Analysing the deep structure of queries: Transfer effect on learning a query language

Abstract This experiment was designed to investigate the impact on performance, while learning a query language, of specific instructions pertaining to the ‘deep-structure’ of the queries. The instructions related to deep-structure comprised training in analysing queries in natural language in terms of logical constituents. The effect of such instructions was investigated for two different query languages. One was a textual and one was a ‘graphic’ query language. The results suggested that the usefulness of explicit training in analysing queries in terms of logical deep-structure is dependent on the semantic constraints in the query language. The textual query language, which had keywords and a syntactical template such as GET_WHERE_EQ_ assumedly induced analysis of the query regardless of pretraining, while the graphic used in this study tended to encourage a trial-and-error approach, especially when no prior training in analysing queries was given.