Infants show keen ear for speech sounds.

Infants show keen ear for speech sounds

Six-month-olds may not utter an articulate word, but a new study indicates they already organize adult vowel sounds into distinct categories and perceive some sounds as better examples of a particular vowel than others. Specific speech sounds apparently serve as "perceptual anchors" from infancy onward, crucially influencing the ability to speak and understand language, say psychologists Di-Anne Grieser and Patricia K. Kuhl of the University of Washington in Seattle.

Grieser and Kuhl base their work on the notion that human perception -- the operation of senses such as seeing and hearing -- sorts diverse incoming stimuli into meaningful categories. Certain stimuli, according to the theory, constitute the best instances, or prototypes, of their particular category. When new stimuli are encountered, the brain assigns them to a perceptual category based on how closely they resemble a category's prototype.

For example, people in a broad array of cultures agree on color prototypes -- say, the hue that best represents the color green. These preferences are thought to be inherently defined by the visual system.

Over the past decade, researchers have found that adults group the smallest units of language -- consonant and vowel sounds -- into acoustic categories and that they perceive some sounds as better examples of a category than others.

Reports also indicate that 6-month-old infants can distinguish between prototypical vowel sounds made by adults and recognize novel variations of the sounds. Grieser and Kuhl, who describe their study in the July DEVELOPMENTAL PSYCHOLOGY, examined how infants accomplish this complex feat.

In their first experiment, the researchers conditioned 16 infants to turn their heads toward a loudspeaker when the prototype for one vowel category (the long "e" in the word "peep") changed to another vowel prototype (the short "e" in the word "pep"). Correct responses caused a box on the loudspeaker to light up and reveal a mechanical animal, such as a bear with a drum. A computer synthesizer created the vowel prototypes and 32 acoustic variations of each sound. The new sounds varied in the degree to which they conformed to the prototypes.

Infants heard eight novel sounds from each of the two categories. They recognized the sounds as belonging to the appropriate category, as indicated by head turns, more than 90 percent of time.

In a second experiment, Grieser and Kuhl tested a new group of 32 infants, all 6 months old. Half the group learned to turn their heads in response to the prototype for the long "e" sound; the rest were conditioned to turn their heads in response to a poor representative of the same vowel category. The researchers generated a set of 32 acoustic variations, 16 around the good "e" and 16 around the poor "e."d Infants heard all 16 variations of the sound on which they were trained.

Recognition of novel sounds in the same sound category, indicated by lack of head turning, proved significantly greater among those trained on the prototype. This finding offers the first support for the idea that infants organize speech sounds around what adults perceive as good examples of units of speech, the investigators maintain.

It is not known whether vowel "goodness" is inherently defined by the nature of the human hearing system or is the result of considerable experience, even among 6-month-olds, in listening to English sound patterns.

The researchers are now testing U.S. infants on non-English vowel variations to see if training on unfamiliar prototype sounds produces similar effects. If so, they say, speech-sound prototypes may be an inherent characteristic of human hearing.