Version classiqueVersion mobile

Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

 | 
Roberto Basili
, 
Malvina Nissim
, 
Giorgio Satta

Toward a bilingual lexical database on connectives: Exploiting a German/Italian parallel corpus

Peter Bourgonje, Yulia Grishina et Manfred Stede

Résumé

We report on experiments to validate and extend two language-specific connective databases (German and Italian) using a word-aligned corpus. This is a first step toward constructing a bilingual lexicon on connectives that are connected via their discourse senses.

Texte intégral

1 Introduction

1An important part of discourse processing deals with uncovering coherence relations that hold between individual, “elementary” units of a text. The lexical items that can signal such a relation are referred to as discourse connectives, and examples of these relations, also called the connectives’ senses, are contrast (e.g., “but”), elaboration (e.g., “in particular”), or cause (e.g., “therefore”). Notice, however, that relations need not always be signalled in text, if the context or world knowledge is sufficient for the reader to infer it, as (1)-(4) demonstrate:

(1) We should hurry, because it’s late.
(2) We should hurry. It’s late.
(3) The red pen costs $2, while the blue one is $2.50.
(4) The red pen costs $2; the blue one is $2.50.

2On the other hand, example (6) is a perfectly grammatical sentence but the meaning is different from (5), so for this case of a Concession relation, the connective is in fact indispensable.

(5) Although it is late, we don’t need to hurry.
(6) It is late; we don’t need to hurry.

3Recognizing these relations, which can hold within a sentence, between two sentences, or between larger spans of text, is a central task for uncovering the structure of a text, as it has been studied in theories like Rhetorical Structure Theory (Mann and Thompson, 1988) or Segmented Discourse Representation Theory (Asher and Lascarides, 2003). While the usage of connectives can sometimes be optional, the set of connectives that a language offers is generally taken as important (if not exhaustive) evidence for the set of coherence relations that should be assumed.

1.1 Background: Connectives

4From a syntactic viewpoint, “connective” is not a homogeneous class, as it contains conjunctions, different kinds of adverbials, as well as certain prepositions. Our underlying definition of discourse connectives is based on (Pasch et al., 2003, p. 331):

(7) Def.: A discourse connective is a lexical item x that exhibits each of the following five properties:
(M1) x cannot be inflected.
(M2) x does not assign case features to its syntactic environment.
(M3) The meaning of x is a two-place relation.
(M4) The arguments of the relation (the meaning of x) are propositional structures.
(M5) The expressions of the arguments of the relation can be sentential structures.

5Following (Stede, 2002), we drop M2 because our lexicon deliberately includes several prepositions that can be used as connectives (in the sense of M1, M3-M5), e.g., trotz (‘despite’) or wegen (‘due to’).

1.2 Motivation and contribution

6Connectives can pose interesting challenges to translation and for language learners, as the differences in meaning between similar connectives can be quite subtle. For these reasons, we are interested here specifically in a bilingual Italian– German lexical resource, to be built on top of two existing single-language lexicons. As a case study, we focus on the subgroup of contrastive/concessive connectives, which we determined to comprise (in the existing lexicons) 31 German connectives and 12 Italian connectives; see Tables 3.2.2 and 3.2.2.

7The main contributions of this paper are (1) suggestions for improving the existing languagespecific resources used in this study through the technique of cross-lingual projection in a parallel corpus, which reveals correspondences between connectives and can point to gaps in either of the resources; and (2) an overview of the distribution of connectives and their senses, to be used in a bilingual database. Section 2 explains the two monolingual lexicons we work with, and Section 3 describes the corpus. Section 4 reviews related work in this area. Section 5 elaborates the idea of bilingual connective databases, and Section 6 summarises our findings.

2 Lexicons: DiMLex and LICo

8We extracted the German contrastive connectives from DiMLex (Scheffler and Stede, 2016), a connective lexicon with several different fields describing orthographical variants, syntactic type, discourse sense, and usage examples. It contains 275 entries. The sense annotations are based on the Penn Discourse Treebank (PDTB) senses (Miltsakaki et al., 2008) in its latest version 3. The lexicon is publicly available 1and aims to exhaustively describe the set of connectives for German, thus providing a basis for our case study.

9The set of Italian contrastive connectives comes from LICo (Feltracco et al., 2016), a similar lexicon for Italian containing 170 entries.2 LICo was inspired by DiMLex and contains annotations on the same attributes and uses essentially the same structure (i.e., the same PDTB senses, orthographic variants, usage examples, etc.). An example entry of LICo is shown in Figure 1. We refer the reader to Feltracco et al. (2016) for details.

Figure 1: al contrario entry in LICo

Figure 1: al contrario entry in LICo

3 Exploiting a parallel corpus

10For the parallel German/Italian corpus we used Europarl (Koehn, 2005), as it still appears to be the biggest resource of this kind, and it is, conveniently, already sentence-aligned. From the 1,832,053 sentences in the German-Italian part of the corpus we extracted the word alignments using MGIZA++ (Gao and Vogel, 2008). In the following, we sketch our method for obtaining the correspondence information on connectives based on these word alignments, and then present the results.

3.1 Method: Iterative lookup

11We approach the problem from two sides: First we look up every German connective (31 in total) to get Italian alignments. 30 of them appeared in our Europarl corpus (with dementgegen missing). Then we look up every Italian connective to get German alignments (all 12 connectives present in the corpus). We end up with a list of target language words or phrases (or empty elements, since a source language connective can also be covert in the target language) that are candidate contrastive connectives. Note that the lookup procedure does not differ structurally between words and phrases. In both cases, single words (stand-alone or in a phrase) can correspond to zero, one or more target words. The target representation is collected in a key-value structure, where the key is the position in the sentence and the value the word. This list is then sorted by position to return the target word or phrase (which is potentially discontinuous). Because the word alignment is not guaranteed to be correct, to filter for unlikely translations we focus on only the 3 most frequent alignments for every connective. We expect to find at least a subset of the already known (contrastive) connectives (from DiMLex or LICo), potentially complemented by a set of words or phrases that can help filling gaps in either of the lexicons.

  • 3 Ideally going back and forth until a stable and exhaustive set of candidates is found. For this stu (...)

12This procedure produces at least some incorrect results for the following two reasons: 1) discourse connectives often can appear in a text with a connective reading or with a non-connective reading; and 2) connectives can have multiple senses, so that a connective may not have the contrastive reading in the particular sentence. The candidates produced hence have to be evaluated manually. Resulting candidates that have a connective reading are added to the seed list, in order to repeat the step back from the target language to the source language3.

3.2 Results

3.2.1 German–Italian

13The results of the first step of the iteration using the 31 German seed connectives are displayed in Table 3.2.2, where an underscore indicates an empty string (meaning that the connective was not aligned to a particular word or phrase in the target language) and the number after the underscore represents the (normalised) frequency of the alignment.

14For the evaluation, we asked a native speaker of Italian with expert knowledge in linguistics to validate the resulting top 3 bilingual mappings. Firstly, we identified several possible connective candidates that were aligned to German contrastive connectives, but were not present in LICo, such as al contempo, solo che, doppo tutto. Secondly, we observed several possible orthographic variants of the already existing Italian connectives: contro or contrario (as possible variants of al contrario), and d’altro canto (as a variant of a discontinious connective da un canto...dall’altro). Finally, we found that several Italian connectives only had the concession sense, while the corresponding German connectives also had the Contrast sense, such as comunque, for which we found the German alignments aber, allerdings and doch, for example.

Figure 2: Most frequent alignments of jedoch

Figure 2: Most frequent alignments of jedoch

15As an example of a visualisation (for a single connective) the above analysis is based on, consider Figure 2, showing the most frequent alignments of jedoch, which always has a connective reading, thus nullifying the first problem mentioned in 3.1.

3.2.2 Italian–German

16The results of the first step of the iteration using the 12 Italian seed connectives are displayed in Table 3.2.2. For 11 of the 12 contrastive connectives from LICo, the top 3 alignments yielded an existing DiMLex entry. The only connective without a DiMLex entry in the top 3 was al contrario, for which a possible new German connective candidate im Gegenteil was found through alignment.

17Upon further investigation of the lower-ranked alignments (not included in Table 3.2.2), we were able to identify several other gaps in the German lexicon. Firstly, we observed that the Italian connective invece is frequently aligned to the German word anstelle, which is not in DiMLex (but anstelle dessen is). After examining the corresponding examples, we conclude that anstelle should be added to DimLex as a separate entry (similarly to the already existing aufgrund vs. aufgrund dessen). Also, we found that DiMLex lacks italic>statt dessen as an orthographic variant of the more canonical stattdessen.

Figure 3: Most frequent alignments of invece

Figure 3: Most frequent alignments of invece

Figure 4: Mapping of connective senses from Italian to German

Figure 4: Mapping of connective senses from Italian to German

18Finally, we identified two interesting cases that are DiMLex candidates: umgekehrt and (ganz) im Gegenteil, which we found aligned to the Italian viceversa and al contrario, respectively, but more corpus evidence is required to decide whether they can indeed serve as connective in the German language.

19As an example visualisation, consider Figure 3, showing the most frequent alignments of invece, which always has a connective reading.

  • 4 Not listed here for reasons of space.

20For Italian–German, we repeated the steps above with the candidates found using the German seed list (projecting the resulting Italian list back to German) to see if any additional connectives or orthographic variants would be found. We again found im Gegenteil through alignment of al contrario and a few alternative lexicalisations for DiMLex connectives4, but no new candidates.

Table 1: German connectives and their Italian alignments

German connective (frequency)

Top 3 Italian alignments

aber(105413)

ma//_(0.24)//tuttavia

alldieweil (3)

finche//perche

allein (6973)

_(0.30)//solo//soltanto

allerdings (16232)

tuttavia//_(0.22)//ma

andererseits (6354)

_(0.30)//dall’ altro//d’ altro canto

bloB dass (117)

_(0.10)//solo che//che solo

dafiir (36895)

_(0.70)//per//per aver

dafur // dass (42)

che//_(0.19)//per

dagegen (5423)

_(0.34)//contro//contrario

dahingegen (24)

_(0.17)//invece//al contrario

dementgegen (0)

demgegenuber (121)

_(0.25)//invece//contro

doch(37423)

_(0.47)//ma//tuttavia

einerseits (4221)

da un lato//_(0.31)//da una parte

freilich (159)

_(0.30)//naturalmente//certo

gleichzeitig (13293)

_(0.35)//al contempo//allo stesso tempo

hingegen (1909)

invece//_(0.26)//tuttavia

immerhin (1360)

_(0.44)//comunque//dopo tutto

indessen (280)

invece//_(0.19)//tuttavia

jedoch (47667)

tuttavia//_(0.27)//ma

nur dass (21617)

che//solo che

sosehr (14)

malgrado tutto

unterdessen (193)

nel frattempo//_(0.21)//intanto

wiederum (2450)

_(0.55)//a sua volta//ancora una volta

wogegen (111)

mentre//_(0.19)//contro cosa

wohingegen (218)

mentre//_(0.14)//ma

wahrend (20388)

_(0.28)//mentre//durante

wahrenddessen (78)

nel frattempo//_(0.17)//mentre

zugleich (3576)

_(0.41)//al contempo//allo stesso tempo

zum anderen (4299)

_(0.09)//altri//altre

zum einen (8848)

un//_(0.10)//una

Table 2: Italian connectives and their German alignments

Italian connective (frequency)

Top 3 German alignments

al contrario (3641)

im gegenteil//_(0.10)//im gegenteil

bensi (7107)

sondern//_(0.12)//sondern vielmehr

contrariamente a (661)

_(0.08)//entgegen//im gegensatz zu

da un canto (352)

einerseits//_(0.11)//andererseits

da un lato (4612)

einerseits//_(0.08)//einerseits die

da una parte (10194)

_(0.07)//und//eine

invece(18778)

_(0.48)//anstatt//stattdessen

ma (135218)

aber//sondern//_(0.15)

mentre (15773)

wahrend//_(0.19)//und

per contro (13468)

gegen//und//_(0.06)

pero (22687)

aber//jedoch//_(0.24)

viceversa (522)

umgekehrt//_(0.19)//hingegen

4 Related work

21Parallel corpora have been successfully exploited before in order to automatically generate or induce connective lexicons in different languages. In particular, Versley (2010) projected discourse connectives across an English–German parallel corpus to train a discourse parser capable of disambiguating connective and non-connective readings. Similarly, Zhou et al. (2012) used an English–Chinese parallel corpus in order to build a Chinese connective lexicon via cross-lingual projection, and Hajlaoui and Popescu-Belis (2013) relied on parallel data to automatically retrieve Arabic counterparts for a subset of English connectives.

22Since our goal was not to build a connective lexicon from scratch, but to extend the connective lists and refine the inventory of senses for the already existing lexicons, the closest approach to ours is the one adopted by Laali and Kosseim (2014), who aimed at automatically inducing a French connective lexicon via English–French parallel corpora using additional filtering rules. Similar to ours, their results have shown that using parallel translations can improve the coverage of the connective lists in both languages; however, since their lexicons used different sets of discourse relations, they were not able to extend their connective database in respect to senses, as opposed to our work.

5 Toward a bilingual connective database

23Our study is meant as a step toward moving from single-language connective lexicons to a bilingual one that provides information about the relationships between the language-specific entries. Both monolingual lexicons are already publicly available on GitHub and in addition an interface allowing bilingual search has been made public in a related project5. Below we sketch additional plans for providing this information on the levels of connective tokens, and senses (coherence relations).

5.1 Connective mappings

24One central purpose of a bilingual database is to assist translators (human or machine) or (human) language learners. For most connectives, there is a complicated m:n mapping between languages, which standard dictionaries do not cover, and the relevant features for making choices are not systematically known yet. A corpus-based inventory of mappings – ideally supplemented by pointers to the corpus instances and their context – can be a very useful resource for undertaking contrastive lexical investigations.

5.2 From connectives to phrases

25The PDTB (Prasad et al., 2008) makes a distinction between connectives (a closed set) and “alternative lexicalizations” (AltLex), which are a non-demarcated set of phrases used to express a coherence relation. Such phrases are so far not part of DiMLex nor LICo. Obviously, they are much harder to detect: Corpus annotation (as done in PDTB) is one way, and we regard our cross-lingual projection method as another promising way. Quite often, connectives in language A have been translated to an AltLex in language B. We plan to study this more systematically by a closer inspection of the alignments and their contexts, in order to extract AltLex candidates as a supplement to the connective lexicons.

5.3 Senses and their distributions

26A bilingual connective database can shed light on the distribution of senses over different languages and the degree of ambiguity that individual connectives exhibit. While we consider such conclusions premature for the current stage of the language-specific resources, we include Figure 4, which shows groups of connectives that share the same sense (or group of senses for ambiguous connectives) and their alignment to similar groups on the target side. The 12 Italian connectives (on the left), when grouped together based on their sense(s), form 4 sets, whereas for German (right side), fewer connectives (11 that were found in DiMLex among the top 3 alignments of the 12 source connectives) group into more sets (10). This suggests more ambiguity in Italian connectives, with less different senses represented by a larger set of connectives.

27In addition, we observed that Italian connectives with a sense Contrast or Concession are frequently aligned to their German counterparts with a sense Substitution, such as anstelle-invece. Having examined the parallel examples more closely, we conclude that assigning both senses would be valid for both German and Italian, although they are placed distantly in the PDTB hierarchy of senses. These findings are confirmed by Feltracco et al. (2016), who acknowledge that the distinction between the two senses was one of the main cases of the inter-annotator disagreement. We conclude that both lexicons could benefit from adding additional senses gained via comparing parallel translations.

6 Summary

28We present, to the best of our knowledge, the first Italian–German investigation of discourse connective lexicons. For the subclass of Contrast (in a wide sense), we were able to identify several missing entries in both lexicons, and provided a start on identifying AltLex items for the two languages (future work). Once the information is organized in a complete bilingual database, it can assist translation and conclusions can be drawn regarding connective distribution, sense distribution and ambiguity in the different languages.

29As prominent steps for future work, we note the disambiguation of connective- and non-connective readings, the implementation of more sophisticated filtering strategies to retrieve more reliable connective candidates and repeating this study for different languages pairs.

Acknowledgments

30We are grateful to the Deutsche Forschungsgemeinschaft (DFG) for funding this work in the project “Anaphoricity in Connectives”. We would like to thank the anonymous reviewers for comments on an earlier version of this manuscript. We also thank Flavia Adani for her help with translation and interpretation of the Italian results, and Tatjana Scheffler for the recent work on DiMLex.

Bibliographie

Nicholas Asher and Alex Lascarides. 2003. Logics of Conversation. Cambridge University Press, Cambridge.

Anna Feltracco, Elisabetta Jezek, Bernardo Magnini, and Manfred Stede. 2016. Lico: A lexicon of italian connectives. In Proceedings of the 3rd Italian Conference on Computational Linguistics (CLiC-it), Napoli, Italy.

Qin Gao and Stephan Vogel. 2008. Parallel implementations of word alignment tool. In Software Engineering, Testing, and Quality Assurance for Natural Language Processing, SETQA-NLP ‘08, pages 49– 57, Stroudsburg, PA, USA. Association for Computational Linguistics.

Najeh Hajlaoui and Andrei Popescu-Belis. 2013. Assessing the accuracy of discourse connective translations: Validation of an automatic metric. In University of the Aegean-14th International Conference on Intelligent Text Processing and Computational Linguistics. Springer.

Philipp Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In Conference Proceedings: the tenth Machine Translation Summit, pages 79–86, Phuket, Thailand. AAMT, AAMT.

Majid Laali and Leila Kosseim. 2014. Inducing discourse connectives from parallel texts. In COLING, pages 610–619.

William Mann and Sandra Thompson. 1988. Rhetorical structure theory: Towards a functional theory of text organization. TEXT, 8:243–281.

Eleni Miltsakaki, Livio Robaldo, Alan Lee, and Aravind Joshi, 2008. Sense annotation in the Penn Discourse Treebank, pages 275–286. Springer Berlin Heidelberg, Berlin, Heidelberg.

Renate Pasch, Ursula Brauße, Eva Breindl, and UlrichĤerrmann Waßner. 2003. Handbuch der deutschen Konnektoren. Walter de Gruyter, Berlin/New York.

Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Miltsakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The penn discourse treebank 2.0. In In Proceedings of LREC.

Tatjana Scheffler and Manfred Stede. 2016. Adding semantic relations to a large-coverage connective lexicon of German. In Nicoletta Calzolari et al., editor, Proc. of the Ninth International Conference on Language Resources and Evaluation (LREC 2016), Portorož, Slovenia, May.

Manfred Stede. 2002. Dimlex: A lexical approach to discourse markers. In Exploring the Lexicon - Theory and Computation. Edizioni dell Orso, Alessandria.

Yannick Versley. 2010. Discovery of ambiguous and unambiguous discourse connectives via annotation projection. In Proceedings of Workshop on Annotation and Exploitation of Parallel Corpora (AEPC). Northern European Association for Language Technology (NEALT).

Lanjun Zhou, Wei Gao, Bin Li, Zhong Wei, and KamFai Wong. 2012. Cross-lingual identification of ambiguous discourse connectives for resource-poor language. In Proceedings of COLING.

Notes

1 https://github.com/discourse-lab/dimlex

2 https://hlt-nlp.fbk.eu/technologies/lico

3 Ideally going back and forth until a stable and exhaustive set of candidates is found. For this study, we only did the first step, and then projected the found Italian connectives back to German.

4 Not listed here for reasons of space.

5 http://connective-lex.info/

Table des illustrations

Titre Figure 1: al contrario entry in LICo
URL http://books.openedition.org/aaccademia/docannexe/image/2360/img-1.jpg
Fichier image/jpeg, 232k
Titre Figure 2: Most frequent alignments of jedoch
URL http://books.openedition.org/aaccademia/docannexe/image/2360/img-2.jpg
Fichier image/jpeg, 76k
Titre Figure 3: Most frequent alignments of invece
URL http://books.openedition.org/aaccademia/docannexe/image/2360/img-3.jpg
Fichier image/jpeg, 80k
Titre Figure 4: Mapping of connective senses from Italian to German
URL http://books.openedition.org/aaccademia/docannexe/image/2360/img-4.jpg
Fichier image/jpeg, 232k

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Lire

Open access

Acheter

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search