FREE ELECTRONIC LIBRARY - Theses, dissertations, documentation

Pages:   || 2 |

«Dirk Heylen OTS, Universiteit Utrecht Kerry Maxwell CL/MT Group, Essex University Lexical Functions and the Translation of Collocations Abstract This ...»

-- [ Page 1 ] --

Dirk Heylen

OTS, Universiteit Utrecht

Kerry Maxwell

CL/MT Group, Essex University

Lexical Functions and the Translation of Collocations


This paper discusses the lexicographical concept of Lexical functions (Mel'chuk

and Zholkovsky, 1984) and their potential exploitation in the development of a

machine translation lexicon designed to handle collocations. We show how lexical

functions can be thought to reflect crosslinguistic meaning concepts for collocational

structures and their translational equivalents, and therefore suggest themselves as some kind of language-independent semantic primitives from which translation strategies can be developed.

1. Introduction Collocations present specific problems in translation, both in human and automatic contexts. If we take the construction heavy smoker in English and attempt to translate it into French and German, we find that a literal translation of heavy yields the wrong result, since the concept expressed by the adjective (something like 'to excess') is translated by grand in French and stark in German. The example is a good illustration of the fact that languages differ in how they express the concepts mediated in collocational structures, e.g. a heavy smoker in English is a 'large' smoker in French (grand fumeur) and a 'strong' smoker in German (starker Raucher). We observe then that in some sense the adjectives stark, grand and heavy are equivalent in the collocational context, but that this is of course not typically the case in other contexts, cf: grande boîte, starke Schachtel and heavy box, where the adjectives could hardly be viewed as equivalent. It seems then that adjectives which are not literal translations of one another may share meaning properties specifically in the collocational context.

How then can we specify this special equivalence in the machine translation dictionary? The answer seems to lie in addressing the concept which underlies the union of adjective and noun in these three cases, i.e.

intensification, and hence establish a single meaning representation for the adjectives which can be viewed as an interlingual pivot for translation. The CEC project ET-10/75: Collocations and the Lexicalisation of Semantic Operations, is concerned with investigating the lexical functions (LF's) of Mel'chuk (Mel'chuk and Zholkovsky 1984), as a candidate interlingual device for the translation of adjectival and verbal collocates. In this paper we Hie way words work together / combinatorics 299 will attempt to provide an overview of the key areas of research entailed by the project, including a characterisation of collocational structures, an evaluation of Lexical Functions and proposals for representation and translation strategies.

2. A characterisation of collocational structures In Mel'chuk's 'Explanatory Combinatory Dictionary' (ECD, see (Mel'chuk et al., 1984)) expressions such as une ferme intention, une résistance acharnée, un argument de poids, un bruit infernal and donner une leçon, faire un pas, commetre un crime are described in the lexical combinatorics zone. These 'expressions plus ou moins figées" are considered to consist of two parts - which we will call the base and the collocate. In the examples above the nouns are the bases and the adjectives and the verbs are the collocates. The idea that all adjective collocates and all the verb collocates share an important meaning component - roughly paraphrasable as intense and do respectively - and the fact that the adjectives and verbs are not interchangeable but are restricted with this meaning to the accompanying nouns, is coded in the dictionary using lexical functions (in this case Magn and Oper).

This class of loosely fixed combinations can be identified as collocations.

We try to determine some properties that fit th s set of expressions and that are compatible with a number of ways the term collocation is used in the literature.

Looking at several characterisations,1 we see the following predicates surface repeatedly, albeit in different guises: recurrent, idiomatic, contextually restricted, cohesive and arbitrary. These notions can be summed up as 'collocations are cohesive, recurrent, arbitrary combinations of words which are not idioms but in which the (figurative) meaning of one part is contextually restricted to the specific combination".

Let us briefly explain these notions and see which properties we have used in our own characterisation. The first notion, recurrence is most typically captured in the definition of collocation as a 'recurrent combination of words that co-occur more often than expected by chance" (Smadja 1993). We have not included this notion in our definition, although we have used it to extract potential collocations from texts (see (Heylen et al., 1993)). It is a surface characteristic related to the occurrence of collocations.

The second notion pertains to the semantic properties of the parts making up the collocation and the way they are combined. In the ODCIE (Cowie and Mackin 1975), it is stated that a collocation is 'not an idiom because the meaning of the whole reflects the meaning of the parts".2 Mel'chuk also says that they are not 'idioms stricto sensu". As far as the meaning of the parts is concerned it is said that 'one word has a figurative sense", but 'the other element appears in a familiar, literal sense". This second notion is not found in every definition of collocations. In most corpusbased approaches to 300 Euralex 1994 collocations, for instance, this property is absent. It is not mentioned in the (Benson et al. 1986) definitions either. Another problem is that it is sometimes difficult to decide whether we are dealing with a figurative meaning or with a meaning which is not the primary meaning. The least we can say is that in most cases the meaning of the collocate is not its most prominent one.

Another way to say that this meaning is special, is by using the third notion.

The particular reading of the collocate is one that is 'not found outside that limited context" (ODCIE). This means that one part is only used in this sense in restricted contexts (although some lexical variation is possible). Also the term 'loosely fixed combination" covers this property. It also ties in with the fourth notion, cohesiveness which means that 'the presence of one or several words of the collocation often implies or suggests the rest of the collocations". (Smadja, also BBI). We combine these notions by stating that the base of a collocation selects a specific word (or a limited set of words) to express a certain meaning and this selected word, the collocate, is only used with a limited set of bases to express this meaning. This combination of properties is responsible for the cohesion between the elements and probably also for their recurrence (certainly if we assume that the meanings are ones that are often expressed).

The fifth notion 'arbitrary' (Benson, Smadja) is sometimes also referred to as 'lexical'. There is no semantic reason why a smoker is 'heavy' in English and 'strong' in German. This is to a large degree determined by coincidental lexical selection.

Our notion of collocation is a combination of these properties (though it considers the first as a derived property rather than a criterial one), and adds the idea that to a large extent the figurative meaning of the part can be identified with lexical functions.

3. Analysis of lexical functions

In this section we provide a brief introduction to Mel'chuk's proposals concerning lexical functions (Mel'chuk et al., 1984), (Mel'chuk and Polguère 1987), (Mel'chuk and Zholkovsky 1988a). Broadly speaking, Lexical Functions (henceforth LF's) are used to describe systematically certain semantic and collocational relations existing between lexemes. They apply at the deep syntactic level of the Meaning-Text Model (MTM),3 and are used to indicate either a set of phraseological combinations related to a keyword (argument lexeme to which they apply) or those words which can replace a keyword under certain conditions. We focus on the former class, called syntagmatic LF's.

A definition of the notion of lexical function can be found in (Mel'chuk and Zholkovsky 1988b: 51): 'A LF is a function in the mathematical sense representing a certain extremely general idea, such as 'very', 'begin', or 'implement', or else a certain semantico-syntactical role. A lexical function The way words work together / combinatorics 301 f associates with a word Wo called its argument, or KEY WORD, the set of words and phrases which express - contingent on WQ - the meaning or role which corresponds to £' We should note that the co-occurrence restrictions these lexical functions are intended to capture are those that are truely linguistic and not the restrictions between lexemes that 'cannot co-occur only because of their meanings and of our knowledge of the world'.4 So the philosophy is that LF's are in no way intended to refer to the semantics of the lexemes over which they operate. We should note however that the relation denoted by the LF itself is somehow semantic in nature, cf: the LF Magn denoting 'intensification' and the LF Degrad denoting a process of 'becoming worse or bad'.

4. Issues in translation

Despite their differences, the classical MT transfer and interlingua architectures share the same basic mechanics. Both fall within the paradigm that is concerned with mappings between symbolic representations that start and end with natural language. Differences between these architectures and their variants can be characterised by the number of representation levels, their interpretation, and the mappings between them. In a transfer architecture an expression in the source language will be analysed up to some specific level of representation. Next, the resulting structure is mapped onto a similar structure of the target language by specific transfer rules. From this structure the target language expression is generated. Interlingua systems assume a level of representation with structures that are shared by both the source and the target language. In this case no transfer mapping is necessary, or one could say that the transfer mapping is the identity function.

The project has tried to investigate the use of Lexical Functions as an interlingual device, i.e. one which is shared by the semantic representations of collocations in the language pairs.5 The typing of a collocation with such a function opens up the way to a treatment of collocations inside a given language module and hence to a substantial reduction in the number of collocations explicitly handled in the multilingual transfer dictionary. The existence of a collocation function is established during analysis. This information is used to generate the correct translation in the target language. To illustrate, the English analysis module might analyse (1) heavy smoker as (2) Magn(smoker). The transfer module maps (2) onto (3) Magn(fumeur) which is then synthesised by the French module to (4) grand fumeur.

The example points out that the translation strategy is a mixture of transfer and interlingua. The bases are transferred but the representation of the collocate is shared between the source and the target representation. This treatment of collocations rests, among others, on the assumptions that there are only a limited number of lexical functions, that lexical functions can be 302 Euralex 1994 assigned consistently, that all (or a significant number of) collocations realise a lexical function, that lexical functions are not restricted to particular languages, etc. In the following paragraphs we discuss two problems. The first deals with the appropriateness of Lexical Functions as an interlingual device.

The second is concerned with the problems that arise when collocations do not translate into collocations.

4.1 Lexical functions as interlingua4.1.1 Overgenerality

An important problem stems from the interpretation of LF's implied by their use as an interlingua - namely that the meaning of the collocate in some ways reduces to the meaning implied by the lexical function. This interpretation is trouble-free if we assume that LF's always deliver unique values; unfortunately cases to the contrary can be readily observed. An example attested from our corpus was the range of verb adverb constructions possible with the verbal head oppose, e.g. adamantly, bitterly, consistently, steadfastly, strongly, vehemently, vigorously, deeply, resolutely.

The function Magn is an appropriate descriptor in all cases since each adverb functions as a typical intensifier in this context. However each adverb also denotes some other meaning aspect(s), e.g. consistently suggests something like 'continuingly', bitterly suggests 'animosity', etc. These meaning aspects are not captured by the function, i.e. Magn refers to the intensification device inherent in each adverb but cannot say how these possible intensifiers differ. Now, one may argue that the difference between these adverbs can be highlighted in the semantic specifications (or definitions) associated with their individual entries in the lexicon. This does not, however, offer any consolation if our aim is to exploit purely LF's as an interlingual device at the point of translation. Their imprecision will mean that we have no means of distinguishing between the various intensifiers possible in the context of a given keyword, and hence will not have sufficient information to choose the correct translation where, correspondingly, multiple possibilities exist in the target language.

4.1.2 Possible enhancements

It is essentially in addressing the issue of overgenerality that Mel'chuk introduces sub- and superscripts to lexical functions, enhancing their precision and making them sensitive to meaning aspects of the lexical items over which they operate. Superscripts are intended to make the meaning of the LF more precise, subscripts are used to reference a particular semantic component of a keyword. The introduction of such devices into the account of LF's demonstrates both the need for precision and the fact that it does seem necessary to address semantic aspects of lexemes standing in co-occurrence relations. In fact it has been suggested by some (e.g. (Anick The way words work together / combinatorics 303 and Pustejovsky 1990), (Heid and Raab 1989)) that collocational systems may be systematically predictable from the lexical semantics of nouns.

Pages:   || 2 |

Similar works:

«ANÁLISIS DIMENSIONAL Y ESTUDIO DE VALORES: EL CAMBIO CULTURAL EN ESPAÑA Mariano Torcal Loríente Universidad Autónoma de Madrid RESUMEN. La hipótesis del cambio cultural es una de las más importantes entre las que han intentado encontrar una explicación al cambio electoral observado en las dos últimas décadas. El estudio de esta hipótesis para el caso español permite comprobar aspectos importantes ya no sólo relacionados con ella, sino sobre cuestiones teóricas y metodológicas de...»

«Journal of Sport Management, 2008, 22, 205-226  © 2008 Human Kinetics, Inc. Brand Image and Fan Loyalty in Professional Team Sport: A Refined Model and Empirical Assessment Hans H. Bauer, Nicola E. Stokburger-Sauer, and Stefanie Exler University of Mannheim This study highlights the importance of brand image for fan loyalty in team sport. A parsimonious 4-factor, 20-indicator structure effectively represents brand image. In contrast to Keller’s proposed model, relationships...»

«Faculty of Science and Technology MASTER’S THESIS Study program/ Specialization: Spring semester, 2013 Offshore Technology/ Subsea Technology Open / Restricted access Writer:. Indra Permana (Writer’s signature) Faculty supervisor: Dr. Daniel Karunakaran (Adjunct Professor) (University of Stavanger, Subsea 7 Norway) External supervisor(s): Dr. Dasharatha Achani (Subsea 7 Norway) Title of thesis: A Study on Engineering Critical Assessment (ECA) of Subsea Pipeline Girth Welds for Reeling...»

«Pro-Training Collar™ Training Guide Streaming-video of training available on starmarkacademy.com Congratulations on choosing the Pro-Training Collar™! The Pro-Training Collar™ is the collar of choice for dog owners who want the most effective and fun Obedience Training Process available on the market today. The Pro-Training Collar™ actually enhances communication between you and your dog. It allows you to guide your dog through the training process in a safe, and enjoyable manner. A...»

«On The Epistemic Value of Photographs∗ Jonathan Cohen†and Aaron Meskin‡ The photograph is the only picture that can truly convey information, even if it is technically faulty and the object can barely be identified. A painting of a murder is of no interest whatever; but a photograph of a murder fascinates everyone. — Gerhard Richter, quoted in [Obrist, 1995], 56–57. Many have held that photographs give us a firmer epistemic connection to the world than do other depictive...»

«UNLIKELY, LIKELY AND IMPOSSIBLE INTERSECTIONS WITHOUT ALGEBRAIC GROUPS ´ ZOE CHATZIDAKIS, DRAGOS GHIOCA, DAVID MASSER, AND GUILLAUME MAURIN Abstract. We formulate function field analogues for the PinkZilber Conjecture and for the Bounded Height Conjecture. The “special” varieties in our formulation are varieties defined over the constant field. We prove our function field Pink-Zilber Conjecture for all subvarieties, and we prove our function field Bounded Height Conjecture for a...»

«“Intermediate Just call me Skipper” Leaving the Dock Steering: Sailboats are steered by a rudder. For a rudder to work there needs to be water flow across it. In forward gear the initial water flow is created by the propeller’s thrust, so a sailboat will respond very quickly when going forward. However, when in reverse gear, the water thrust is away from the rudder, thus the only water flow over the rudder will be once the sailboat begins to move, so sailboats respond a lot slower when...»

«Racial Disparity in NYPD Stops-and-Frisks: THE CENTER FOR CONSTITUTIONAL RIGHTS PRELIMINARY REPORT ON UF-250 DATA FROM 2005 THROUGH JUNE 2008 Released January 15, 2009 www.CCRjustice.org Racial Disparity in NYPD Stops-and-Frisks: The Center for Constitutional Rights Preliminary Report on UF-250 Data Obtained for 2005 through June 2008 TABLE OF CONTENTS I. EXECUTIVE SUMMARY II. INTRODUCTION III. KEY FINDINGS OF THE UF-250 DATA A. Stops B. Frisks C. Arrests, Summons, Weapons & Contraband Yield...»

«Structure of Three-Manifolds – Poincar´ and geometrization conjectures e Shing-Tung Yau1,2 Ladies and gentlemen, today I am going to tell you the story of how a chapter of mathematics has been closed and a new chapter is beginning. Let me begin with some elementary observations. A major purpose of Geometry is to describe and classify geometric structures of interest. We see many such interesting structures in our day-to-day life. Let us begin with topological structures of a two dimensional...»

«PROSPECTO DEFINITIVO. Los valores descritos en este prospecto han sido registrados en el Registro Nacional de Valores que lleva la Comisión Nacional Bancaria y de Valores. Dichos valores no podrán ser ofrecidos ni vendidos fuera de los Estados Unidos Mexicanos, a menos que sea permitido por las leyes de otros países. VOLKSWAGEN BANK, S.A., INSTITUCIÓN DE BANCA MÚLTIPLE PROGRAMA DE CERTIFICADOS BURSÁTILES BANCARIOS, CERTIFICADOS DE DEPÓSITO BANCARIO DE DINERO A PLAZO, PAGARÉS CON...»

«AUTHOR STYLE GUIDELINES How to submit a manuscript using our online submission process Submissions to Corporate Governance: An International Review can be made online using Manuscript Central. To submit to the journal go to http://mc.manuscriptcentral.com/cgir. Full instructions on making your submission are provided at this website. You should receive an e-mail acknowledgement within a few minutes after completing your submission. Thereafter, the system will keep you informed of the progress...»

«U.S. Department of Energy Collegiate Wind Competition 2016 RULES AND REQUIREMENTS Revision 2, 9/4/15 Preface The U.S. Department of Energy Collegiate Wind Competition 2016 will be governed and adjudicated by this rules and requirements manual, which is intended to establish fair contest rules. The organizers reserve the right to change contest criteria, rules, and measurable outcomes as needed. In addition, we encourage the teams to bring to our attention rules that are unclear, misguided, or...»

<<  HOME   |    CONTACTS
2016 www.theses.xlibx.info - Theses, dissertations, documentation

Materials of this site are available for review, all rights belong to their respective owners.
If you do not agree with the fact that your material is placed on this site, please, email us, we will within 1-2 business days delete him.