Universal quantification in machine translation

نویسنده

  • Cornelia Zelinsky-Wibbelt
چکیده

This approach has been developed in the context of the E U R O T R A machine translation (MT) project and thus has been d e s i g n e d w i t h r e spec t to a s y n t a x based s t r a t i f i c a t i o n a l translation process? We assume that in a semantic representation determiners are deleted and that their semantic funct ion which is represented by semantic features is percolated into the m o t h e r n o d e . The s e m a n t i c f u n c t i o n s of d e t e r m i n e r s are explicated. The interact ion between grammatical and lexical quant i f ieat iun is outlined. Ensemble theory is applied to the " coun t " / "mass " noun d i s t i n c t i o n . T r a n s f e r of quant i f ica t ion between German, English, and French is i l lustrated with respect to the "count"/"mass" distinction. The article closes with an outlook on [he relevance of generalized quantif iers for Machine Translation. 1. Semantic representation of determiners in EUROTRA E U R O T R A aims at def in ing a semantic representat ion which guarantees s imple transfer between all European languages, that is, it should be "euroversal". The concept of "euroversality" implies, amongst others, a semantic representation in a canonical form out or" which all European languages may be generated. With respect to this canonical form it is reasonable to delete the determiners dur ing translation into the semantic representation and to represent their meaning by semantic features of the NP. This step may be motivated pr imari ly by two facts: (1) Languages vary with respect to the use of 0determiners. (2) The set properties realized by an enti ty are expressed dif ferent ly in d i f ferent languages. Tim idea that determiners are not deep structure constituents, but that their surface structure consti tuents have to be generated from a semantic representation is not new. It can already be found for example in P E R L M U T T E R 1970. Moreover, to represent the quant i f ier by means of semantic features of the NP implies that the ent i ty which is focussed by the process of determinat ion cannot be referred to directly, but only as the argument of the determiner which provides a new referent (cf. P I N K A L 1986). BARWISE & C O O P E R (1981) c o n s i d e r determiners as two-place predicates which take the noun which is the domain of quant i f icat ion as one argument , and the rest of the sentence, which is the predicate quantif ied as the other argument. With respect to the EUROTRA MT system this has important implicat ions for the translation between the syntactic dependency level the E U R O T R A Relat ional Structure (ERS) and the s e m a n t i c l eve l the i n t e r f a c e S t r u c t u r e (IS). Determiners which have the function of modify ing nouns at ERS on the basis of several syntactic conditions establish d i f fe ren t types of determination. Those types of determinat ion are the ba,;is for d e d u c i n g (i .e. t r a n s l a t i n g ) e x a c t l y tha t informat ion which yields the new referent in the NP by unifying with the semantic features of the noun. Although both determiners and quantif iers have characterist ic functions, they have others in common, so that a borderline is d i f f icul t to draw. Cases of crossclassification exist in many languages, as for example the one consti tuted by German der/ein/O, French le/un/O, English the/a/O (cf. VATER 1963). This is why we describe both determiners and quantif iers by a common set of semantic features. 2. The semantic functions It is agreed in the l i terature that determiners and quantif iers share the function of DELIMITATION (cf. VATER 1980). This del imita t ion consists in the localisation of a referent in the speech or textual context or the non-l inguis t ic situation or in relation to the presupposed knowledge of the hearer or reader (only the first of these functions, and this again in a rather restricted way, may be represented in the E U R O T R A system). B A R W I S E & C O O P E R (1981) r e f e r to this f u n c t i o n of d e l i m i t a t i o n as the p r o p e r t y " l ives on" and d e f i n e that determiners "assign to common count noun denotations (i.e. sets) A a quant i f ier that lives on A." (BARWISE & COOPER 1981.179) 2.1. Quant i f ica t ion over whole sets: "generic" versus "identifying" It is the function of determiners and quantif iers to quant i fy over gets of enti t ies. The writer 's motivation to create sets is that the enti t ies which should be members of the set share one or several p r o p e r t i e s . F o l l o w i n g the t r a d i t i o n of the M O N T A G U E approach, BARWISE & COOPER treat all NPs as quantif iers which denote sets of properties of individuals. There are two basic types of WHOLE SETS, which may be created. (l) The enti ty 's e x t e n s i o n is created "generically" by means of it 's inherent lexical meaning as in the following example: Die Linguisten sind in formalen Sprachen geiibt (Linguists are practised in formal languages.). (2) Here the NP quantif ies exactly over the complete set of l inguists of the actual world. An lntensional property of the enti ty set makes possible it 's "identification". In this case a WItOLE SET is referred to which is precisely delimited (cf. VATI?;R 1963, P L A T T E A U 1980). This type of ent i ty set may only be established context-sensi t ive. It is thus a set which may be referred to as a WHOLE SET only with respect to a certain domain of interpretation, which is the intensional property: The linguists of EUROTRA ... This NP quant i f ies exactly over that set of linguists who work for EUROTRA, 2.2. The semantic functions of determiners: determiners as variables and as var iab le-b inding fnnctions It is the funct ion of determiners to select one or several enti t ies from a set of ent i t ies (cf. PLATTEAU 1980). The salient function of indefini te determiners is equivalent to that of the existential quant i f ier (cf. L A N G E N D O N C K 1980, P L A T T E A U 1980); they introduce new entit ies into the speech or text situation. Thus they only express that entit ies exist in the speech situation, wi thout "specifying" which. It is an inf ini te set of a p o t e n t i a l of e n t i t i e s (cf. HAWKINS 1978.198). We may therefore say that indef ini te determiners in their salient function are variables. This yields a PARTIAL SET of entities which is "existential". Beside this salient function the indefini te determiner may also "specify" entities, if it is clear in the universe of discourse which ent i ty is designated (ef. OOMEN 1977 and DI EUGENIO 1986).

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Rule-based Translation of Quantifiers for Chinese-Japanese Machine Translation

Quantifiers and numerals often cause mistakes in Chinese-Japanese machine translation. In this paper, an approach to quantifier translation is proposed based on the syntactic features after classification. First, morphological analysis is performed on sentences extracted from a Chinese-Japanese aligned corpus, which consists of quantifiers and numerals. Next, statistical information is obtained...

متن کامل

Two-Dimensional Specification of Universal Quantification in a Graphical Database Query Language

We propose a technique for specifying universal quantification and existential quantification (combined with negation) in a two-dimensional (graphical) database query language. Unlike other approaches that provide set operators to simulate universal quantification, this technique allows a direct representation of universal quantification. We present syntactic constructs for specifying universal...

متن کامل

Merged bilingual trees based on Universal Dependencies in Machine Translation

In this paper, we present our new experimental system of merging dependency representations of two parallel sentences into one dependency tree. All the inner nodes in dependency tree represent source-target pairs of words, the extra words are in form of leaf nodes. We use Universal Dependencies annotation style, in which the function words, whose usage often differs between languages, are annot...

متن کامل

Collocations, Dictionaries and MT

Collocations pose specific problems in translation (both human and machine translation). For the native speaker of English it may be obvious that you ’pay attention’, but for a native speaker of Dutch it would have been much simpler if in English people ’donated attention.’ Within an MT system, we can deal with these mismatches in different ways. Simply adding the entry to our bilingual diction...

متن کامل

Toward Verbalizing Ontologies in isiZulu

IsiZulu is one of the eleven official languages of South Africa and roughly half the population can speak it. It is the first (home) language for over 10 million people in South Africa. Only a few computational resources exist for isiZulu and its related Nguni languages, yet the imperative for tool development exists. We focus on natural language generation, and the grammar options and preferen...

متن کامل

A Comparative Study of English-Persian Translation of Neural Google Translation

Many studies abroad have focused on neural machine translation and almost all concluded that this method was much closer to humanistic translation than machine translation. Therefore, this paper aimed at investigating whether neural machine translation was more acceptable in English-Persian translation in comparison with machine translation. Hence, two types of text were chosen to be translated...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1988