free/en/concepts/universal-dependencies/) (UD), content words are the heads of a tree and function words hang beneath them ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/)). The relation case follows this rule. A word like *on* in *on the table* does not govern *table*. *Table* is the head, and *on* is its dependent with the label case. The phrase then attaches upward as a nominal, usually through obl or nmod. The adposition is therefore a sister of the determiner, not a node above the noun.

This choice comes from the Universal Stanford Dependencies, which UD v1 merged with other standards ([Marneffe et al. 2014](https://md.research.elamur.ai/free/en/papers/marneffe-2014-universal-stanford-dependencies-cross-linguistic/); [Nivre et al. 2016](https://md.research.elamur.ai/free/en/papers/nivre-2016-universal-dependencies-v1-multilingual-treebank/)). The relation survived into the v2 inventory of 37 universal types ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)).

The case relation has a fixed head and a fixed dependent:

- The head is a nominal: a noun, proper noun, pronoun or numeral, and sometimes an adverb or adjective used nominally.
- The dependent is the marker itself: an adposition (ADP), or a particle or clitic. English possessive *'s* is split off as its own syntactic word, following the lexical integrity principle ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/)).

Several neighbouring relations are easy to confuse with case:

- mark: the same kind of word, but it introduces a clause rather than a nominal (*after he left*).
- compound:prt: verbal particles such as *up* in *pick up*.
- fixed: links the parts of a multiword marker (*because of*, *in front of*). The first part still bears case.
- obl and nmod: the relations of the nominal itself, never of the marker.

The guidelines define no universal subtypes. Languages may add their own, as UD allows in general ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)). One example is case:loc, which some Chinese treebanks use for postposed localizers. Such subtypes are uncommon.

## Examples

Arcs are written relation(head, dependent):

1. *She sat on the bench.* gives case(bench, on) and obl(sat, bench).
2. *Anna 's bicycle was stolen.* gives case(Anna, 's) and nmod:poss(bicycle, Anna).
3. French *le livre de Paul* ("Paul's book") gives case(Paul, de) and nmod(livre, Paul).

In the enhanced representation, the marker's lemma is copied into the head relation. Sentence 1 then carries obl:on, and the multiword *in front of* yields a single nmod:in_front_of edge ([Schuster & Manning 2016](https://md.research.elamur.ai/free/en/papers/schuster-2016-enhanced-english-universal-dependencies-improved/)).

## Across languages

The relation is the analytic side of case. Where case is inflectional, as in Russian or Finnish endings, nothing is split off and the information sits in the morphological feature Case. Where it is carried by separate words, it appears as a case arc. This covers prepositions in English and French, postpositions in Hindi or Turkish, and particles such as Japanese *ga* and *o*.

Treating free and bound markers this way makes trees more parallel across languages. A Finnish noun in the inessive and an English *in*-phrase have the same shape above the noun ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/)).

How core arguments are coded interacts with this relation. When a language marks subjects and objects with particles, those particles also take case. The distinction between core and oblique is then made by the head relation, not by the marker (Zeman 2017). Recent work treats such markers as one strategy among several for expressing a construction (Nivre 2025). Other work asks whether labels like case are cross-linguistic comparative concepts or language-particular categories (Boyer & Dehouck 2025).

The English and French data of UD 2.18 differ in degree but not in kind. In both, the pair noun head with adposition dependent dominates by far, followed by proper nouns and pronouns with adpositions. Case takes a larger share of all tokens in French, which uses *de* and *à* heavily. English has a distinctive pattern of proper noun with particle, which is the possessive *'s*. French shows adverbs taking adpositions, as in *d'ici*.

## In treebanks and research

For [dependency parsing](https://md.research.elamur.ai/free/en/concepts/dependency-parsing/), case is one of the most frequent and easiest relations, since markers form a short closed list. Its errors cluster where a word could also be mark, a particle or part of a fixed expression. Annotation checks across UD corpora look for exactly these inconsistencies ([Marneffe et al. 2017](https://md.research.elamur.ai/free/en/papers/marneffe-2017-assessing-annotation-consistency-universal-dependencies/)).

The content-head design was chosen so that predicate–argument structure is visible for downstream use. It underpinned the multilingual parsing shared tasks run by CoNLL in 2017 and 2018 ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/); [Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)). Tutorials on UD present the case analysis as a central example of these principles (Zeman 2021). The relation also figures in UD's treatment of nonverbal predicates, where an oblique nominal is the predicate (Nivre et al. 2026).

Beyond parsing, case marking is studied as a typological variable in [machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/). Experiments with synthetic languages found two results (Bisazza et al. 2021):

- Case suffixes let models resolve subject–object ambiguity when word order is free.
- Fixed-order sources still translated better when training data were limited.

See also [treebanks](https://md.research.elamur.ai/free/en/topics/treebanks/) and [morphology](https://md.research.elamur.ai/free/en/concepts/morphology/).

How often the relation case occurs: its share of all tokens Counted by code over the Universal Dependencies 2.18 treebanks of English and French (every token once, subtypes with their base relation).

**The numbers**

| Treebanks | Share |
| --- | --- |
| English treebanks | 8.43% of tokens (94,904 tokens) |
| French treebanks | 11.77% of tokens (128,358 tokens) |

What case joins: the most frequent head tag → dependent tag The five most frequent in each language's treebanks, by tokens.

**The numbers**

| Pair | Tokens |
| --- | --- |
| English: NOUN → ADP | 56,903 |
| English: PROPN → ADP | 20,226 |
| English: PRON → ADP | 7,323 |
| English: PROPN → PART | 2,100 |
| English: NUM → ADP | 1,996 |
| French: NOUN → ADP | 91,773 |
| French: PROPN → ADP | 21,379 |
| French: PRON → ADP | 7,258 |
| French: NUM → ADP | 3,471 |
| French: ADV → ADP | 1,529 |

## 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.

- [Universal Dependencies v1: A Multilingual Treebank Collection](https://md.research.elamur.ai/free/en/papers/nivre-2016-universal-dependencies-v1-multilingual-treebank/) — Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter et al. · 2016 · 1,741 citations
- [Universal Stanford dependencies: A cross-linguistic typology](https://md.research.elamur.ai/free/en/papers/marneffe-2014-universal-stanford-dependencies-cross-linguistic/) — Marie-Catherine de Marneffe, Timothy Dozat, Natalia Silveira et al. · 2014 · 627 citations
- [Universal Dependencies](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/) — Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre et al. · 2021 · 480 citations
- [Enhanced English Universal Dependencies: An Improved Representation for Natural Language Understanding Tasks](https://md.research.elamur.ai/free/en/papers/schuster-2016-enhanced-english-universal-dependencies-improved/) — Sebastian Schuster, Christopher D. Manning · 2016 · 336 citations
- [Universal Dependencies v2: An Evergrowing Multilingual Treebank Collection](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/) — Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter et al. · 2020 · 325 citations
- [Assessing the Annotation Consistency of the Universal Dependencies Corpora](https://md.research.elamur.ai/free/en/papers/marneffe-2017-assessing-annotation-consistency-universal-dependencies/) — Marie-Catherine de Marneffe, Matias Grioni, Jenna Kanerva et al. · 2017 · 30 citations

## Related concepts

- [Universal Dependencies](https://md.research.elamur.ai/free/en/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](https://md.research.elamur.ai/free/en/concepts/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](https://md.research.elamur.ai/free/en/topics/universal-dependencies/) · [🌳 Dependency parsing](https://md.research.elamur.ai/free/en/topics/dependency-parsing/)

Shelves: [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/)

- [← previous concept BLEU score](https://md.research.elamur.ai/free/en/concepts/bleu/)
- [next concept → cc: coordinating conjunction](https://md.research.elamur.ai/free/en/concepts/ud-cc/)
- [↑ back to concepts](https://md.research.elamur.ai/free/en/concepts/)