Ling 250 part 2 Flashcards
what is a phoneme
Perceptually distinct units of sound that distinguish one word
from another in a given language
what is an allophone
phonetic variations of
a phoneme, appear in predictable contexts
how do you argue that sounds are 2 different phonemes
lexical contrast (minimal pairs), overlapping distribution
how do you argue that 2 sounds are 2 allophones of the same phoneme
distribution analysis: what context does the sound appear in, if they are in complementary distribution then they are allophones
define minimal pair
pair of words with different meanings and just one sound difference
complementary distribution
no overlapping contexts where a set of sounds appear
overlapping distribution
minimal pairs exist (2 sounds can appear in the same context)
underlying representation
the phoneme with the most simple/broad contexts
assimilation
make 2 sounds sound more similar
dissimilation
make sounds sound less similar
epenthesis
add a sound (insertion)
deletion
delete a sound
metathesis
reordering of sounds
fortition
sounds get stronger
lenition
sounds get weaker
morphemes
meaningful pieces of a word (prefix, stem, suffix)
allomorphs
morpheme variants, same meaning, different forms depending on the context
describe the sonority hierarchy
low-high: voiceless stops, voiced stops, voiceless fricatives, voiced fricatives, nasals, liquids, glides, vowels
describe the components of a syllable
onset, rhyme (nucleus, coda)
what does the mora signify
syllable weight
light syllable
one mora
heavy syllable
2 moras
superheavy syllable
3+ moras
what sounds get a mora
vowels -1, diphthongs- 2, stressed syllables with a coda - 1
onset
first sound(s) in a syllable
nucleus
vowel
coda
comes after the nucleus
rhyme
made up of the nucleus and coda
sonority sequencing principle
the nucleus of a syllable is the sonority peak, usually progressively decreases from the nucleus on both sides (upside down V shape), some violations
tone system
the use of pitch for lexical contrast
intonation system
the use of pitch to convey a discourse-level meaning
tone shift
tone can spread from its segmental anchor and shift to other segments and syllables
tone spread
a single tone feature may spread out over a string of successive syllables
obligatory tone contour principle
identical tones adjacent to each other are prohibited in some languages
tone sandhi
observed in East Asian tone languages, phonological rules that change tone assigned to a syllable depending on tone of adjacent syllable
boundary tone
occur at the right edge of a phrase and mark the type of speech act
pitch accent
encode the informational status of a word or phrase (highlights new information, broad or contrastive focus)
african languages
level tone languages, relative pitch heights
asian languages
contour tone languages
distributional analysis of language data
determines if 2 sounds are 2 different phonemes or allophones of the same phoneme
steps for a distributional analysis
look at what contexts each sound appears in
format for writing phonological rules
/x/ –> [y] / [context])
response patterns in identification tasks
continuous perception is linear, categorical is non-linear (_/-)
response patterns in discrimination tasks
continuous is a flat line around 50, categorical peaks between 4 and 5 but flat around
continuous perception
gradient
catagorical speech
one or the other
origins of perceptual catagorization
temporal response patterns in the auditory cortex (auditory system), and learning!
steps in generating speech from text
breaks a word down into pronouncable units, transformed into electronic sound signals by several methods (parametric, concatenative, generative)
parametric speech synthesis
define parameters of human speech signal and use this information to generate speech (amplitude and freq of formants)
synthesis by rule
computer synthesized sound wave where only f1/f2/f3 frequency and f1/f2/f3 amplitude are manipulated
articulatory synthesis
model of the vocal tract, speech is created by digitally simulating air flow through the vt
concatenative speech synthesis
synthesize sounds by concatenating short samples of pre-recorded sound
steps in automatic speech recognition
speech input, spectral encoding, look-up, output
word error rate
number of errors divided by the number of words
challenges of automatic speech synthesis and recognition
segmentation (don’t know word boundaries), ambiguity (homophones, semantics, pragmatics, syntax), variability (across speakers, speech styles, accents, etc.)
identification experiment
what category does a sound belong to, do we find a clear boundary?
discrimination experiment
can you discriminate one sound from another
