Speech Graph Analysis in REM and non-REM Dream Reports
Speech Graph Analysis; dreaming; dream report; non-REM sleep; REM sleep.
Despite a growing body of literature, it is still unclear to what extent mentation recorded immediately after Rapid Eye Movement (REM) sleep is different to that recorded after non-REM (NREM). One way to explore potential disparities has been to investigate the linguistic differences in dream reports across the respective REM and NREM sleep stages. A particularly promising method, yet to be applied in this regard, is the analysis of speech graph attributes (SGA). Such a technique has been able to distinguish verbal reports of schizophrenics, and is particularly revealing of psychosis when analysing dream reports. The present study aims to use SGA to investigate possible structural differences in dream reports from the three major sleep stages. In this regard, 22 healthy participants slept in a laboratory and were awoken during N2, N3, and REM. Upon awakening, participants were asked to report any dream content remembered and then to answer subjective rating scales relating to their dream content. The final sample for analysis comprised 161 dream reports (N2 – 82, N3 – 33, REM – 46). The initial results indicate that graphs from REM dream reports are larger and more connected than those from N3. They also suggest that N2 graphs represent an intermediate form between REM and N3 – smaller and less connected than REM graphs, but larger and more connected than N3 ones. A correlation analysis demonstrated a significant relationship between SGA attributes and subjective ratings of dream intensity, dream recall, auditory perception and emotionality. Altogether, these results indicate a complex relationship between the graph structures of dream reports and the sleep stage from which they are obtained. Future analyses incorporating more detailed sleep measures, as well as complementary semantic based techniques can provide a more complete picture of the present findings.