Grammatical Framework (GF) is a special-purpose functional language for defining grammars. It uses a Logical Framework (LF) for a description of abstract syntax, and adds to this a notation for defining concrete syntax. GF grammars themselves are purely declarative, but can be used both for linearizing syntax trees and parsing strings. GF can describe both formal and natural languages. The key notion of this description is a grammatical object, which is not just a string, but a record that contains all information on inflection and inherent grammatical features such as number and gender in natural languages, or precedence in formal languages. Grammatical objects have a type system, which helps to eliminate run-time errors in language processing. In the same way as an LF, GF uses...

In this mlicle, we 1)reseu! a statistical approach to machine translatiou that is based on Data-Oriented Parsing: l)ata-Oriented Translatiou (DOT). In DOT, we use liuked subtree pairs for creating a derivation of a source seutencc. Each linked subtree lmir has a certain probability, and cousists of two trees: one in the source language and oue in the target lan~ gnage. When a dcriwtfion has been formed with these subtree pairs, we can create a lrauslation fi'om this deriwltion. Since there are typically mauy dif- ferent dcrivatious of lhe same sentence in the source language, there can be a, many different translations for it. The probability of a translatiou can be calculated as the total probability of fill the derivations that form this translatiou. We give the computatioual aspects for tiffs model, show that we can convcrt each subtree pair into a productive rewrite rule, and that the most probable translatiou can be computed by means of Monte Carlo disambiguation. Finally, we discuss some pilot experimeuts with the Veal)mobil corpus.

This paper describes a multi-level design, i.e., a non-uniform approach to interlingual machine translation (MT), in which distinct representational languages are used for different types of knowledge. We demonstrate that a linguistically-motivated "division of labor" across multiple representation levels has not complicated, but rather has readily facilitated, the identification and construction of systematic relations at the interface between each level. Our approach assumes an interlingua derived from the lexical semantics and predicate decomposition approaches of Jackendoff (1983; 1990) and Levin and Rappaport-Hovav (1995a; 1995b). We describe a model of interpretation and representation of natural language sentences which has been implemented as part of an interlingual MT system called PRINCITRAN.

This paper describes the use of verb class memberships as a means of capturing generalizations about manner-of-motion verbs in Synchronous Tree Adjoining Grammars, STAGs, [20, 21, 22]. This approach allows STAGs, which are essentially transfer-based, to take advantage of the same types of generalizations which are generally thought of as wholly the domain of interlingua systems- without giving up any of the lexical specificity unique to transfer-based systems. In this way a machine translation system based on STAGs can respond with seamless flexibility to a wide spectrum of phenomena being presented for translation ranging from idioms and idiosyncratic lexical items to well-behaved verbs that follow lexical rules. 1

this document.

This paper discusses a mathematical concept of language that models both artificial and natural languages and thus provides a framework for a unified language processing methodology. This concept of a language is regarded as a communication tool that allows language users to develop knowledges, while interacting with their universe of discourse, and to communicate with each other, while exchanging knowledges. Criteria for consistent usage of a language are established using a Galois connection between language syntax and language semantics. Solutions to ambiguity, paraphrase, attitude, and other problems concerning the relationship between syntax and semantics are addressed. A general schema for language specification is introduced and algorithms that perform language generation and language analysis are discussed as universal tools defined by the specification schema. Language transformations performed by various kinds of translators are examined and correctness criteria of these translators are defined using the language Galois connection. The paper is structured as follows: Section 1 introduces the framework and justifies the necessity of a unified methodology for language processing. Section 2 presents the mathematical concept of a language. Section 3 illustrates the mathematical concept of a language with three kinds of language structures: natural language, logical language, and programming language. Section 4 discusses the algebraic mechanism of language specification that unifies the methodology for language processing tool development. Section 5 formalizes the criterion for the consistency of the language usage, defines the architecture of a unified language processing system, and shows how the consistency criteria for language usage can be employed as correct...

The automated translation of one natural language to another, known as machine translation (MT), typically requires successful modeling of the grammars of the languages and the relationship between them. Rather than hand-coding these grammars and relationships, some machine translation e#orts employ data-driven methods, where the goal is to learn from a large amount of training examples of accurate translations. One such data-driven approach is statistical MT, where language and alignment models are automatically induced from parallel corpora. This work has also been extended to probabilistic finite-state approaches, most often via transducers.

Translating idioms is one of the most di cult tasks for human translators and translation machines alike. The main problems consist in recognizing an idiom and in distinguishing idiomatic from non-idiomatic usage. Recognition is di cult since many idioms can be modi ed and others can be discontinuously spread over a clause. But with the help of systematic idiom collections and special rules the recognition of an idiom candidate is always possible. The distinction between idiomatic and non-idiomatic usage is more problematic. Sometimes this can be done by means of special words that are only used in an idiom. But in general this distinction is a question of semantics and pragmatics and therefore beyond the abilities of current translation systems. In this paper we investigate the requirements for automatically recognizing idioms and we check whether idiom recognition is possible within current translation systems, i.e. machine translation and translation memory systems. This is of current interest since the developers of translation systems have started to include huge idiom collections in their products. 1

This paper traces the history of efforts to develop computer programs (software) for the translation of natural languages, commonly and traditionally called ‘machine translation ’ (MT), or, in non-English-speaking countries, ‘automatic translation ’ (traduction automatique, avtomaticheskij perevod). A brief history can of course mention only the most significant research systems and projects

summary – This paper aims to show how Montague-style grammars can be completely formalized and thereby declaratively implemented by using the Grammatical Framework GF. The implementation covers the fundamental operations of Montague’s PTQ 4 model: the construction of analysis trees, the linearization of trees into strings, and the interpretation of trees as logical formulas. Moreover, a parsing algorithm is derived from the grammar. Given that GF is a constructive type theory with dependent types, the technique extends from classical Montague grammars to ones in which the Curry-Howard isomorphism is used to explain anaphoric reference. On the other hand, GF has a built-in compositionality requirement that is stronger than in PTQ and prevents us from formulating quantifying-in rules of Montague style. This leads us to alternative formulations of such rules in terms of combinators and discontinuous constituents. The PTQ fragment will moreover be presented as an example of how a GF grammar is modified by replacing English with another target language, French. The paper concludes by a discussion of the complementary rôles of logically and linguistically oriented syntax.