Newborns are sensitive to multiple cues for word segmentation in continuous speech.

Area of Science:

• Developmental Psychology
• Neuroscience
• Linguistics

Background:

• Infants must segment continuous speech to learn words, a challenge due to the lack of clear speech boundaries.
• While older infants show word segmentation skills, it's unclear if these abilities are present from birth or develop later.

Purpose of the Study:

• To investigate whether newborns possess the ability to segment words from continuous speech.
• To examine the role of statistical co-occurrences and prosody in early speech segmentation.

Main Methods:

• Functional near-infrared spectroscopy (fNIRS) was used to measure brain responses in neonates.
• Two experiments familiarized newborns with continuous speech for approximately 3 minutes.
• A recognition test assessed brain responses to words that either adhered to or violated learned statistical or prosodic boundaries.

Main Results:

• Newborns exhibited differential brain responses to words violating statistical boundaries (Experiment 1).
• Newborns also showed distinct brain responses to words violating prosodic boundaries (Experiment 2).
• Word recognition occurred even without prosodic cues, indicating phonological encoding independent of prosody.

Conclusions:

• Human infants are born with operational language processing and memory capacities.
• Newborns can utilize both statistical and prosodic cues to segment continuous speech.
• These innate abilities represent a fundamental first step in language acquisition.