<?xml version="1.0" standalone="yes"?>
<Paper uid="W00-0204">
  <Title>An interlingua aiming at communication on the Web: How language-independent can it be? Ronaldo Teixeira Martins ronaIdo @nilc.icmsc.sc.usp.br</Title>
  <Section position="2" start_page="0" end_page="24" type="intro">
    <SectionTitle>
1. Introduction
</SectionTitle>
    <Paragraph position="0"> The widespread use of the Web and the growing Intemet facilities have sparked enormous interest in improving the ways people use to communicate. In this context multilingual Machine Translation systems become prominent, for they allow for a huge information flow. To date, MT systems have been built under limited conditions, of which we highlight two: i) in general, they mirror one-to-many(languages) or many(languages)to-one approaches, often involving English at the &amp;quot;one&amp;quot; end; ii) communication is reduced to basic information exchange, ignoring richness and flexibility implied by human mind. The first limitation has been seldom overcome, since it requires a robust environment and research teams that can cope with knowledge of several languages 1, to derive precise automatic language analyzers and synthesizers. The second limitation follows up the first: adding up communicative issues to linguistic processing/modeling makes still harder to overcome MT limitations.</Paragraph>
    <Paragraph position="1"> In this article, we elaborate on work using an interlingua conceived to overcome the first limitation, i.e., to allow for a many-to-many information exchange environment, which shall be plugged in a nontraditional Internet platform. The goal is to allow interlocutors to entangle communication even if they do not share the same mother tongue or the English Standing, most often, for natural language, or NL.  language, unlike MT systems that have just one language at one of their edges. As the main component of a Knowledge-Base MT system (hereafter, KBMT), the interlingua approach has been developed under the Universal Networking Language Project, or simply UNL Project. What makes the interlingua UNL special is its intended use: as an electronic language for networks, it has to allow for high quality 2 conversation systems involving many languages. As the main component of a KBMT system, it has to be sufficiently robust to ground research and development (R&amp;D) of the language-specific modules to be attached to the system. It is this latter perspective that is undertaken here: from the viewpoint of R&amp;D, we discuss how broad, or language-independent, the interlingua UNL is, especially focusing on its syntax and coverage. In addition to being consistent and complete to represent meaning, we also consider its sharing by researchers all around the world, which is an important bottleneck of the UNL Project, since information exchange by researchers during R&amp;D brings about the problems introduced by the interlingua UNL itself, concerning both its formalism and foundational issues. Before discussing this topic in Section 5, we present an overview of the UNL Project (Section 2) and describe the main features of the interlingua UNL (Section 3). In Section 4, we describe the UNL system architecture. Hereafter, 'interlingua UNL' will be simply referred to as UNL, the acronym for Universal Networking Language. Also, the viewpoint presented here is that of interlingua users who experience R&amp;D for a given NL, and not of its authors.</Paragraph>
  </Section>
class="xml-element"></Paper>