Freud's explicit recognition in the Traumdeutung of the existence of abundant verbal material in dreams, and his implicit recognition of the high degree of grammaticality which characterizes it, seems to have posed a theoretical problem for his overall dream model: How can an apparently logico-grammatical (=secondary) process be operating effectively within the framework of primary-process dreaming? (Or, in modern Chomskyan terms: How can the strictly-defined formal transformational rules for sentence generation be applied so correctly and consistently within the dream-work, whose 'rules' for the generation of the manifest dream from the dream-thoughts are notoriously ill-defined and haphazardly applied?)

Freud sought to resolve this apparent anomaly by degrading (what is today called) 'linguistic competence' in dreams by denying any novel aspect to dream-speeches, viewing them rather as 'replays' of utterances heard or spoken previously in waking life and 'recorded' in memory.