Summary

UD_French-FIB (French Interrogative Bank) is an extraction of sentences exhibiting an interrogative from French UD corpora.

Introduction

UD_French-FIB (French Interrogative Bank) is an extraction of sentences exhibiting an interrogative from French UD corpora:

• UD_French-FQB (written)
  • Contributors: Seddah, Djamé; Candito, Marie; Guillaume, Bruno
• UD_French-GSD (written)
  • Contributors: de Marneffe, Marie-Catherine; Guillaume, Bruno; McDonald, Ryan; Suhr, Alane; Nivre, Joakim; Grioni, Matias; Dickerson, Carly; Perrier, Guy
• UD_French_ParisSotires (spoken)
  • Contributors: Gerdes, Kim; Kahane, Sylvain; Mahamdi, Menel
• UD_French-ParTUT (written)
  • Contributors: Bosco, Cristina; Sanguinetti, Manuela
• UD_French-PUD (written)
  • Contributors: Uszkoreit, Hans; Macketanz, Vivien; Burchardt, Aljoscha; Harris, Kim; Marheinecke, Katrin; Petrov, Slav; Kayadelen, Tolga; Attia, Mohammed; Elkahky, Ali; Yu, Zhuoran; Pitler, Emily; Lertpradit, Saran; Strnadová, Jana; Caron, Gauthier; Popel, Martin; Zeman, Daniel; de Marneffe, Marie-Catherine; Guillaume, Bruno
• UD_French-Rhaposide (spoken)
  • Contributors: Gerdes, Kim; Kahane, Sylvain; Nakhlé, Mariam; Yan, Chunxiao; Etienne, Aline; Courtin, Marine
• UD_French-Sequoia (written)
  • Contributors: Candito, Marie; Seddah, Djamé; Perrier, Guy; Guillaume, Bruno

The identification of interrogatives clauses and the extraction were performed using the FUDIA programme. 21 sentences were removed by hand because they were badly annotated, either due to an original annotation mistake, or due to limitations of FUDIA's heuristics.

Origin

Here is the number of FIB sentences by corpus origin.

Corpus	Number of sentences	Percentage in FIB	Percentage in original corpus
FQB	2230	75.0	97.4
GSD	322	10.8	2.0
ParisStories	131	4.4	4.7
Rhapsodie	130	4.4	4.1
Sequoia	96	3.2	3.1
ParTUT	39	1.3	3.8
PUD	24	0.8	2.4
Total	2972	100	10.0

Changes wrt. original UD corpora

Some changes wrt. the original UD corpora have been made during the identification process. For more information about these changes, please check out the FUDIA documentation.

Added features :

• some missing PronType="Rel"
• some missing PronType="Int"
• ClauseType="Int" on interrogative clause heads
• PhraseType="Int" on interrogative phrase heads

Changed annotations

• tel quel changed to fixed (with ExtPos="ADJ") when it is not already the case
• n'importe + WH changed to fixed: with ExtPos="DET" if WH[lemma="quel"], ExtPos="PRON" if WH[lemma="quoi"|"qui"|"lequel"] and ExtPos="ADV" if WH[lemma="comment"|"quand"|"où"], and with consequent head shift
• est-ce que changed to fixed with ExtPos="SCONJ", the head shifted to the interrogative clause head CL_HEAD, and the main relation changed to mark(CL_HEAD,est)
• qu'est-ce que changed to fixed with ExtPos="PRON", the head shifted to the interrogative clause head CL_HEAD, and the relations changed to obj, obl or nsubj from CL_HEAD to qu', depending on the context
• WH + que + S' with que attached to WH, changed to govern que by the head of S'

Splitting

The whole corpus contains 38,884 tokens in 2,972 sentences.

In UD_French-FIB, data are split into:

• fr_fib-ud-test.conllu: 2,386 tokens in 115 sentences
  • using GSD test, ParisStories test, ParTUT test, Rhaposodie test and Sequoia test
• fr_fib-ud-dev.conllu: 2,365 tokens in 130 sentences
  • using GSD dev, ParisStories dev, ParTUT dev, Rhaposodie dev and Sequoia dev
• fr_fib-ud-train.conllu: 34,133 tokens in 2,727 sentences
  • using FQB test, GSD train, ParisStories train, ParTUT train, PUD test, Rhaposodie train and Sequoia train

Enriched versions

For each set, an enriched version is provided. The enriched version contains the same sentences but with additional edges and features:

• edges cue:wh from the head of the interrogative clause to the interrogative proform(s) associated
• edges cue:wh from the head of the interrogative phrase to the interrogative proform(s) associated
• edged cue:mark from the head of the interrogative clause to the interrogative marker
• feature Quoted="Yes" for quoted segments

For more information, see the FUDIA documentation.

Genres

The corpus contains written sentences and trasncriptions of spoken sentences. More about the different genres is to be found in the documentation of the original corpora.

References

Valentin D. Richard. Est-ce que l'extraction des interrogatives du français peut-elle être automatisée ?. 5èmes journées du Groupement de Recherche CNRS « Linguistique Informatique, Formelle et de Terrain » (LIFT 2023), Nov 2023, Nancy, France. pp.69-76. ⟨hal-04359947⟩

Changelog

Licenses

Caution, due to the originical corpora, the parts of this corpus have different licenses:

• CC BY-NC-SA 4.0 for ParTUT
• CC BY-SA 4.0 for Rhapsodie, ParisStories, PUD, GSD
• LGPL-LR (see LGPL-LR_LICENSE.txt) for Sequoia, FQB

The original corpus is prepended to all sentence id.

=== Machine-readable metadata (DO NOT REMOVE!) ================================ Data available since: UD v2.11 License: multiple Includes text: yes Genre: wiki medical news nonfiction legal blog reviews spoken Lemmas: converted from manual UPOS: converted from manual XPOS: not available Features: converted from manual Relations: converted from manual Contributors: Richard, Valentin D. Contributing: elsewhere