TL;DR
This paper explores the challenges and potential of using computers to aid in the creation and updating of bilingual dictionaries. It proposes a phased approach to computer-aided lexicography, moving from basic word lists to fully automated typesetting, aiming to overcome the inherent complexities of semantic equivalence and the rapid obsolescence of dictionary content.
What Problem Does This Paper Address?
Creating comprehensive bilingual dictionaries is a complex undertaking. Unlike monolingual dictionaries that focus on a single language's orthography, grammar, and definitions, bilingual dictionaries must also grapple with accurate translations, which requires a deep understanding of both languages and their nuances. This complexity is amplified when aiming for a "modern" bilingual dictionary, which goes beyond simple word-for-word glossaries to provide a rich set of information for translators. The paper highlights the difficulties in managing this vast amount of data, ensuring semantic equivalence, and dealing with "faux amis" (deceptive cognates). Furthermore, traditional manual editing processes are time-consuming and lead to dictionaries quickly becoming outdated. The paper addresses the need for a more efficient and dynamic approach to bilingual lexicography, specifically through the application of computer-aided editing.
Methods
The paper outlines a three-phase approach to computer-aided bilingual lexicography:
- Phase 1: "Skeleton" Word Lists: This initial phase focuses on creating basic word lists with essential features like grammar, subject field, and dialect. The goal is to produce lists for verification by consultants and to create check lists without disrupting alphabetical order.
- Phase 2: Conversion to Machine-Readable Form: This phase involves adding textual material and converting the entire corpus into a machine-readable format. The paper discusses the development of "software fiches" designed for efficient keypunching and comprehension, and explores solutions for handling accented words and homographs in sorting and printing. Two solutions are presented:
- LEXAUTOM approach: Representing the entry word twice in separate fields, one without accents and one with special characters for internal representation.
- "Code" field approach: A more efficient method where a "code" field following the "entry" field uses special characters to indicate how letters should be treated, avoiding duplication and saving storage space.
- Phase 3: Feasibility Study of Automatic Typesetting: This final phase investigates the use of automatic typesetting procedures for the final published version of the dictionary.
The paper also touches upon the development of a variable-length format for data files to address the issue of storage waste associated with fixed-length formats, drawing inspiration from the "catalogue" system used for the Random House Dictionary.
Datasets
Not discussed in the paper. The paper focuses on the methodology and challenges of creating a bilingual dictionary rather than evaluating specific datasets.
Key Insights and Ideas
- The inherent complexity of bilingual dictionaries: They require not only linguistic accuracy but also a deep understanding of semantic equivalence and the potential pitfalls of "faux amis."
- The limitations of manual lexicography: Traditional methods are slow, labor-intensive, and lead to rapid obsolescence of dictionary content.
- The potential of computer-aided editing: Automation can significantly accelerate editorial tasks, improve accuracy, and enable more frequent updates.
- The challenge of representing linguistic nuances in digital formats: Handling accents, capitalization, and homographs requires careful design of data structures and coding schemes.
- The need for flexible data formats: Variable-length formats are crucial for efficient storage and management of diverse dictionary entries.
- The "catalogue" system as a model: This system, adapted from monolingual dictionary production, offers a promising structure for managing complex bilingual lexicographical data.
Key Results
The paper does not present quantitative results in terms of dictionary accuracy or speedup metrics. Instead, it focuses on the conceptualization and proposed methodology for computer-aided bilingual lexicography. The key "results" are the identification of specific challenges and the proposal of solutions, such as the "code" field approach for handling linguistic variations and the adoption of a variable-length format inspired by the Rand Corporation's "catalogue" system. The paper suggests that these methods, when implemented, will lead to more efficient editorial processes and a more up-to-date final product.
Limitations and Future Work
The authors acknowledge several limitations and suggest future directions:
- Limited Corpus in Early Stages: The "code" field solution was initially devised for a limited corpus of 100 words, and its scalability to a corpus of 100,000 words is noted as a potential challenge, though the paper implies it is manageable.
- Storage Waste with Fixed-Length Formats: The paper highlights the inefficiency of fixed-length formats for variable-sized data and proposes variable-length formats as a solution.
- Complexity of Data Conversion: The conversion of complex textual data to machine-readable form is a significant undertaking.
- Future Work: The paper explicitly mentions the ongoing study of variability by programmers at the University of Victoria and IBM (Canada) Ltd. to develop a variable-length format. It also points to the third phase of the project, which is a feasibility study into automatic typesetting procedures for the final published version. The overarching future goal is to enable the rapid production of revised editions, reducing the time from decades to months or weeks, and ensuring a proliferation of up-to-date consultative material.
Summary updated