{
"paper_id": "2020",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T04:33:39.944976Z"
},
"title": "Polish Lexicon-Grammar Development Methodology as an Example for Application to other Languages",
"authors": [
{
"first": "Zygmunt",
"middle": [],
"last": "Vetulani",
"suffix": "",
"affiliation": {},
"email": "vetulani@amu.edu.pl"
},
{
"first": "Gra\u017cyna",
"middle": [],
"last": "Vetulani",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Adam Mickiewicz University",
"location": {
"settlement": "Pozna\u0144"
}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "In the paper we present our methodology with the intention to propose it as a reference for creating lexicon-grammars. We share our long-term experience gained during research projects (past and ongoing) concerning the description of Polish using this approach. The above-mentioned methodology, linking semantics and syntax, has revealed useful for various IT applications. Among other, we address this paper to researchers working on \"less\" or \"middle-resourced\" Indo-European languages as a proposal of a long term academic cooperation in the field. We believe that the confrontation of our lexicon-grammar methodology with other languages-Indo-European, but also Non-Indo-European languages of India, Ugro-Finish or Turkic languages in Eurasiawill allow for better understanding of the level of versatility of our approach and, last but not least, will create opportunities to intensify comparative studies. The reason of presenting some our works on language resources within the Wildre workshop is the intention not only to take up the challenge thrown down in the CFP of this workshop which is: \"To provide opportunity for researchers from India to collaborate with researchers from other parts of the world\", but also to generalize this challenge to other languages.",
"pdf_parse": {
"paper_id": "2020",
"_pdf_hash": "",
"abstract": [
{
"text": "In the paper we present our methodology with the intention to propose it as a reference for creating lexicon-grammars. We share our long-term experience gained during research projects (past and ongoing) concerning the description of Polish using this approach. The above-mentioned methodology, linking semantics and syntax, has revealed useful for various IT applications. Among other, we address this paper to researchers working on \"less\" or \"middle-resourced\" Indo-European languages as a proposal of a long term academic cooperation in the field. We believe that the confrontation of our lexicon-grammar methodology with other languages-Indo-European, but also Non-Indo-European languages of India, Ugro-Finish or Turkic languages in Eurasiawill allow for better understanding of the level of versatility of our approach and, last but not least, will create opportunities to intensify comparative studies. The reason of presenting some our works on language resources within the Wildre workshop is the intention not only to take up the challenge thrown down in the CFP of this workshop which is: \"To provide opportunity for researchers from India to collaborate with researchers from other parts of the world\", but also to generalize this challenge to other languages.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "In the linguistic tradition a crucial role in language description was typically given to dictionaries and grammars. The oldest preserved dictionaries were in form of cuneiform tablets with Sumerian-Akkadian word-pairs and are dated 2300 BC. Grammars are \"younger\". Among the first were grammars for Sanskrit attributed to Yaska (6th century BC) and P\u0101\u1e47ini (6-5th century BC). In Europe the oldest known grammars and dictionaries date from the Hellenic period. The first one was Art of Grammar by Dyonisus Thrax (170-90 BCE), in use in Greek schools still some 1,500 years later. Until recently, these tools were used for the same purposes as before -teaching and translation, and ipso facto were supposed to be interpreted by humans. The formal rigor was considered of secondary importance. The situation changed recently with development of computer-based information technologies. For machine language processing (as machine translation, text and speech understanding, etc.) it appeared crucial to adapt language description methodology to the technologyimposed needs of precision. Being human-readable was not enough, new technological age required from grammars and dictionaries to become machine-readable. New concepts of organization of language description for better facing technological challenges emerged. One among them was the concept of lexicon-grammar. This paper addresses two cases. Firstlanguages with a rich linguistic tradition and valuable preexisting language resources, for which the methods described in this paper will be easily applicable and may bring interesting results. 1 We do not believe that basic linguistic research is avoidable on the base of technological solutions only. (See the historical statement addressed by Euclid of Alexandria Among Indian languages this will be the case of Sanskrit, Hindi and many other. On the other hand, a multitude of languages in use on the Indian subcontinent do not dispose of such a privileged starting position. In this case, in order to benefit from the methodology we describe in this paper, an effort must first be done to complete existing gaps. This is a hard work, and the paper, we hope will give some idea on the priorities on this way. Still, an important basic research effort will be necessary 1 .",
"cite_spans": [
{
"start": 1600,
"end": 1601,
"text": "1",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1."
},
{
"text": "Development of computational linguistics and resulting language technologies made possible passage from the fundamental research to the development of real-scale applications. At this stage availability of rigorous, exhaustive and easy to implement language models and descriptions appeared necessary. The concept of lexicongrammar answers to these needs. Its main idea is to link an important amount of grammatical (syntactic and semantic) information directly to respective words. Within this approach, it is natural to keep syntactic and semantic information stored as a part of lexicon entries together with other kinds of information (e.g. pragmatic). This principle applies first of all to verbs, but also to other words which \"open\" syntactic positions in a sentence, as e.g. certain nouns, adjectives and adverbs. Within this approach, we include into the lexicon-grammar all predicative words (i.e. words that represent the predicate in the sentence and which open the corresponding argument positions).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Why Lexicon-Grammars?",
"sec_num": "2."
},
{
"text": "(365 BC -270 BC) to Ptolemy I (367 BC -282 BC): \"Sir, there is no royal road to geometry\".)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Why Lexicon-Grammars?",
"sec_num": "2."
},
{
"text": "The idea of lexicon-grammar is to link predicative words with possibly complete grammatical information related to these words. It was first systematically explored by Maurice Gross (Gross 1975 (Gross , 1994 , initially for French, then for other languages. Gross was alsoto the best of our knowledgethe first to use the term lexicon-grammar (fr. lexique-grammaire)).",
"cite_spans": [
{
"start": 182,
"end": 193,
"text": "(Gross 1975",
"ref_id": null
},
{
"start": 194,
"end": 207,
"text": "(Gross , 1994",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Why Lexicon-Grammars?",
"sec_num": "2."
},
{
"text": "The EUREKA GENELEX 2 was a European initiative to realize the idea of lexicon-grammar in form of a generic model for lexicons and to propose software tools for lexicons management (Antoni-Lay et al., 1994) . Anoni-Lay presents two reasons to build large-size lexicons as follows. \"The first reason is that Natural Language applications keep on moving from research environments to the real world of practical applications. Since real world applications invariably require larger linguistic coverage, the number of entries in electronic dictionaries inevitably increases. The second reason lies in the tendency to insert an increasing amount of linguistic information into a lexicon. (\u2026) In the eighties, new attempts were made with an emphasis on grammars, but an engineering problem arose: how to manage a huge set of more or less interdependent rules. The recent tendency is to organize the rules independently, to call them syntactic/semantic properties, and to store this information in the lexicon. A great part of the grammatical knowledge is put in the lexicon (\u2026). This leads to systems with fewer rules and more complex lexicons.\" (ibid.).",
"cite_spans": [
{
"start": 180,
"end": 205,
"text": "(Antoni-Lay et al., 1994)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "The genericity of the GENELEX model is assured by:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "-\"theory welcoming\", what means openness of the GENELEX formalism to various linguistic theories (respecting the principle that its practical application will refer to some, well defined linguistic theories as a basis of the 2 GENELEX was followed by several other EU projects, such as LE-PAROLE (1996 -1998 , LE-SIMPLE (1998 -2000 and GRAAL (1992 GRAAL ( -1996 . 3 The GENELEX creators make a clear distinction between independence with respect to language theory, and the necessity for any particular application to be covered by some language theory compatible with the GENELEX model (this is in order to organize correctly the lexicographer's work). 4 Polish, like all other Slavic languages, Latin and, in some respect, also Germanic languages, has a developed inflection system. Inflectional categories are case and number for nouns, gender, mood, number, person tense, and voice for verbs, case, gender, number and degree for adjectives, degree alone for adverbs, etc. Examples of descriptive categories are gender for nouns and aspect for verbs. The verbal inflection system (called conjugation) is simpler than in most Romance or Germanic languages but still complex enough to precisely situate action or narration on the temporal axis. The second of the two main paradigms (called declension) is the nominal one. It is based on the case and number oppositions. The declension lexicographer's research workshop). It should allow encoding phenomena described in different ways by different theories 3 ; -possibility to generate various application-oriented lexicons;",
"cite_spans": [
{
"start": 286,
"end": 301,
"text": "LE-PAROLE (1996",
"ref_id": null
},
{
"start": 302,
"end": 307,
"text": "-1998",
"ref_id": "BIBREF0"
},
{
"start": 310,
"end": 325,
"text": "LE-SIMPLE (1998",
"ref_id": null
},
{
"start": 326,
"end": 331,
"text": "-2000",
"ref_id": null
},
{
"start": 336,
"end": 347,
"text": "GRAAL (1992",
"ref_id": null
},
{
"start": 348,
"end": 361,
"text": "GRAAL ( -1996",
"ref_id": null
},
{
"start": 364,
"end": 365,
"text": "3",
"ref_id": null
},
{
"start": 654,
"end": 655,
"text": "4",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "-capacity of generation of lexicons apt to serve applications demanding a huge linguistic coverage.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "The second important property of GENELEX besides genericity was the requirement of high precision and clarity of GENELEX-compatible lexicon-grammars.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "GENELEX was first dedicated to a number of West-European languages, among other French, English, German, Italian. Although Polish 4 was not directly addressed by GENELEX, it was covered together with Czech and Hungarian by two EU projects (COPERNICUS projects CEGLEX -COPERNICUS 1032 (1995 -1996 and GRAMLEX -COPERNICUS 621 (1995 -1998 )) 5 whose objective was testing the potential of the extension of the novel GENELEX-based LT solutions to highly inflectional (as Polish) and agglutinative (as Hungarian) languages. Positive results obtained within this project demonstrated potential usefulness of the lexicon-grammar approach for so far less-resourced languages, Indo-European or not. In particular, the case of Polish demonstrated the need to take into account, within the lexicon-grammar approach, the specificity of highly inflected languages, like Lain or Sanskrit, with complex verbal and nominal morphology.",
"cite_spans": [
{
"start": 239,
"end": 283,
"text": "(COPERNICUS projects CEGLEX -COPERNICUS 1032",
"ref_id": null
},
{
"start": 284,
"end": 289,
"text": "(1995",
"ref_id": null
},
{
"start": 290,
"end": 295,
"text": "-1996",
"ref_id": null
},
{
"start": 300,
"end": 329,
"text": "GRAMLEX -COPERNICUS 621 (1995",
"ref_id": null
},
{
"start": 330,
"end": 335,
"text": "-1998",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "GENELEX project (1990-1994)",
"sec_num": "3."
},
{
"text": "Already in our early works on question-understanding-andanswering systems (Vetulani, Z. 1988 (Vetulani, Z. , 1997 we capitalized the advantages of the lexicon-grammar approach. In addition to information typically provided in system of Polish strongly marks Polish syntax; as the declension case endings characterize the function of the word within the sentence, therefore the word order is more free than in, e.g., Romance or Germanic languages where the position of the word in a sentence is meaningful. Main representatives of the Polish declension system are nouns, but also adjectives, numerals, pronouns and participles. Polish inflected forms are created by combining various grammatical morphemes with stems. These morphemes are mainly prefixes and suffixes (endings). Endings are considered as the typical inflection markers and traditional classifications into inflection classes are based on ending configurations. Endings may fulfil various syntactic and semantic functions at the same time. A large variety of inflectional categories for most of parts of speech is the reason why inflection paradigms are complex and long in Polish. For example, the nominal paradigm has 14 positions, the length of the verbal paradigm is 37 and the length of the adjectival one is 84 (Vetulani, G. 2000) . 5 Some of the outcomes of these project are described in (Vetulani, G. 2000 ).",
"cite_spans": [
{
"start": 74,
"end": 92,
"text": "(Vetulani, Z. 1988",
"ref_id": null
},
{
"start": 93,
"end": 113,
"text": "(Vetulani, Z. , 1997",
"ref_id": null
},
{
"start": 1281,
"end": 1300,
"text": "(Vetulani, G. 2000)",
"ref_id": null
},
{
"start": 1303,
"end": 1304,
"text": "5",
"ref_id": null
},
{
"start": 1360,
"end": 1378,
"text": "(Vetulani, G. 2000",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "dictionaries we managed to explore structural, as well as morpho-syntactic-and-semantic information directly stored with predicative words, i.e. words which are surface manifestation of sentence predicates. In Polish, as in many (all?) Indo-European languages, these are typically verbs, but also nouns, adjectives, participles and adverbs. The content of lexicon-grammar entries informs about the structure of minimal complete elementary sentences supported by the predictive words, both simple and compound. This information may be precious in order to substantially speed-up sentence processing 6 (see e.g. Vetulani, Z. 1997) . Taking this into account, the text processing stage requires a new kind of language resource which is electronic lexicon-grammar. In opposition to small text processing demo systems developed so far, this requirement appears demanding when starting to build real size applications within the concept of predicate-argument approach to syntax of elementary sentences that we applied in our rule-based text analyzers and generators. The rulebased approach dominating still at the turn of the centuries remains important in all cases where high processing precision is essential.",
"cite_spans": [
{
"start": 610,
"end": 628,
"text": "Vetulani, Z. 1997)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "Concerning digital language resources Polish was clearly under-resourced at those days, however with a good starting position due to well-developed traditional language descriptions. For example, since 1990s the high quality lexicon-grammar in the form of Generative Syntactic Dictionary of Polish Verbs (Pola\u0144ski 1980 (Pola\u0144ski -1982 was to our disposal. This impressive resource of 7,000 most widely used Polish simple verbs, being addressed first of all to human users, was hardly computer-readable. As simplified example of an entry we propose the description of the polysemic predicative verb POLECIE\u0106 (meaning to fly). One of its meanings is represented by the following entry (lines ad):",
"cite_spans": [
{
"start": 304,
"end": 318,
"text": "(Pola\u0144ski 1980",
"ref_id": null
},
{
"start": 319,
"end": 334,
"text": "(Pola\u0144ski -1982",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "(a) POLECIE\u0106 (English: FLY) 7 (b) NPNominative+NPI+(NPAblative)+(NPAdlative) (c) NPNominative [human]; NPInstrumental [flying object]; NPAblative [location]; NPAdlative [location] (d) Examples: ..., Ja(NPN) z Warszawy (NPAbl) do Francji (NPAdl) POLEC\u0118 samolotem (NPI),\u2026 \u2026, I (NPN) WILL FLY from Warsaw(NPAbl) to France(NPAdl) by plane(NPI)),... , where:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "6 E.g. in heuristic parsing in order to limit the grammar search space explored by the parser (Vetulani, Z. 1997 ). 7 \"We do not claim that the set of semantic features we propose is exhaustive and final. Besides features commonly accepted we considered necessary to introduce such distinction words as nouns designing plants, elements, information etc.\", cf. (Pola\u0144ski 1992). 8 \"We do not claim that the set of semantic features we propose is exhaustive and final. Besides features commonly accepted we considered necessary to introduce such distinction words as nouns designing plants, elements, information etc.\", cf. (Pola\u0144ski 1992).",
"cite_spans": [
{
"start": 94,
"end": 112,
"text": "(Vetulani, Z. 1997",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "(a) is the entry identifier (verb in infinitive) (b) is the sentential scheme showing the syntactic structure and syntactic requirements of the verb with respect to obligatory and facultative (in brackets) arguments (it may be considered as a simple sentence pattern) (c) is the specification of semantic requirements of the verb for obligatory and facultative arguments (ontology concepts in brackets) The dictionary entry for ODBY\u0106 LOT in the above format will be: Information contained in lexicon-grammar entries appeared very useful in various NLP tasks. For example, an important part of information useful for simple sentence understanding may be easily accessed through basic forms of words identified in the sentence. Parts (b) and (c) of the dictionary entries for the identified predicative word will help to make precise hypotheses 9 about the syntacticsemantic pattern of the sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "Despite their merits, the traditional syntactic lexicons, as is the above presented Syntactic Generative Dictionary, are not sufficient to supply all necessary linguistic information to solve all language processing problems. The case of highly inflected Polish (but also other Slavonic languages, Latin, German etc.) demonstrates the need of precise and complete description of morphology. For Polish we delivered within the project POLEX (1994-1996) a large 9 The concept of syntactic hypothesis is crucial for our methods of heuristic parsing making a right choice of hypothesis about the sentence structure may considerably reduce the parsing cost (in time and space). With good heuristics, in some cases it is possible to reduce the grammatical search space considerably and as a result turning the nondeterministic parser into a de facto deterministic one. We explored this idea with very good effects in our rule-based question-answering systems POLINT (see e.g. section Preanalysis in (Vetulani, Z. 1997) .",
"cite_spans": [
{
"start": 460,
"end": 461,
"text": "9",
"ref_id": null
},
{
"start": 993,
"end": 1012,
"text": "(Vetulani, Z. 1997)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "electronic dictionary (Vetulani, Z. et al. 1998 ; Vetulani, Z. 2000) of over 120,000 entries. 10 This resource is easily machine treatable and was used as Polish Lexicon-Grammar complement.",
"cite_spans": [
{
"start": 22,
"end": 49,
"text": "(Vetulani, Z. et al. 1998 ;",
"ref_id": "BIBREF0"
},
{
"start": 50,
"end": 68,
"text": "Vetulani, Z. 2000)",
"ref_id": null
},
{
"start": 94,
"end": 96,
"text": "10",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Lexicon-Grammar of Polish",
"sec_num": "4."
},
{
"text": "Within our real-size application projects 11 Przepi\u00f3rkowski, 2004) in order to select the most important vocabulary, for the purpose of the application expanded with the application specific terminology 14 . Development of PolNet was organized in an incremental way, starting with general and frequently used vocabulary 15 . By 2008, the initial PolNet version based on noun synsets related by hyponymy/hyperonymy relations was already rich enough to serve as core lexical ontology for real-size application developed in the project (POLINT-112-SMS system cf. Vetulani, Z. et al. 2010) . Further extension with verbs and collocations, operated after the 2009, contributed to transform PolNet into a lexicongrammar intended to ease implementation of AI systems with natural language competence and other NLPrelated tasks.",
"cite_spans": [
{
"start": 42,
"end": 44,
"text": "11",
"ref_id": null
},
{
"start": 45,
"end": 66,
"text": "Przepi\u00f3rkowski, 2004)",
"ref_id": null
},
{
"start": 560,
"end": 585,
"text": "Vetulani, Z. et al. 2010)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Citing \u00ab PolNet -Polish Wordnet \u00bb as lexical ontology",
"sec_num": "5."
},
{
"text": "
Warszawy do Tokio samolotem in a Breguet 19 w roku |
1926 (In 1926, Oli\u0144ski flew/made a flight from Warsaw to |
Tokyo in a Breguet 19). |