Background: Speech recognition in noise testing has been conducted at least since the 1940s (Dickson
et al, 1946). The ability to recognize speech in noise is a distinct function of the
auditory system (Plomp, 1978). According to Kochkin (2002), difficulty recognizing
speech in noise is the primary complaint of hearing aid users. However, speech recognition
in noise testing has not found widespread use in the field of audiology (Mueller,
2003; Strom, 2003; Tannenbaum and Rosenfeld, 1996). The audiogram has been used as
the “gold standard” for hearing ability. However, the audiogram is a poor indicator
of speech recognition in noise ability.
Purpose: This study investigates the relationship between pure-tone thresholds, the articulation
index, and the ability to recognize speech in quiet and in noise.
Research Design: Pure-tone thresholds were measured for audiometric frequencies 250–6000 Hz. Pure-tone
threshold groups were created. These included a normal threshold group and slight,
mild, severe, and profound high-frequency pure-tone threshold groups. Speech recognition
thresholds in quiet and in noise were obtained using the Hearing in Noise Test (HINT)
(Nilsson et al, 1994; Vermiglio, 2008). The articulation index was determined by using
Pavlovic's method with pure-tone thresholds (Pavlovic, 1989, 1991).
Study Sample: Two hundred seventy-eight participants were tested. All participants were native
speakers of American English. Sixty-three of the original participants were removed
in order to create groups of participants with normal low-frequency pure-tone thresholds
and relatively symmetrical high-frequency pure-tone threshold groups. The final set
of 215 participants had a mean age of 33 yr with a range of 17–59 yr.
Data Collection and Analysis: Pure-tone threshold data were collected using the Hughson-Weslake procedure. Speech
recognition data were collected using a Windows-based HINT software system. Statistical
analyses were conducted using descriptive, correlational, and multivariate analysis
of covariance (MANCOVA) statistics.
Results: The MANCOVA analysis (where the effect of age was statistically removed) indicated
that there were no significant differences in HINT performances between groups of
participants with normal audiograms and those groups with slight, mild, moderate,
or severe high-frequency hearing losses. With all of the data combined across groups,
correlational analyses revealed significant correlations between pure-tone averages
and speech recognition in quiet performance. Nonsignificant or significant but weak
correlations were found between pure-tone averages and HINT thresholds.
Conclusions: The ability to recognize speech in steady-state noise cannot be predicted from the
audiogram. A new classification scheme of hearing impairment based on the audiogram
and the speech reception in noise thresholds, as measured with the HINT, may be useful
for the characterization of the hearing ability in the global sense. This classification
scheme is consistent with Plomp's two aspects of hearing ability (Plomp, 1978).
Key Words
Articulation index - attenuation - distortion - Hearing in Noise Test (HINT) - pure-tone
average - speech recognition in noise - speech recognition in quiet
