Generating from a Deep Structure

: Noncanonical semantic representations are representations which cannot be derived by some grammar G although they are semanticMly equivalent to representations which can be derived by G. This paper presents a generation Mgorithm which deals with noncanonical input. The proposed approach also enhances portability and language independence in that (i) linguistic decisions made by independent modules (e.g., planner, transfer component) can be communicated to the generater in a naturM way and (ii) the same algorithm coupled with different grammars will yield sentences in the corresponding languages.

• Type: Conference
• Year: 1990
• Where: COLING
• Authors: Claire Gardent, Agnès Plainfossé
• Keywords: COLING 1990, COLING 2008, Generation Mgorithm, Noncanonical Input, Noncanonical Semantic Representations