A Survey of Word Reordering in Statistical Machine Translation: Computational Models and Language Phenomena

Abstract

Word reordering is one of the most difficult aspects of statistical machine translation (SMT), and an important factor of its quality and efficiency. Despite the vast amount of research published to date, the interest of the community in this problem has not decreased, and no single method appears to be strongly dominant across language pairs. Instead, the choice of the optimal approach for a new translation task still seems to be mostly driven by empirical trials.To orientate the reader in this vast and complex research area, we present a comprehensive survey of word reordering viewed as a statistical modeling challenge and as a natural language phenomenon. The survey describes in detail how word reordering is modeled within different string-based and tree-based SMT frameworks and as a stand-alone task, including systematic overviews of the literature in advanced reordering modeling.We then question why some approaches are more successful than others in different language pairs. We argue that, besides measuring the amount of reordering, it is important to understand which kinds of reordering occur in a given language pair. To this end, we conduct a qualitative analysis of word reordering phenomena in a diverse sample of language pairs, based on a large collection of linguistic knowledge. Empirical results in the SMT literature are shown to support the hypothesis that a few linguistic facts can be very useful to anticipate the reordering characteristics of a language pair and to select the SMT framework that best suits them.

Neural Machine Translation with Reordering Embeddings

“…In PBSMT, there has been a substantial amount of research works about reordering model, which was used as a key component to ensure the generation of fluent target translation. Bisazza and Federico (2016) divided these reordering models into four groups: Phrase orientation models (Tillman, 2004;Collins et al, 2005;Nagata et al, 2006;Zens and Ney, 2006;Galley and Manning, 2008;Cherry, 2013), simply known as lexicalized reordering models, predict whether the next translated source span should be placed on the right (monotone), the left (swap), or anywhere else (discontinuous) of the last translated one.…”

Machine Translation using Semantic Web Technologies: A Survey

“…Commonly, SMT systems are trained using reference translations by which ML algorithms are able to analyze the data and find patterns by themselves, thus being able to translate text without any rules created by humans. Although some basic linguistics mistakes have been solved by Tree-based and Neural Networkbased approaches, the lack of complex linguistic rules still causes ambiguity problems (e.g., errors on relative pronouns)- [149]. An additional problem of the latter approach is the complexity of performing error analysis over outputs.…”

“…2. Structural divergence: By definition, structural reordering is reorganizing the order of the syntactic constituents of a language according to its original structure [149]. It in turn becomes a critical issue because it is the core of the translation process.…”

Integration of a Multilingual Preordering Component into a Commercial SMT Platform

“…They may be deterministic (i.e., leading to a single reordered variant of the given source sentence) or non-deterministic (i.e., leading to several variants of the source sentence). An extensive overview of different preordering approaches is presented by Bisazza and Federico (2016). Xu et al (2009) and Nakagawa (2015) proposed preordering methods which can be applied to many different language pairs.…”

“…English and Japanese differ in many syntactic aspects: the order of the clauses is different, as well as the order of the words within the clauses. An extensive overview of the differences on various syntactic levels can be found in Bisazza and Federico (2016). The rule set for Japanese is taken from Lee et al (2010).…”