--- title: "AUX: auxiliary — definition and key papers — Elamur Research" description: "AUX (auxiliary) is the Universal Dependencies part-of-speech tag for function words that accompany a predicate and express grammatical…" canonical: "https://research.elamur.ai/free/en/concepts/upos-aux/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/concepts/upos-aux/) [home](https://md.research.elamur.ai/free/en/) / [concepts](https://md.research.elamur.ai/free/en/concepts/) / AUX: auxiliary # 🏷️ AUX: auxiliary Universal Dependencies · 7 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/) AUX (auxiliary) is the Universal Dependencies part-of-speech tag for function words that accompany a predicate and express grammatical categories such as tense, mood, aspect, voice or evidentiality, rather than lexical meaning. Copulas are included as well. Also: auxiliary verb, helping verb ## What it is In [Universal Dependencies](https://md.research.elamur.ai/free/en/concepts/universal-dependencies/) (UD), every word receives one tag from a closed list of 17 universal part-of-speech (UPOS) tags ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)). AUX marks the "helping" words of a clause. These words carry grammatical information about the event (when it happens, how certain it is, whether it is active or passive). The event itself is left to a content word. The UD guidelines define an auxiliary as a function word, usually verbal in origin, that attaches to a predicate. Several groups fall under the tag: - tense auxiliaries (English *will*, *have*) - passive auxiliaries (*be*, *get*) - modals (*can*, *must*) - the copula linking a subject to a non-verbal predicate (*is* in *she is tired*) In some languages, uninflected particles that mark verbal categories are also tagged AUX. This follows from UD's design. Content words are the heads of dependency relations. Function words such as auxiliaries, copulas and adpositions attach to them as dependents ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/)). The universal tagset of Slav Petrov, Dipanjan Das and Ryan McDonald had only twelve categories and no separate auxiliary class, so auxiliaries fell under VERB ([Petrov et al. 2012](https://md.research.elamur.ai/free/en/papers/petrov-2012-universal-part-speech-tagset/)). UD v1 extended this twelve-tag set to 17 tags, and two of the additions were AUX and PROPN ([Nivre et al. 2016](https://md.research.elamur.ai/free/en/papers/nivre-2016-universal-dependencies-v1-multilingual-treebank/)). ### Borderline cases The tag sits between several neighbouring categories: - AUX vs. VERB. The same lemma can take either tag, depending on its use. In *I have finished*, *have* is AUX; in *I have a cat*, it is VERB. English *do* is AUX in do-support (*did you go?*) and VERB as a main verb. In English, copula-like verbs such as *become* or *seem* are VERB, because only the basic copula, *be*, is tagged AUX. - AUX vs. PART. The infinitival *to* and the negator *not* are PART, even though they cluster with auxiliaries. ### Features and relations Auxiliaries often carry most of a clause's inflection. Typical morphological features, stored in the FEATS column of CoNLL-U files, are: - VerbForm, Mood, Tense and Aspect - Person and Number (agreement) - Voice - Polarity, which in v2 replaced the v1 feature Negative ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)) The feature inventory goes back to Interset, a universal set of morphosyntactic features ([Zeman 2008](https://md.research.elamur.ai/free/en/papers/zeman-2008-reusable-tagset-conversion-tagset-drivers/); [Nivre et al. 2016](https://md.research.elamur.ai/free/en/papers/nivre-2016-universal-dependencies-v1-multilingual-treebank/)). An AUX word normally attaches to its predicate through one of three relations: - aux, for tense, aspect, mood and evidentiality, that is, non-passive auxiliaries - aux:pass, a subtype of aux for passive auxiliaries ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)) - cop, for copulas ## Examples | Language | Sentence | AUX word | Relation | | --- | --- | --- | --- | | English | *She will leave.* | *will* | aux | | English | *It was stolen.* | *was* | aux:pass | | English | *He is happy.* | *is* | cop | | French | *Il a mangé.* | *a* | aux | | French | *Elle est partie.* | *est* | aux | | French | *Elle est malade.* | *est* | cop | | Czech | *Udělal bych to.* | *bych* | aux | In the English copular sentence, the adjective *happy* heads the clause and *is* depends on it. The same holds in French *Elle est malade*. French *être* therefore takes both relations: aux in compound tenses such as *Elle est partie*, and cop with a non-verbal predicate. In elliptical answers such as *Yes, I do*, no content word is left, so the auxiliary itself can become the root. ## Across languages Languages differ in how large and how verbal their AUX class is: - English has a fairly closed set: modals plus *be*, *have*, *do* and *get*. - French uses *avoir* and *être* to form compound tenses. *Être* also serves as the passive auxiliary and as the copula. - Slavic languages such as Czech have conditional and past auxiliaries that behave like clitics. - Many languages of Asia mark tense or aspect with invariant particles. These may be tagged AUX when they function as auxiliaries. A frequent point of disagreement concerns modal verbs. Some treebanks tag modals as AUX, while others keep them as VERB with a full clausal complement. Such cross-language choices are recorded in each language's documentation, as the framework permits ([Nivre et al. 2020](https://md.research.elamur.ai/free/en/papers/nivre-2020-universal-dependencies-v2-evergrowing-multilingual/)). ## In treebanks and research Besides the regular aux, aux:pass and cop attachments, AUX words occasionally take other relations: - root, mostly in ellipsis - reparandum, for disfluencies in spoken data Because the AUX/VERB and AUX/PART boundaries depend on conventions, auxiliaries can be a source of inconsistency within a language. General methods for checking part-of-speech consistency apply here. One such method compares UPOS trigram distributions between treebanks of the same language using a symmetric divergence score. It covers all tags together rather than singling out AUX ([Aggarwal & Zeman 2020](https://md.research.elamur.ai/free/en/papers/aggarwal-2020-estimating-pos-annotation-consistency-different/)). Annotation consistency across UD corpora has also been assessed more broadly ([Marneffe et al. 2017](https://md.research.elamur.ai/free/en/papers/marneffe-2017-assessing-annotation-consistency-universal-dependencies/)). For [part-of-speech tagging](https://md.research.elamur.ai/free/en/concepts/part-of-speech-tagging/) and [dependency parsing](https://md.research.elamur.ai/free/en/concepts/dependency-parsing/), the AUX tag works together with UD's content-word-as-head convention. The lexical predicate heads the clause, and the auxiliary depends on it. A central aim of the scheme was to make clauses with and without auxiliaries look alike, both within a single language and across languages ([Marneffe et al. 2021](https://md.research.elamur.ai/free/en/papers/marneffe-2021-universal-dependencies/)). How often the tag AUX 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 | 6.45% of tokens (72,609 tokens) | | French treebanks | 3.54% of tokens (38,625 tokens) | The relations AUX words take most often The five most frequent in each language's treebanks, by tokens. **The numbers** | Pair | Tokens | | --- | --- | | English: aux | 42,565 | | English: cop | 26,649 | | English: root | 1,611 | | English: parataxis | 501 | | English: reparandum | 345 | | French: aux | 24,682 | | French: cop | 13,413 | | French: reparandum | 181 | | French: root | 170 | | French: advcl | 52 | ## 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 - [A Universal Part-of-Speech Tagset](https://md.research.elamur.ai/free/en/papers/petrov-2012-universal-part-speech-tagset/) — Slav Petrov, Dipanjan Das, Ryan McDonald · 2012 · 1,180 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 - [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 - [Reusable Tagset Conversion Using Tagset Drivers](https://md.research.elamur.ai/free/en/papers/zeman-2008-reusable-tagset-conversion-tagset-drivers/) — Daniel Zeman · 2008 · 181 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 - [Estimating POS Annotation Consistency of Different Treebanks in a Language](https://md.research.elamur.ai/free/en/papers/aggarwal-2020-estimating-pos-annotation-consistency-different/) — Akshay Aggarwal, Daniel Zeman · 2020 · 7 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… - [Part-of-speech tagging](https://md.research.elamur.ai/free/en/concepts/part-of-speech-tagging/) — Part-of-speech tagging is the task of assigning each token in a text a grammatical category, such as noun, verb, adjective or determiner… - [Morphology](https://md.research.elamur.ai/free/en/concepts/morphology/) — Morphology is the branch of linguistics that studies the internal structure of words. It covers how words are built from meaningful units… ## More papers Topics: [🔤 Part-of-speech tagging](https://md.research.elamur.ai/free/en/topics/part-of-speech-tagging/) · [🌐 Universal Dependencies](https://md.research.elamur.ai/free/en/topics/universal-dependencies/) Shelves: [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) - [← previous concept aux: auxiliary](https://md.research.elamur.ai/free/en/concepts/ud-aux/) - [next concept → BLEU score](https://md.research.elamur.ai/free/en/concepts/bleu/) - [↑ back to concepts](https://md.research.elamur.ai/free/en/concepts/)