A Linguistic Approach to the Design of a Language for Computational Linguistics

V. AndrewshtshenkoCOLING 1982

TL;DR

This paper proposes LICOL, a novel programming language designed for computational linguistics. It leverages a semiotic system approach, treating the language as a tool for man-machine communication in natural language processing and linguistic research, aiming to unify diverse computational needs.

What Problem Does This Paper Address?

The paper addresses the challenge of creating a unified language for computational linguistics, a field bridging linguistics and computer science. This field requires tools for both natural language data processing (like machine translation) and automation of linguistic research. Existing approaches often lack the richness to satisfy the diverse computational conditions arising in these applications. The authors aim to design a language that can serve as a powerful interface for man-machine communication within this domain.

Methods

The core of the paper's approach is viewing LICOL as a semiotic system. This means the language's units are formed based on R. Jakobson's linguo-semiotic functions. LICOL is designed for communication between humans and computers in situations involving automatic and automatized natural language data processing.

Key methodological aspects include:

  • Function-Argument Form: Relations between concepts are expressed using a function-argument structure.
  • Reversed Polish Notation: The language employs a reversed Polish form for its notation, which is convenient for representing tree structures linearly.
  • Computational Constructions (CCs): The fundamental concept in LICOL's "world" is the Computational Construction (CC). These are the building blocks for both programs and data.
  • Multibase Classification of CCs: CCs are classified based on several criteria:
  • Signifiants: Representing (names, pictures) and Processing (controllers, operations).
  • Signifiés: Real (external data), Virtual (user notions of processing), and Notational (constitutive parts of signifiants).
  • Value Form: Scalars, Vectors, and Lists.

This classification allows for flexible operations on various data types and structures, including sequences, sets, and graphs.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Semiotic System as Foundation: Treating a programming language as a semiotic system, drawing parallels to natural language functions, is a novel approach to designing languages for computational linguistics.
  • Unifying Programs and Data: The paper suggests that by representing programs and data as IC trees (which can be linearized using Polish notation), programs can be interpreted as data and vice versa. This blurs the lines between data representation and procedural knowledge.
  • Computational Constructions (CCs) as Universal Units: The concept of CCs as the fundamental building blocks for both programs and data provides a unified framework for computational linguistics tasks.
  • Flexibility through Multibase Classification: The multi-faceted classification of CCs allows for a highly flexible and expressive language capable of handling diverse computational needs.
  • Bridging Natural and Artificial Languages: The design aims to facilitate a smoother transition from natural language expressions to LICOL, leveraging the metalanguage function of natural language.

Key Results

The paper presents a conceptual design for the LICOL language. It outlines its theoretical underpinnings, structural components, and intended applications. Specific quantitative results or performance benchmarks are not provided, as the paper focuses on the linguistic and semiotic principles guiding the language's design. The key "result" is the proposed architecture and philosophy of LICOL itself.

Limitations and Future Work

The paper acknowledges that it is an "inevitably short review." It presents the theoretical framework and conceptual design of LICOL.

Limitations:

  • The paper does not provide concrete examples of LICOL code or demonstrate its application in solving specific computational linguistics problems.
  • Performance metrics, efficiency, and practical implementation details are not discussed.

Future Work:

The paper implies that further development and practical implementation of LICOL are necessary. Future work would likely involve:

  • Developing a full-fledged compiler or interpreter for LICOL.
  • Creating comprehensive documentation and tutorials.
  • Testing LICOL on a range of computational linguistics tasks to validate its effectiveness and compare it with existing tools.
  • Exploring the full potential of the semiotic approach in language design for AI and NLP.

Summary updated

BibTeX
@inproceedings{andrewshtshenko1982linguistic,
  author = {V. Andrewshtshenko},
  title = {{A Linguistic Approach to the Design of a Language for Computational Linguistics}},
  booktitle = {International Conference on Computational Linguistics},
  year = {1982},
}

Loading related content...

Figure