Elamur: **Mova** · [Parse](https://mova.elamur.ai/free/en/parse/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Translate](https://md.translate.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · [MCP](https://mcp.elamur.ai/) · [Groupants](https://md.groupants.elamur.ai/free/en/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/repo-docs/parsers/)

[home](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [repo-docs](https://md.mova.elamur.ai/free/en/docs/repo-docs/) / Open parsers: witnesses and tests

# Open parsers: witnesses and tests

This page covers the open parsers Mova uses as witnesses and test references: for English, French, Ukrainian, Latin, Ancient Greek, Biblical Hebrew and Modern Hebrew. It says which licences allow what, how the parsers are used, and which parser serves which role. The papers behind the parsers are in the library.

## What it is

Open parsers and their models are supplied separately; Mova does not ship them. They serve three purposes:

- **Witnesses** for the teacher's trees: where independent parsers agree on a head and the teacher disagrees, the word is a suspect to check first.
- **Test references** for Mova's own `en` parser and for the parsers of the language packages.
- **Candidates** for transparent rule-based components (Link Grammar, ACE + ERG, Grammatical Framework).

A witness runner is configured through `CYCLE_NEURAL` ([tools/cycle](https://md.mova.elamur.ai/free/en/docs/tools/cycle/)). Its inputs and outputs must keep token alignment, and each model's licence is checked apart from its code's licence.

## Licences: build vs test

By the data-licence rule ([Data and model licences](https://md.mova.elamur.ai/free/en/docs/licence/DATA/)), share-alike, non-commercial and research-only models are for **tests and measurements only**, never built into Mova or shipped. A permissive code licence does not settle it: a model trained on restricted treebanks needs a data-licence check before any build use.

| parser | tree | model licence | use |
| --- | --- | --- | --- |
| UDPipe 1 (UD 2.5), UDPipe 2 (UD 2.15; straka-2018) | UD | CC BY-NC-SA 4.0 | tests only |
| UDify | UD (multilingual BERT) | LINDAT, NC | tests only |
| Stanza (qi-2020) | UD; PTB constituents | Apache-2.0 code; models trained on UD/PTB (some UD treebanks NC: grc/hbo PTNK, he HTB) | tests; build only after a data-licence check |
| Trankit (XLM-R; nguyen-2021) | UD | Apache-2.0; trained on UD | tests |
| spaCy `en_core_web_*` | ClearNLP-style dependencies | MIT (trained on OntoNotes) | tests; build only after a data-licence check |
| spaCy `fr_core_news_*`, `fr_dep_news_trf` (honnibal-2020) | UD-style, Sequoia labels | LGPL-3.0 | tests |
| hopsparser (code) + almanach CamemBERT 2.0 models `camembertv2-base-{gsd,sequoia,rhapsodie}`, `camembertav2-base-{gsd,sequoia,rhapsodie}` | UD (biaffine over CamemBERT/CamemBERTa encoders; grobol-2021, antoun-2024) | MIT code; weights MIT per model card; trained on French GSD (CC BY-SA 4.0), Sequoia (LGPL-LR), Rhapsodie (CC BY-SA 4.0) | tests; build only after a data-licence check |
| COMBO `combo-nlp` French (XLM-R base, GSD, UD 2.17) | UD (klimaszewski-2021) | GPL-3.0 code; model CC BY-SA 4.0 | tests only |
| Trankit `ukrainian` (XLM-R base) | UD (nguyen-2021) | Apache-2.0 code; weights licence not stated; trained on UD Ukrainian IU 2.5 (CC BY-NC-SA 4.0) | tests only |
| COMBO `combo-nlp` Ukrainian IU / ParlaMint (XLM-R base, UD 2.17) | UD (klimaszewski-2021) | GPL-3.0 code; IU model CC BY-NC-SA 4.0, ParlaMint model CC BY-SA 4.0 | tests only |
| Stanza `uk` packages `iu_*`, `parlamint_*` | UD (qi-2020) | Apache-2.0 code; weights licence not stated; trained on UD Ukrainian IU (CC BY-NC-SA 4.0) or ParlaMint (CC BY-SA 4.0) | tests only |
| spaCy `uk_core_news_lg`, `uk_core_news_trf` (honnibal-2020) | UD-style | MIT; trained on the silver Ukr-Synth (MIT) with CC0 vectors; encoder `ukr-models/xlm-roberta-base-uk` | tests; build only after a data-licence check of Ukr-Synth |
| Yasuoka `*-ukrainian-ud-goeswith` (ModernBERT large and base, RoBERTa large WECHSEL) | UD (POS and `goeswith` subword merging) | card says CC BY 4.0 / MIT; trained on all Ukrainian UD treebanks, so NC-SA applies | tests only |
| UDPipe 2 `uk_all` (UD 2.15) | UD | CC BY-NC-SA 4.0 | tests only; not run (needs the embedding service) |
| LatinCy `la_core_web_*` | UD (harmonised) | MIT | tests |
| OdyCy `grc_odycy_joint_*` | UD (Perseus/PROIEL) | MIT | tests |
| GreCy `grc_dep_*_trf` | UD | not stated | tests only |
| DictaBERT joint (he) | UD (HTB or IAHLT style) | CC BY 4.0; training data not checked | tests |
| YAP (he) | SPMRL labels | Apache-2.0 code; MILA models and lexicon | tests only |
| Stanford CoreNLP | Stanford/UD dependencies, PTB constituents | GPL-3.0 | tests only |
| Berkeley Neural Parser (benepar) | PTB constituents | MIT (trained on PTB) | tests |
| SuPar | dependencies, constituents, semantic dependencies | MIT (trained on PTB/UD/SDP) | tests |
| MaltParser + `engmalt` | Stanford-basic dependencies | BSD code; maltparser.org model terms | tests |
| AllenNLP, EasyCCG, depccg | dependencies, constituents, CCG | Apache-2.0 / MIT (trained on PTB, CCGbank) | tests |
| BLLIP (Charniak–Johnson) | PTB constituents | unclear (trained on PTB/LDC) | tests only |
| Link Grammar | links (hand-written rules) | LGPL-2.1 | candidate for build: transparent rules |
| ACE + ERG 2018 | HPSG + MRS semantics | MIT-like | candidate for build: transparent rules |
| Grammatical Framework + RGL + gf-wordnet | abstract syntax trees | LGPL/BSD | candidate for build: real ASTs |
| syntaxdot (Rust) | dependencies | MIT/Apache-2.0 | code to read; no English model published |

Code only, no pretrained English model: Enju (HPSG), TurboParser, Hexatagger, DiaParser. CLTK has no parser of its own (it wraps Stanza or a large model) and is not set up.

## How they are used

- **School witnesses.** `tools/cycle` runs the neural parsers over school sentences (runner path in `CYCLE_NEURAL`), pretokenized so trees align token by token; [`tools/parsecmp`](https://md.mova.elamur.ai/free/en/docs/tools/parsecmp/) compares them and `tools/treeagree` merges their agreement with the teacher's trees ([Training Mova](https://md.mova.elamur.ai/free/en/docs/repo-docs/training/)). Stanza and Trankit (UD, GPU) are the witnesses; a word where both agree on another head than the teacher is a suspect, and spaCy `trf` agreeing as well puts it at the top of the list. Agreement with a participating witness is not independent accuracy evidence; linguistic quality is measured on an admissible held-out gold set.
- **Convention, not error.** spaCy (ClearNLP) and SuPar (Stanford dependencies) make prepositions and copulas heads of their phrases. Low agreement with UD trees from them is a head-rule difference; use them as witnesses only after conversion. Constituency parsers (Stanza, benepar, SuPar CRF) serve bracket comparisons.
- **Language packages.** For each language, dev sets are parsed with gold syntactic words (multiword tokens already split, tokenizer off) and scored on all words, punctuation included: LAS on the full label and LAS-b on the label before the colon, each on the parser's own tags. A model trained on the same treebank is not a fair witness; a model from another treebank is fair but measures convention differences as much as quality.
- **Mixed-language text.** Mova's own parsers are one model per language; [`tools/mova-parse`](https://md.mova.elamur.ai/free/en/docs/tools/mova-parse/) cuts a text into language spans with `langseg` (English and French by dictionary) and sends each span to its language's model, writing `# lang` per sentence and `Lang=` on switched tokens. The neural witnesses above stay single-language.

Which parser serves which role:

| language | witness for in-treebank style | fair witness (other treebank) | notes |
| --- | --- | --- | --- |
| English | Stanza, Trankit | spaCy `trf` (after head-rule differences) | constituency: Stanza, benepar, SuPar |
| French (GSD) | hopsparser `camembertav2-base-gsd`, COMBO, Stanza `gsd`; UDPipe 2, Trankit `french` | Stanza `sequoia` (other treebank) | spaCy labels in Sequoia style; measurements in «French witnesses» |
| Ukrainian (IU) | Trankit `ukrainian`, COMBO IU, spaCy `trf` (silver Ukr-Synth), Stanza `iu_charlm` | COMBO ParlaMint, Stanza `parlamint_charlm` (other treebank); Yasuoka `modernbert-large` (both) | measurements in «Ukrainian witnesses»; ParlaMint is a second convention, not a better one |
| Latin (LLCT) | Stanza `llct`, UDPipe 2, LatinCy | none good | LLCT is medieval with its own conventions; for other text compare across treebanks |
| Ancient Greek (PTNK) | Stanza `ptnk`, UDPipe 2 `grc_all` | Stanza/Trankit `proiel`, OdyCy | Perseus/PROIEL models show the register gap of the Greek Pentateuch |
| Biblical Hebrew (PTNK) | Stanza `hbo`, UDPipe 2 `hbo_ptnk` | none (the only treebank) | vowel points and cantillation fed as they are |
| Modern Hebrew | UDPipe 2 `he_all`; Stanza `htb` (HTB style), Stanza `combined` and DictaBERT `iahlt_ud` (IAHLT style) | across the two conventions | YAP only after a label mapping |

Not measured here: raw-text parsing with multiword-token recovery.

## French witnesses (measured 2026-11-06)

Dev sets of UD 2.18 French GSD (1476 sentences, 35721 words) and Sequoia (412 sentences, 9999 words), gold tokens (multiword tokens split into syntactic words, tokenizer off), LAS and UAS over all words, punctuation included, on each parser's own output. Speed is sentences per second on the RTX 4000 SFF Ada (GSD dev, model already loaded, one warm-up batch excluded; batch 500 sentences for Stanza and Trankit, 64 for hopsparser, 1 for COMBO). GPU is the peak of the process as seen by `nvidia-smi` (it includes the allocator cache, so it overstates the need). spaCy `lg` runs on the CPU. Runner: [`tools/treeagree/bridge/fr_neural.py`](https://md.mova.elamur.ai/free/en/docs/tools/treeagree/); raw results `results.jsonl` in the run directory fr-parsers-2026-11-06. One run per parser, not repeated; the numbers are deterministic up to GPU arithmetic.

Bold marks a model trained on the treebank of that column (in-treebank style; a development set may have been used for model selection, so these are slightly optimistic); the rest are fair witnesses that carry the convention gap of another treebank. Sorted by the sum of the two LAS.

| parser | GSD LAS / UAS | Sequoia LAS / UAS | sent/s (GSD) | GPU MiB |
| --- | --- | --- | --- | --- |
| COMBO `combo-nlp` (GSD) | **94.7 / 96.6** | 90.5 / 94.3 | 23 | 3548 |
| hopsparser `camembertav2-base-gsd` | **94.5 / 96.6** | 90.1 / 94.4 | 116 | 10140 |
| hopsparser `camembertv2-base-gsd` | **94.3 / 96.5** | 90.1 / 94.5 | 97 | 9740 |
| Stanza `combined` | **93.3 / 95.6** | **90.7 / 94.0** | 64 | 8100 |
| Stanza `gsd` | **93.7 / 95.9** | 88.8 / 93.3 | 68 | 7604 |
| Trankit `french` (GSD) | **91.0 / 95.5** | 88.2 / 93.7 | 488 | 2588 |
| Stanza `sequoia` | 87.0 / 92.6 | **91.6 / 94.1** | 58 | 7376 |
| hopsparser `camembertv2-base-sequoia` | 83.8 / 89.7 | **87.3 / 90.9** | 188 | 8786 |
| hopsparser `camembertav2-base-sequoia` | 84.0 / 89.5 | **87.2 / 90.8** | 118 | 9206 |
| Trankit `french-sequoia` | 83.3 / 89.0 | **86.6 / 90.2** | 459 | 2606 |
| spaCy `fr_dep_news_trf` | 83.2 / 88.9 | 85.5 / 89.4 | 156 | 1788 |
| spaCy `fr_core_news_lg` | 74.9 / 81.5 | 79.7 / 84.7 | 182 | CPU |
| hopsparser `camembertav2-base-rhapsodie` | 76.6 / 87.2 | 76.0 / 86.2 | 107 | 9206 |
| hopsparser `camembertv2-base-rhapsodie` | 75.8 / 87.4 | 74.9 / 86.3 | 123 | 8786 |
| Stanza `rhapsodie` | 75.3 / 84.6 | 74.5 / 82.9 | 57 | 7498 |
| Trankit `french-partut` | 75.4 / 86.7 | 74.0 / 84.5 | 341 | 2600 |
| Stanza `partut` | 74.5 / 87.6 | 74.9 / 86.7 | 53 | 6658 |
| Stanza `parisstories` | 72.2 / 82.2 | 72.1 / 81.5 | 32 | 7188 |
| Trankit `french-spoken` | 67.8 / 83.6 | 67.9 / 83.4 | 375 | 2526 |

Reading the table:

- **Treebank version drift.** The hopsparser models were trained on UD 2.9 (model cards: GSD LAS 94.5 for `camembertav2-base-gsd`). On GSD dev 2.18 they reproduce that (94.5), but the Sequoia models drop to 87 on Sequoia dev 2.18 (UAS 90.9) although the runner reproduces each card's own parsed dev file exactly (99.99% agreement): the Sequoia annotation changed between the versions, so the Sequoia-trained models are measured against a moved target and Stanza `sequoia` (retrained on a later version) is the better Sequoia-style witness. Compare these columns only for the version stated.
- **Spoken and small treebanks** (Rhapsodie, ParTUT, ParisStories, Trankit `spoken`) are useful only as register witnesses; they score about 75 LAS on written text.
- **Agreement of the two best GSD-style GPU models** (hopsparser `camembertav2-base-gsd` and COMBO, both trained on GSD, so not independent): mutual UAS 97.3 on GSD dev; both right on 95.4% of words; at least one right on 97.8%. On Sequoia dev, hopsparser `camembertav2-base-sequoia` with Stanza `sequoia` agree on 88.9% (UAS).
- Not measured: raw-text parsing, the 2.19 restoration of GSD features, Rhapsodie/ParisStories dev as test sets, UDPipe 2 (needs its server; NC).

Models, revisions and checksums of the hopsparser downloads are in `hops/manifest.tsv` in the parsers directory of the data tree (outside Git; its own environment with hopsparser 0.8.0; COMBO has another). The COMBO batch mode fails on pretokenized input with combo-nlp 4.0.14 (a lemma-head size mismatch), so the runner parses one sentence at a time.

**Recommended French witness set for the Eel's agreement step.** GPU pair: hopsparser `camembertav2-base-gsd` (fast, 94.5 LAS on GSD) and COMBO (94.7); a third GSD-style vote from Stanza `gsd` if three are wanted. Fair witness: Stanza `sequoia` (other treebank, 87.0 LAS / 92.6 UAS on GSD dev, and the best Sequoia-style model at 91.6); Trankit `french` is the fast cheap alternative (488 sent/s, 2.6 GB). Agreement with a participating witness is not independent accuracy evidence (see above); the table says which witnesses carry the GSD convention and which carry another one.

Sources: hopsparser (grobol-2021), CamemBERT 2.0 (antoun-2024), Stanza (qi-2020), Trankit (nguyen-2021), spaCy (honnibal-2020), UDPipe 2 (straka-2018), COMBO (klimaszewski-2021), Universal Dependencies v2 (nivre-2020); GSD (guillaume-2019), Sequoia (candito-2012), Rhapsodie (lacheret-2014).

## Ukrainian witnesses (measured 2026-11-06)

Treebanks: UD 2.18 Ukrainian IU (dev 673 sentences, 12606 words; **CC BY-NC-SA 4.0** by its README and `LICENSE.txt`: tests and counts only; the README offers a commercial licence on request) and ParlaMint (dev 739 sentences, 10842 words; CC BY-SA 4.0: tests, as little as possible). Both are the `r2.18` tags of `UniversalDependencies/UD_Ukrainian-IU` and `UD_Ukrainian-ParlaMint`, the only Ukrainian treebanks in that release, kept outside Git in the research raw-data tree (`raw/uk-ud/ud-treebanks-v2.18/`). Measured as in the French section: gold tokens (Ukrainian has no multiword tokens), tokenizer off, LAS and UAS over all words, punctuation included, each parser on its own output. Speed is sentences per second on the RTX 4000 SFF Ada (IU dev, model already loaded, one warm-up batch excluded; batch 500 for Stanza and Trankit, 64 for spaCy `trf`, 1 for COMBO and the `goeswith` models). GPU is the process peak seen by `nvidia-smi` (it includes the allocator cache, so it overstates the need). spaCy `lg` runs on the CPU. Runner: [`tools/treeagree/bridge/uk_neural.py`](https://md.mova.elamur.ai/free/en/docs/tools/treeagree/); raw results `results.jsonl` in the run directory uk-parsers-2026-11-06. One run per parser, not repeated; the numbers are deterministic up to GPU arithmetic.

Bold marks a model trained on the treebank of that column (in-treebank style; a development set may have been used for model selection, so these are slightly optimistic); the rest are fair witnesses that carry the convention gap of another treebank. Sorted by the sum of the two LAS.

| parser | IU LAS / UAS | ParlaMint LAS / UAS | UPOS IU / PM | sent/s (IU) | GPU MiB |
| --- | --- | --- | --- | --- | --- |
| Trankit `ukrainian` (XLM-R base; IU, UD 2.5) | **91.98 / 94.15** | 84.74 / 92.77 | 98.4 / 98.1 | 306 | 3044 |
| Yasuoka `modernbert-large-ukrainian-ud-goeswith` | **88.56 / 91.63** | **87.38 / 91.46** | 95.8 / 95.7 | 3.3 | 5216 |
| COMBO `combo-nlp` IU (XLM-R base; UD 2.17) | **91.95 / 94.01** | 83.86 / 91.57 | 98.5 / 97.8 | 14.7 | 3546 |
| spaCy `uk_core_news_trf` | 91.91 / 94.15 | 83.62 / 91.63 | 97.9 / 97.5 | 134 | 1994 |
| COMBO `combo-nlp` ParlaMint | 83.91 / 90.85 | **91.02 / 94.39** | 97.2 / 98.9 | 22.6 | 3546 |
| Yasuoka `roberta-large-wechsel-ukrainian-ud-goeswith` | **87.82 / 91.39** | **86.66 / 91.09** | 96.0 / 94.9 | 3.5 | 3822 |
| Stanza `iu_charlm` | **87.74 / 90.83** | 80.49 / 88.68 | 98.0 / 97.9 | 44 | 9148 |
| Yasuoka `modernbert-base-ukrainian-ud-goeswith` | **84.18 / 88.55** | **83.67 / 87.77** | 96.4 / 94.4 | 2.6 | 2814 |
| Stanza `parlamint_charlm` | 78.38 / 85.78 | **88.47 / 92.11** | 96.7 / 99.0 | 43 | 9036 |
| spaCy `uk_core_news_lg` | 86.49 / 89.10 | 80.29 / 88.62 | 97.4 / 97.7 | 176 | CPU |
| Stanza `iu_nocharlm` | **85.40 / 88.88** | 78.56 / 86.64 | 97.1 / 97.6 | 45 | 7682 |
| Stanza `parlamint_nocharlm` | 74.68 / 82.24 | **87.25 / 91.41** | 95.1 / 98.7 | 60 | 7566 |
| Mova `en` parser, own tags (train → dev after `en xpos-fill`) | **73.33 / 79.60** | **79.28 / 83.48** | n/a | CPU, about 20000 words/s | CPU |
| Mova `en` parser, gold tags | **78.17 / 83.02** | **81.34 / 85.01** | n/a | same | CPU |

Reading the table:

- **Two conventions.** The IU-trained models (Trankit, COMBO IU, Stanza `iu_*`) lose 7 to 9 LAS on ParlaMint at nearly the same UAS (88.7 to 92.8), and the ParlaMint-trained ones lose 7 to 13 LAS on IU: the head structure transfers, the labels do not. We have not analysed which relations differ. Compare LAS columns only with the training treebank in mind; UAS is the fairer cross-treebank number.
- **spaCy `trf` was not trained on a gold treebank** but on the silver Ukr-Synth set (MIT licence; encoder `ukr-models/xlm-roberta-base-uk`) and still scores 91.9 LAS on IU, level with the IU-trained Trankit and COMBO, which suggests the silver labels follow the IU convention. We did not trace how Ukr-Synth was labelled, so the model is neither build-cleared nor independent of IU.
- **Yasuoka's `goeswith` models** (2025; ModernBERT large and base, RoBERTa large WECHSEL) were trained on all Ukrainian UD treebanks together (the repository's `maker.py` clones every `UD_Ukrainian-*`), so they sit between the two conventions. They parse raw text and split words themselves: with space-joined gold tokens, 4 to 22 sentences per run (up to 3%) came back with a different token count, or crashed in the repository's `ud.py`, and are scored as wrong, which lowers their figures by up to about 2 LAS (the `mismatch` field of `results.jsonl`). They are slow (3 to 6 sent/s: one masked forward pass per subword) and run remote code (`trust_remote_code`). Their cards say CC BY 4.0 or MIT although the training data includes the NC-SA IU: treat them as NC-SA.
- **Stanza without the character language model** loses 2.3 LAS on IU (85.4 against 87.7) and 1.2 on ParlaMint. Stanza peaks at 6 to 9 GB because of its allocator cache.
- **Agreement** (UAS between parsers, IU dev, gold tokens): COMBO IU with Trankit 95.7, Trankit with spaCy `trf` 95.2, COMBO IU with spaCy `trf` 94.2, Stanza `iu_charlm` with the others 91.3 to 91.7, `modernbert-large` with COMBO IU or Trankit 91.6. All five of {COMBO IU, Trankit, Stanza `iu_charlm`, spaCy `trf`, `modernbert-large`} are right on 83.3% of words and at least one is right on 98.2%. ParlaMint dev, with COMBO ParlaMint and Stanza `parlamint_charlm` in place of the first and third: COMBO ParlaMint with Trankit 93.6, with Stanza `parlamint_charlm` 92.4; all right 81.7%, at least one right 98.5%. Models that share a training treebank agree because of the shared data and convention, not because they are independent.
- **Mova's own parser** is the `en` tree parser trained in memory on each Ukrainian train set after `en xpos-fill` (nothing is saved; the IU run is a test-only measurement and no model of it is kept). It is about 19 LAS behind the best neural parser on IU with its own tags (14 with gold tags) and 12 behind on ParlaMint (10 with gold tags); it has no Ukrainian lexicon or morphology.
- **Not measured:** raw-text parsing; UDPipe 2 (`uk_all` of the 2.15 release is extracted, CC BY-NC-SA 4.0, but its prediction mode reads word embeddings from a separate embedding service on TensorFlow, which is not cheap to set up); Stanza `ner languk`; `lang-uk/flair-uk-pos` (POS only, downloaded); Yasuoka's `embeds` variants and its BERT-large and RoBERTa-base models.

Models, revisions, licences, training data and checksums are in `uk/manifest.tsv` in the parsers directory of the data tree (outside Git; Stanza and Trankit are under `stanza/resources/uk` and `trankit/cache`, COMBO under `combo/hf`, the Yasuoka models under `uk/hf`). The Hugging Face search for 2023 to 2026 Ukrainian UD parsers found the COMBO models (October 2026) and Yasuoka's families (2025) used above, and only encoders otherwise (`lang-uk/electra-base-ukrainian-*`, `Goader/modern-liberta-*`), which are not parsers.

**Recommended Ukrainian witness set for the Eel's agreement step.** GPU trio in the IU convention: Trankit `ukrainian` (92.0 LAS, 306 sent/s, 3 GB), COMBO IU (92.0, 15 sent/s, 3.5 GB) and spaCy `uk_core_news_trf` (91.9, 134 sent/s, 2 GB; the only one with an MIT licence on the weights); their mutual UAS is 94 to 95, but they share the IU convention. For a second convention add COMBO ParlaMint (91.0 on ParlaMint; UAS 90.9 on IU) or the faster Stanza `parlamint_charlm` (88.5), and, when a slow vote is acceptable, Yasuoka `modernbert-large` (another architecture, trained on both treebanks). The fast cheap pair is Trankit plus spaCy `trf`. All are tests-only: the IU-trained models and Yasuoka's carry the NC-SA data, and the silver data of spaCy is unchecked for build use. Agreement with a witness that shares the training treebank is not independent accuracy evidence.

Sources: Stanza (qi-2020), Trankit (nguyen-2021), spaCy (honnibal-2020), COMBO (klimaszewski-2021), UDPipe 2 (straka-2018), Universal Dependencies v2 (nivre-2020); the Ukrainian treebanks, Ukr-Synth and Yasuoka's models are listed in `docs/usage.tsv`.

## Teacher and dispute interfaces

The annotation teacher is Codex CLI through [`mova-teacher`](https://md.mova.elamur.ai/free/en/docs/teacher/) ([training](https://md.mova.elamur.ai/free/en/docs/repo-docs/training/)). `treeagree` also has a **dispute mode** (`dispute-ask`, `dispute-apply`): only the words whose head the parsers dispute are sent, as short multiple-choice questions. The teacher uses the full mode (a little world context, the sentence, and the SLM's tree and pragmatics); dispute mode stays available. Neither interface establishes annotation quality or authorizes a teacher run.

- [← previous Add-on modules](https://md.mova.elamur.ai/free/en/docs/repo-docs/modules/)
- [next → Pragmatics: frames — who did what to whom, why, and what came next](https://md.mova.elamur.ai/free/en/docs/repo-docs/pragmatics/)
- [↑ back to Documentation](https://md.mova.elamur.ai/free/en/docs/repo-docs/)