Fairbanks, Grant. Experimental Phonetics – T06

Auditory Comprehension
Of Repeated High-speed Messages *1

Grant Fairbanks
Newman Guttman
Murray S. Miron **2

The experiment here reported is related
to a general investigation of the
effects of time compression upon the
comprehension of connected spoken
messages (2). In the basic experiment
two relatively long technical messages
were read by an experienced speaker
at a representative rate. These original
readings were recorded, compressed
in time by selected amounts, and presented
to groups of subjects. Comprehension
of factual details was tested.
The technique for compression was
automatic and operated upon the original
recording only in the time dimension
(1), so that the various versions
of the message differed in total listening
time or, reciprocally, in listening
rate. Variation or comprehension was
significant.

It was found that listeners could
tolerate substantial amounts of compression
before reduction of comprehension
became excessive. For example,
subjects who heard the messages
with 50% compression at a rate
of 282 words per minute yielded a
mean score which was approximately
90% of that for subjects who heard
the original versions at 141 wpm. If
the amount of comprehension of a
double-rate message adequately satisfied
the requirements of a given communication
situation, then 50% of the
original message time obviously becomes
available for double-rate presentation
of a second message which
had an original length equal to that of
the first message. If it should be desired,
however, to increase the comprehension
of one message, the finding
suggests that this might be done,
without exceeding the original speaking
rime for the message, by means of
a special arrangement; this would consist
53of a compressed version of the
message plus reinforcing material
added in some manner within the time
‘saved.’ It is with such a proposition
that the present experiment was concerned.

Among various arrangements the
case chosen for experimentation was
double presentation of the complete
message. The most important of the
reasons for making this choice was
that it was considered possible that
the comprehension (not the intelligibility)
of extended material presented
at fast rate might be so close to ceiling
at the rate in question that it might
not be susceptible to improvement by
additional stimulation at fast rate. Another
factor in selection was that complete
verbatim repetition of a recording
avoids all questions of content and
speaking effectiveness, while use of the
complete message as the unit of repetition
is a limiting case of ‘dovetailed’
repetition versions (e.g., word-word,
sentence-sentence). It should also be
noted that complete double presentation
in this form would not be thought
to be a powerful method of increasing
comprehension and constitutes a
strong test of the basic proposition for
that reason.

Procedure

The design of the experiment envisioned
an analysis of variance of the
mixed factorial type, Number of Presentations
by Compression by Stanine
by MEL, following the pattern of the
main study, although only the first
two and their first-order interaction
were of particular concern.

Results

Table 1 shows mean scores expressed
in percentage correct. The
means in the row designated Total,
each based on the scores of 36 subjects
on the 60-item test, are of main interest.
The third value, that for uncompressed
single presentation in a relative
message time T, serves as a reference
to which the other means may be
compared. It will be observed that the
effect of double presentation was in
the expected direction at both compressions,
and that in the two conditions
where stimulus time equalled T,
single 0% and double 50%, the latter
had a small advantage.

The results of the analysis of variance
were straight-forward and will
not be presented in detail. The effect
of number of presentations, tested
with the between-subjects error, was
found to be significant; F was 6.98,
with 1 and 128 degrees of freedom.
Neither compression nor the interaction
of number of presentations and
compression was significant. As would
be expected from the basic experiment,
the effects of stanine and MEL
were both significant. All tests of the
numerous interactions failed. Since the
evidence supports and at no point controverts
the general proposition advanced
above, it is concluded that such
a possibility for increasing auditory
comprehension of a given content in
a given rime merits further study.

References

1. Fairbanks, G., Everitt, W. L., and
Jaeger, R. P. Method for time or
frequency compression-expansion of
speech. Trans. I.R.E.-P.G.A. 1954, AU-2,
7-11.

2. Fairbanks, G., Guttman, N., and
Miron, M. S. Effects of time compression
upon the comprehension of connected
speech. JSHD, 22, 1957, 10-19.55