Language and Phonemes

Psy315h - Class 23

(This page last updated 25 November, 2005.)
  

Hierarchy of Linguistic Units

• phonemes
• morphemes
• words
• phrases
• sentences

Phonology and the Vocal Tract

• vocal cords in larynx
  • vibration of the vocal cords provides voicing for vowels and some consonants
• vowels
  • Vowel sounds are made by allowing air to pass freely through the vocal tract while the vocal cords are vibrating.
  • Different vowel sounds are produced by changing the shape of the vocal tract.
• consonants
  • Consonants are produced by restricting the air flow at some point along the vocal tract.
  • Three factors are varied to produce the different consonants.
    • Voicing
      • unvoiced: p, f, t
      • voiced: b, v, d
    • Location of constriction ("Place of Articulation")
      • Where along the visual tract is the air flow constricted to produce the consonant?
        • Bilabial
        • Labiodental
        • Alveolar
        • etc.
    • Nature of constriction ("Manner of Production")
      • stops
        • Air flow is completely blocked for a brief time
        • p, b, f, v, t, d
      • fricatives
        • Small opening is maintained for air flow
        • s, z
      • etc.

A Subset of English Consonants

	bilabial	labiodental	alveolar
voiced stops	b	v	d
unvoiced stops	p	f	t
voiced fricatives			z
unvoiced fricatives			s

Categorical Perception of Voiced and Unvoiced Consonants

 What is the difference between "pa" and "ba".

• For both, the vocal cords must be vibrating to produce the vowel.
• Both begin with a bilabial stop.
• For "ba", the cords start to vibrate at the very beginning of the word, during the consonant.
• For "pa", the vibration is delayed so that the consonant is unvoiced.
• The vocal tract is capable of producing a wide range of VOT's.
  • For voiced consonants, as in "ba", the VOT is 0 milliseconds.
  • For unvoiced consonants, as in "pa", the VOT is about 60 milliseconds.
• The VOT can vary continuously between these values. However, we tend to perceive all these different physical sound patterns as being from one of two categories.
  • Voiced: "b"
  • Unvoiced: "p"

Experiment in Perception of Phonemes

• Synthetically produce bilabial stops with VOT varying across a wide range of values.
• Subjects listen to each sound and categorize it as "p" or "b".
• As VOT increases, subjects continue to hear "b".
• Then, at some intermediate VOT, the perception suddenly shifts to "p".
• Discriminations across categories are easier than discriminations within a category.
  • Subjects have difficulty distinguishing between a 10 msec VOT and a 30 msec VOT.
  • They both sound like "b".
  • Subjects have difficulty distinguishing between a 40 msec VOT and a 60 msec VOT.
  • They both sound like "p".
  • Subjects can easily distinguish between a 20 msec VOT and a 40 msec VOT.
  • Physically, the difference in sound for all these comparisons is similar (20 msec), but the difference is much easier to detect when it crosses a category boundary.
• Our language perception system takes physical values such as VOT, that vary smoothly, and organizes them into categories.

   

Segmenting Phonemes

• Perceiving phonemes is complicated by the fact that there are usually no gaps between phonemes, even when they are parts of different words.
• Our perceptual system must figure out where the boundaries belong.
• Separating phonemes from one another is complicated by coarticulation.
  • While one phoneme is being produced, the vocal tract will begin to change in order to produce the next phoneme.
  • The physical sound of a given phoneme will vary depending on what phonemes come before and after it.
    • Yet we generally do not perceive any difference.

   

Rules Governing the Combination of Phonemes into Words

Example: For regulars plurals (adding "s" sound), the added phoneme will be voiced or unvoiced depending onthe voicing of the final phoneme of the word.

next class: Syntax

Psych 315H: Cognitive Psychology Kyle Cave Psychology Dept. U. Mass.