What is it about?
This is a proposal for a model of the kinds of semantic relation that exist between entities. It draws primarily on the linguistic domain of binominal lexemes (noun-noun compounds and their functional equivalents), as in “bird song” -- a song that is produced by a bird -- which is an instance of the PRODUCTION relation. In addition, it draws on metonymic patterns of the type “I’m reading Shakespeare”, an instance of the PRODUCER FOR PRODUCT metonymy. Further refinements come through the integration of conceptual relations typically found in polysemy, which can be traced back to Aristotle's three principles of remembering: similarity, contrast and contiguity. The PHAB model is also informed by models of knowledge representation and thus has applicability outside of the domain of natural language.
Featured Image
Photo by Adarsh Kummur on Unsplash
Why is it important?
This is the first attempt to create a general, reusable model covering multiple domains of linguistics and extending beyond natural language to knowledge representation. As such it shows how linguistics can inform the design of artificial intelligence systems, as well as providing an out-of-the-box classification for use in linguistics research.
Perspectives
For me personally this model represents a synthesis of my earlier work on Topic Maps and my more recent research into the semantics of noun-noun compounds (and their functional equivalents).
Steve Pepper
Universitetet i Oslo
This paper connects the modification relations in binominal compounds and other lexical units with two underlying "thing"-concepts and other types of relations between concepts, such as metonymy. It contributes to the understanding of lexical semantics and, with its two-tier classification of relations, has practical applications for research.
Pierre J.L. Arnaud
Read the Original
This page is a summary of: Absolutely PHAB, The Mental Lexicon, October 2020, John Benjamins,
DOI: 10.1075/ml.00016.pep.
You can read the full text:
Resources
Contributors
The following have contributed to this page
