Vowels in consonantal context are perceived more linguistically than are isolated vowels: Evidence from an individual differences scaling study

Categorial discrimination of vowels produced in syllable context and in isolation

“…Target theories that assume vowels are cued exclusively by information present in a single spectral cross section (e.g., Joos, 1948;Ladefoged, 1967) are inadequate to account for the results of many of the vowel-identification studies and Rakerd's (1984) experiments. If the information for vowel identity is the target, then the isolated vowels would provide clearer information for vowel perception than would vowels in the context of consonants.…”

“…Indeed, Assmann, Nearey, and Hogan (1982) found that programs to differentiate vowels acoustically were most successful when spectral information from more than one spectral cross section was utilized. The precise nature of the important acoustic parameters that contribute to the more "linguistic" nature of coarticulated vowels has yet to be delineated, although it is clear that relative duration is very important (e.g., Jenkins, Strange, & Edman, 1983;Peterson & Lehiste, 1960;Rakerd, 1984;Strange et al, 1979;Strange et al, 1983). A detailed specification of the important acoustic parameters in vowel identification also depends in part on the discovery of more adequate acoustic descriptions ofthe dynamic information that is present in the transition portion of the syllables (see Strange et al, 1983 …”

“…This study suggests that there are genuine differences in the information available for the identification of isolated and coarticulated vowels in English. A study by Rakerd (1984) showed that the similarity judgments of American English listeners may be more consistently scaled (using an individual-differences scaling procedure) according to linguistic dimensions when the stimuli are coarticulated vowels rather than isolated vowels. He found that this difference between coarticulated and isolated vowels was not based on previously acquired knowledge, but was attributable to the acoustic characteristics of the vowel stimuli themselves.…”

A multidimensional scaling analysis of vowel discrimination in humans and monkeys

“…The rhotic vowel131 is unique in having a very low F3 and is produced by "bunching" the tongue so that two constrictions, one palatal and one pharyngeal, are produced in the supraglottal vocal tract (Delattre & Freeman, 1967). • a order to correspond with the phonetic dimensions as traditionally portrayed in the "vowel quadrilateral" (see, e.g., Rakerd, 1984;Rakerd & Verbrugge, 1985): The high front vowels are in the upper left, the low central vowels are at the bottom center, and the high back vowels are in the upper right. Figure Procedure All subjects were tested with a minimal-uncertainty repeatingbackground same-different AX discrimination procedure that has been extensively used over the last 20 years in studies of animal psychoacoustics and speech discrimination (Sinnott, 1994(Sinnott, , 1995.…”

“…Multidimensional scaling (MDS) has been in use for over 25 years to study the psychological space of human vowel systems (see, e.g., Fox, Flege, & Munro, 1995;Pols, van der Kamp, & Plomp, 1969;Rakerd, 1984;Rakerd & Verbrugge, 1985;Shepard, 1972;Singh & Woods, 1970). MDS is a mathematical tool that enables us to represent the similarities of perceptual objects spatially, as in a map (Schiffman, Reynolds, & Young, 1981).…”