The Speech-to-Song-Illusion revisited (Simone Falk, Tamara Rathcke)

We are exploring the boundaries of speech and song in an
acoustic-perceptive perspective.  We investigate a musical illusion
first described by Diana Deutsch (1995, Deutsch et al. 2008). In this
"speech-to-song illusion" a phrase read by Diana Deutsch shifts to be
heard as sung without changing any acoustic characteristics of the
signal. This illusion is achieved by simply repeating the phrase
several times in exactly the same way. As far as we know, the effect
was found only with this single phrase as piece of evidence. We assume
that some acoustic characteristics of this read phrase supported the
perceptive shifting from speech to song in a specific way. Our
assumption is that the shifting will occur earlier, i.e. after fewer
repetitions, when specific acoustic characteristics are present in a
sound signal. We have set up a reaction time experiment using the
speech-to-song illusion as a method to test several hypotheses about
the nature of the acoustic characteristics that will support the
perceptive drift. Our hypotheses are as follows: We suggest that the
drift will occur earlier if (i) pitch accents occur isochronically
(ii) if intervocalic intervals are equalized in duration resembling
note durations, (iii) the targets of pitch accents are level-like (as
opposed to contour-like targets as in natural speech) and (iv) if
scalar tonal relations are suggested by the melody especially at the
phrase boundary. In the experiment, we manipulate these factors (on
the basis of natural spoken sentences) individually and cumulatively
to assess their relative importance for the perceptual shift. This
study is still in progress. It will give useful evidence for modelling
the acoustic basis of speech and song perception and will reveal
shared ressources.