Speech Perception 1

Which is NOT an example of Bottom-up Processing?

• when we use lexical information to help in noisy environments
• when we use redundancy in the signal - coarticulation helps
• when we perceve categories of sounds (phonemes), not allophones
• when we perceive ratios of vowel formants, not absolutes

What is the Ganong Effect?

• given a speech continuum like gift-kift: a category boundary (k or g?) will shift toward the real word
• given conflicting auditory & visual speech information: our brains take the visual info into account when trying to parse the signal
• when sounds missing from a speech signal can be restored by the brain and may appear to be heard
• when we perceive ratios of vowel formants, not absolutes

What are the articulatory features of consonants?

• place, manner, voicing
• vertical/horizontal position of the tongue
• nasality & frication
• degree of liquidness & affrication

What are allophones of a phoneme?

• versions of speech sounds that are predictable by their environment
• the smallest unit of speech that changes meaning of a word - an abstract representation
• the smallest categories of speech sounds that we can perceive
• the length of time between the release of a stop & the onset of the following vowel

What is coarticulation?

• the influence of one phoneme on another
• markings that provide very precise articulatory information
• dark bands on spectrograms that reflect the energy of the sound
• versions of speech sounds that are predictable by their environment

Why can babies less than a year old perceive phonemes from foreign languages that adults can't?

• They haven't developed phoneme categories for their native tongue(s) yet.
• They haven't developed vowel formant ratios for their native tongue(s) yet.
• They haven't figured out coarticulation yet.
• They haven't figured out redundancy in the signal yet.

What is Top-Down Processing?

• previously stored information (syntax, context, etc.) helps you parse what you hear
• brand-new, raw, auditory information gets processed & words are built
• when a component is duplicated so if it fails there will be a backup
• when we ask whether 2 sounds are the same or different

What can we see with speech spectrograms?

• time, sound frequencies, & formants
• place, manner, voicing
• vertical/horizontal position of the tongue
• light frequencies & absorption lines

What does “Speech Perception is multi-modal.” mean?

• we can use both auditory & visual information when we're parsing speech
• we can use both auditory & olfactory information when we're parsing speech
• we can use both auditory & extra-sensory perception when we're parsing speech
• we can perceive both phonemes and allophones in the speech stream

Everyone has a different vocal tract & therefore no one's vowels are exactly alike. How do we figure out what vowels are being said?

• by perceiving ratios of vowel formants
• by perceiving them categorically
• by perceiving their diacritic markings
• by restoring their phonemes