Báo cáo khoa học: The Position of Prepositional Phrases in Russian

A problem frequently encountered in the automatic parsing of Russian texts is the correct structuring of prepositional phrases in sentences. Studies of text samples indicate that, when other criteria are absent, the syntactic governors of prepositions can be determined with a high degree of accuracy by reference to the relative position and part-ofspeech of elements in the clausal environment.
Nội dung trích xuất từ tài liệu:
Báo cáo khoa học: "The Position of Prepositional Phrases in Russian" [Mechanical Translation, vol. 8, No. 1, August 1964]The Position of Prepositional Phrases in Russianby Kenneth E. Harper, The Rand Corporation, Santa Monica, California A p roblem frequently encountered in the automatic parsing of Russian texts is the correct structuring of prepositional phrases in sentences. Studies of text samples indicate that, when other criteria are absent, the syntactic governors of prepositions can be determined with a high degree of accuracy by reference to the relative position and part-of- speech of elements in the clausal environment. One of the primary goals of computational linguistics necessary to select as governor one word that willis the development of automatic parsing programs for represent the group.use in processing written texts. There is an enormous The automatic parsing program for Russian, as devel- oped at RAND, has been described elsewhere.2 For pres-utility for computer programs that will produce struc-tural descriptions of sentences comparable to the de- ent purposes, we note only that a sample of somescriptions produced by humans. Although both prod- 10,000 sentences of text from Russian physics journalsucts are admittedly imperfect, given the present has been subjected to the program; the resulting de-inadequacies of grammatical theory, the information scriptions have been verified and corrected by humans.generated in the course of automatic syntactic analysis All dependency relations between the word pairs inis of immediate use in language study: the parsing this text sample are recorded on magnetic tape. Auto-programs themselves can be improved, and a data matic retrieval programs applied to this processedbase is provided for testing the theoretical principles text enable researchers to conduct distributional studiesunderlying the program. The parsing routine is a re- (e.g., through concordances for specified words or syn-search tool for the automatic assembling of facts about tactic constructions).the combinatorial properties of sentence elements; in The present study deals with a common but difficultparticular, it is a means of achieving specificity in problem in machine parsing: the automatic structuringsyntactic description. (In addition, automatic parsing of prepositional phrases in sentences. As with otherhas practical applications in such activities as machine sentence elements, the preposition is said to dependtranslation, indexing, and abstracting.) on one other word (its governor); in turn, it governs A parsing program must be based on a model of at least one other word (its dependent, or object). Thelanguage, however imperfect and tentative that model difficulty lies in the assignment of the correct governormay be. The program described in the present paper is for the preposition. We should stress the fact that webased on a simple dependency grammar, adopted for are not seeking to determine the governor of the prepo-linguistic research in Russian at The RAND Corporation.1 sitional phrase. Machine parsing proceeds by a com-In this model, the structure of a sentence is conceived parison of the respective morphological and syntacticas a tree-like set of relations among the words in the properties of pairs of word-tokens in the sentence; thesesentence. One word in every clause is said to be inde- properties are stored in the machine dictionary, andpendent (in our convention, the predicate); except for unless we create an unmanageably large dictionary bythis item, every other word in the clause depends on storing prepositional phrases, we can parse only inone and only one other word. (Double dependency is terms of word pairs (preposition/governor and prepo-allowable in special instances, e.g., with relative pro- sition/dependent). Can we construct a computer pro-nouns.) The word on which a word depends is said to gram capable of accounting for the numerous relation-be its governor; the latter term is vised merely as a ships that a preposition bears to its syntactic governor?complement to dependent, and does not necessarily Can we ignore, or can we utilize, the semantic infor-correspond to the usage of the term in traditional gram- mation contained in the prepositional phrase,—informa-matical description. The syntactic relationships d ...