The translation process

This report describes a logic grammar formalism, Modular Logic Grammars, exhibiting a high degree of modularity between syntax and semantics. There is a syntax rule compiler (compiling into Prolog) which takes care of the building of analysis structures and the interface to a clearly separated semantic interpretation component dealing with scoping and the construction of logical forms. The whole system can work in either a one-pass mode or a two-pass mode. In the one-pass mode, logical forms are built directly during parsing through interleaved calls to semantics, added automatically by the rule compiler. In the two-pass mode, syntactic analysis trees are built automatically in the first pass, and then given to the (one-pass) semantic component. The grammar formalism includes two devices which cause the automatically built syntactic structures to differ from derivation trees in two ways: (1) There is a shift operator, for dealing with left-embedding constructions such as English possessive noun phrases while using right-recursive rules (which are appropriate for Prolog parsing). (2) There is a distinction in the syntactic formalism between strong non-terminals and weak non-terminals, which is important for distinguishing major levels of grammar.

The formal specifications of the Rosetta modules do not fully determine the algorithms that implement them. This leaves ample room for consideration about performance optimisation or subjective criteria such as elegance to influence the actual algorithms. Decisions must be made about the algorithms in each of the Rosetta modules. These algorithms, some of which were discussed in the previous chapter, are not the only subject of design decisions that have to be made, nor are they the most important ones. Many questions must be posed and answered while building a system like Rosetta. What does the user interface look like? What debugging facilities must be offered for developing linguistic software? What are the most convenient interfaces between the Rosetta modules, and what do S-trees, syntactic derivation trees (D-trees), semantic derivation trees and dictionaries look like? Can a translation system be run as one process, or is it better to split it into several subsystems? Each module of the Rosetta system produces a set of results. How does one represent such sets if they are very large? In what order do we process their elements? What programming language is chosen? Is linguistic data written in this language or in a special-purpose language that is subsequently translated into the programming language? Many of these problems can be dealt with in various ways. Their solutions are a matter of taste rather than of principle and do not deserve a detailed discussion here. The design of user interfaces is the subject of another discipline and is also beyond the scope of this book. Below we touch upon issues of software engineering that either involve a problematic aspect of the formalism or have proved to be essential in building our translation system.

NOMINALIZATIONS IN ENGLISH have received much attention from transformational grammarians.' In recent years, the correct description of a particular type of English nominalization, known as the DERIVED NOMINAL, has been an issue in the controversy between interpretive semantics and generative semantics. The claim has been made in support of interpretive semantics that verbs derived transformationally from other verbs do not have corresponding derived nominals. I will show that for the class of causative verbs this claim is wrong. Interpretive semantics is a variant of the transformational theory of grammar developed by Noam Chomsky and bthers.2 It emphasizes the role of phrase structure rules and rules of semantic interpretation that apply to pairs of deep and surface phrase markers, and it reduces the role of transformations in grammatical descriptions.3 Generative semantics derives from the same transformational theory, but has rejected phrase structure rules, rules of semantic interpretation like those mentioned above, and the concept of deep structure as that point separating the syntactic and semantic components of the grammar, where all lexical items are inserted. No distinction between syntactic and semantic components is made in generative semantics, and lexical insertion is not restricted to one point in the grammar.4 Chomsky, who has been one of the foremost proponents of interpretive semantics, has claimed that derived nominals like its growth and their arrival constitute a part of English grammar that cannot be described transformationally.5 He supports this claim by contrasting such derived nominals

A new type of grammar for generating two-dimensional patterns is introduced. The generation of a pattern occurs in two phases. First, a string grammar generates a “row” of syntactic variables (the sequential phase). Then a battery of instances of string grammars—one for each symbol produced in the first phase—operate in tandem to create the columns of the final pattern (the parallel phase). Several unrelated properties of the classes of patterns generated by various varieties of matrix grammars are announced.

Multiagent systems (MASs) constitute a powerful methodological platform for dealing with the decision-making process underlying individual travel behaviour. A major challenge is the implementation of a strong notion of agency, through the building of agents capable of drawing information from their environment, processing what is important, and engaging in activities. This paper explores this issue. The notion of mental worlds—that is, cognitive entities involving both syntactic and semantic components—is invoked for accounting the way in which individuals understand their environment. A MAS model (CogMob) is illustrated. It has been developed to simulate the application of a palette of cognitive strategies in urban mobility, that is, habitual behaviour: agents choose the most familiar route; learning by instruction: agents secure information about the less congested routes; reasoning: agents select the most convenient route from a set of those previously used; visioning: agents explore a novel route. Cog...

This article discusses the figurative phraseological units, namely, the idioms of English, German and Kazakh languages in comparative aspect. It appeals to the phraseology of imagery due to the fact that the imagery—a defining component of semantic derivation and semantics of phraseological units in particular. The main goal is to make a contribution to the theory of phraseology, based on the tradition of comparative phraseology. Also, a brief review of domestic and foreign publications in the field of comparative phraseology is done and the article describes the results of practical analysis of idioms in three languages. The main features of the analysed units are multi-component structure, stability, idiomaticity that distinguishes them from other expressions. They can be interpreted in two different conceptual levels: literal sense and figurative sense. In the analysis are used the method of field simulation. The study aims to identify and develop a model describing visual images as elements of the cognitive system. Keywords: Imagery, visuality, imagery means, imagery structure.

Elliptical sentences are frequently seen in Chinese, especially in some particular situations, such as dialogues, which is challengeable to understand specific semantic. Chinese zero pronoun resolution, which recovers a noun phrase in the elliptical position, is an effective method to help machines understand natural languages. Traditional methods use the features, which are extracted from syntactic parsing trees manually. However, the long running time and the inaccuracy of automatic parsing algorithms have a bad influence on practical applications. In this work, we propose a new method based on long-short-term memory network that calculates dense vector representations for mention pairs without using features from syntactic parsing trees. These representations, which capture significant semantics for zero pronoun resolution, are built on distributed representation of words in surrounding contexts and candidate antecedents. Our method contributes to reducing the manual work of extracting features from parsing tress, which improves the F1-score of Chinese zero pronoun resolution system. Experimental results on OnotoNotes 5.0 Chinese dataset show our method achieves better performance compared with the state-of-the-art method.

Statistical machine translation has made tremendous progress over the past ten years. The output of even the best systems, however, is often ungrammatical because of the lack of sufficient linguistic knowledge. Even when systems incorporate syntax in the translation process, syntactic errors still result. To address this issue, we present a novel approach for detecting and correcting ungrammatical translations. In order to simultaneously detect multiple errors and their corresponding words in a formal framework, we use feature-based lexicalized tree adjoining grammars, where each lexical item is associated with a syntactic elementary tree, in which each node is associated with a set of feature-value pairs to define the lexical item’s syntactic usage. Our syntactic error detection works by checking the feature values of all lexical items within a sentence using a unification framework. In order to simultaneously detect multiple error types and track their corresponding words, we propose a new unification method which allows the unification procedure to continue when unification fails and also to propagate the failure information to relevant words. Once error types and their corresponding words are detected, one is able to correct errors based on a unified consideration of all related words under the same error types. In this paper, we present some simple mechanism to handle part of the detected situations. We use our approach to detect and correct translations of six single statistical machine translation systems. The results show that most of the corrected translations are improved.

LangPro is an automated theorem prover for natural language (https://github.com/kovvalsky/LangPro). Given a set of premises and a hypothesis, it is able to prove semantic relations between them. The prover is based on a version of analytic tableau method specially designed for natural logic. The proof procedure operates on logical forms that preserve linguistic expressions to a large extent. %This property makes the logical forms easily obtainable from syntactic trees. %, in particular, Combinatory Categorial Grammar derivation trees. The nature of proofs is deductive and transparent. On the FraCaS and SICK textual entailment datasets, the prover achieves high results comparable to state-of-the-art.

Artifacts (e.g., chair, desk, books) are closely related with peoples' daily life.Evidences have been accumulated from behavior experiment, brain damage patient, and neural functional imaging to show that the cognitive neural processing mechanisms are different for artifacts and natural kinds.The intention-based model assumes that people categorize an objects' by reasoning the creator's intent.Evidence for this model are about "intended function" effect in categorization, naming, and reasoning studies.However, the utility-based model assumes that using-goal is determining how people categorize an object.The interaction between people and objects make object's properties related the using-goal so that the semantic organization of an artifact concept is coherent.This model can explain evidence from study on both intended function and non-intended function.From a third-party perspective, the commonalty of these two models is that they are both purpose-based and the key difference between them is that the creator's versus the user's purpose influences artifact concept's categorization and reasoning.Previous studies used a natural language depiction-based free naming task but seldom separate syntactic component from the semantic component in these depictions.In this study, we recruited 96 participants and manipulated the plausibility of 6 objects' function by creating plausible and implausible versions of function depiction to explore how the plausibility of an object's function affects participants' categorization judgment.The study had two key findings: The plausibility of an object's function affects participants' naming, rating, and reasoning of an artifact.Participants categorize ambiguous depictions of objects as artifacts when objects' function is plausible and categorize them as natural kinds when their function is implausible.The correlation analysis showed that the more plausible an object's function is the higher possibility that participants categorize it as an artifact.In summary, presenting the semantic component of an object's function make ambiguous object categorized more as artifacts.Wang et al. found the role effect which showed that positive role word "you" can illuminate object history effect on categorization judgment.Namely, the role word can affect both the amount of function listed and the object categorization.These findings suggest that role word, object function, and categorization judgment are in a linear mediation relation: role word activate object function, and object function activation categorization judgment.In this current study, we directly probe and proved that the semantic component of object function has effect on artifact concept processing.If with the assumption that both the object man-made formation history and the second person role word can activate object function, then the purpose-based theory can explain both the formation effect and the role effect indicating that the two effects have a common mediating step: first to activate object's function, then have effect on next cognitive output.Therefore, this point developed the purpose-based model.

Urban water systems across the United States are facing a variety of challenges to existing supply and demand dynamics. Responding to these challenges in complex socio-environmental systems (SES) requires integrating various types of information – ranging from hydrologic data to political considerations and beyond – into policy and management decisions. However, the design of institutions, i.e., the formal rules in which urban water utilities are embedded, impact the flow of information, especially across diverse actor groups critical to developing and implementing policy or programmatic responses to signal error. This study develops a Bayesian application of the Robustness of Coupled Infrastructure Systems (CIS) Framework to analyze how the institutional design of a major U.S. urban water system impacts information flow and, ultimately, the goal of resource-delivery robustness. We utilize process-tracing along with an institutional analysis approach called the Institutional Grammar Tool (IGT) to parse formal institutions into their semantic and syntactic components and assess how they may influence a system’s capacity to respond to changing stressors. Our findings have important implications for the (re)design of institutions that better facilitate information flow among key policy actors and support policy changes that promote sustainable long-term urban water supply.

This paper presents the development of a syntactic component for the Vietnamese language. We first discuss the construction of a lexicalized tree-adjoining grammar using an automatic extraction approach. We then present the construction and evaluation of a deep syntactic parser based on the extracted grammar. This is a complete system integrating necessary tools to process Vietnamese text, which permits to take as input raw texts and produce syntactic structures. A dependency annotation scheme for Vietnamese and an algorithm for extracting dependency structures from derivation trees are also proposed. At present, this is the first Vietnamese parsing system capable of producing both constituency and dependency analyses with encouraging performances: 69.33% and 73.21% for constituency and dependency analysis accuracy, respectively. The parser also compares favourably to a statistical parser which is trained and tested on the same data sets.

The OntoNL Framework provides an architecture and re-usable components for automating as much as possible the building of natural language interfaces to information systems. In addition to the syntactic analysis components, OntoNL has semantic analysis components which exploit domain ontologies to provide better disambiguation of the user input. We present in this paper the algorithms used for semantic processing of the natural language queries, as well as an ontology-driven semantic relatedness measure developed for this purpose. We also present extensive evaluation results with different ontologies using human subjects.

This work extends and evaluates a two-dimensional automatic evaluation metric for machine translation, which is designed to operate at the sentence level. The metric is based on the concepts of adequacy and fluency, aiming at decoupling both semantic and syntactic components of the translation process to provide a more balanced view on translation quality. These two elements are independently evaluated by using continuous space and <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$n$</tex> </formula> -gram language modeling frameworks, respectively. Two different implementations are evaluated: a monolingual version that fully operates on the target language side, and a cross-language version that has the main advantage of not requiring reference translations. Both implementations are evaluated by comparing their performance with state-of-the-art automatic metrics over a dataset involving five different European languages.

We define sentence generalization and generalization diagrams as a special sort of conceptual graphs which can be constructed automatically from syntactic parse trees and support semantic classification task. Similarity measure between syntactic parse trees is developed as a generalization operation on the lists of sub-trees of these trees. The diagrams are representation of mapping between the syntactic generalization level and semantic generalization level (anti-unification of logic forms). Generalization diagrams are intended to be more accurate semantic representation than conventional conceptual graphs for individual sentences because only syntactic commonalities are represented at semantic level.