Class Notes (1,100,000)
CA (620,000)
UTSC (30,000)
LINA01H3 (200)
Lecture

LINA01H3 Lecture Notes - George Bernard Shaw, Speech Recognition, Vocal Folds


Department
Linguistics
Course Code
LINA01H3
Professor
Chandan Narayan

This preview shows pages 1-3. to view the full 20 pages of the document.
PHONETICS: THE SOUNDS OF LANGUAGE
[1]
TO USE LANGUAGE
- do not need to speak to use it
- can be
- written
- manually signed
- mechanically reprod'ed
- syn'zed by computers w/ some success
SPEECH
- still is main way that humans express themselves via lang
- it was what was used by our species before ability to write arose
- anatomical specialization for speech gives us idea of long history behind human speech
- there exists neural mchsms in humans for peception of speech sounds
PHONETICS
[D] = branch of linguistics that examines the complete list of sounds, and the structures of sounds of speech
[2]
[D] PHONES (aka [D] SPEECH SOUNDS)
= any sound used in human lang
- finite list of them
- portion of entire set found in any human lang
- any human, child or adult can learn to prod. any human speech sound
- about 600 consonants and 200 vowels
- sounds prod'ed using vocal cord may be ones not present in speech
- ex. sound made by sticking tongue out and blowing hard across it
[3]
TWO APPROACHES TO PHONETICS
ARTICULATORY PHONETICS (emphasis of this
chap)
ACOUSTIC PHONETICS
[D] = studies physiol. mcshms of speech prod'ion [D] = focus on measuring & analyzing phys prop's of
sound waves prod'ed when we speak
- both are needed to understand speech
p15
2.1 PHONETIC TRANSCRIPTION
[1]
[D] INTERNATIONAL PHONETIC ALPHABET (IPA) = best-known universal sys. known so far for transcribing
sounds of speech
- dvping since 1888
=> still additions are being made to it
- attempts to rep. each sound of human speech w/ single symbol
- enclosed in [ ] to indicate that transcription is phonetic, and does not rep. spelling sys. of any specific lang

Only pages 1-3 are available for preview. Some parts have been intentionally blurred.

(ex) [ð]
= eth (as in weather)
- sound spelled th in English this is transcribed as above symbol
- this symbol is used by IPA to rep. sound in whichever lang it is heard
===[LANGUAGE MATTERS: SOUNDS AND SPELLING]====
[1]
- diff. b/ween IPA and writing sys of English:
- IPA has one-to-one correspondence b/ween sound and symbol
(ex) [ð] stands for eth as in weather or this
(ex-English)
rough, through, bough though, cough
- all have -ough ending, but letters rep. diff. sounds for each word
[2]
George Bernard Shaw
- claimed that Eng word ghoti can be pronounced as “fish”
- how?
- look at pronounciations of following words:
enough<-- gh has 'f' sound
women <-- o for 'i' sound
[?] How does this o have an i sound in fish?
nation <-- ti for 'sh' sound
- saw this as English spelling sys being “inadequate”
p16
[1]
IPA
- one-to-one correspondence b/ween sound and symbol
- allows linguists to transcribe lang's
- consistently
- accurately
- but in North American usage, some phonetic symbols differ from those used in IPA transciption
- ex. IPA would be use [ ] while NA usage would be [ ]
- but this book will use IPA transcription
[!] Memorize IPA (Tables 2.16-2.17, p39-40 with Companion Website)
2.11 UNITS OF REPRESENTATION
[1]
- when hearing own lang spoken, our focus usually is on meanings of words, pharses & sentences, and not
much at'n to individual sounds
- but the focus of this section IS the individual sounds
[2]
[D] SEGMENTS
= individual speech sounds (phones)
- IPA rep. speech in form of segments

Only pages 1-3 are available for preview. Some parts have been intentionally blurred.

- ex. [p], [s]. [m]
- prod'ed by coordinating # of individual articulatory gestures
- ex. jaw movement, lip shape, tongue placement
- many of these individual activities rep. by
[D] FEATURE
= smallest unit of analysis of phonological structure
- combinations of these units comprise SEGMENTS
(eg) [nasal]
- almost never rep.'ed in writing sys's
- but still impt consistuents of linguistic rep'ion
- depict
- individual aspects of articulatory control
- certain acoustic effects prod'ed by articulation
- (in this chap, segmental transcription used b/c it is most popular used way of rep'ing speech)
2.12 SEGMENTS
[1]
RECALL: SEGMENT = INDIVIDUAL PHONE
-what evidences exist which suggest that speakers able to break down stream of speech into sound segments?
[2]
1. errors in speech prod'ion
- give eviedence for existence of segments
-> slip of the tongue phenomenon
(ex) instead of saying welcome mat, you say melcome wat
- suggests that segments are individual units of linguistic structure & should be rep'ed individually in sys. of
transcrip.
[3]
2. relative unchanging of speech sounds in human lang
- also suggests that segmental phonetic transcrip. is well-motivated way of transcribing speech
- impossible to rep. all variants of human speech sounds, b/c no one person says same sound exactly
- the same way twice
- the same as another person
- but sounds of speech remain unchanging from one lang to another s.t. can transcribe them consistently
- transcribe = put into written form
(ex)
p sound
- same whether u look at English or nanother lang
- same symbol is used b/c there is v.little diff. as to how the effort is to made to pronounce it, and thus the sounds
are not diff. enough to have diff. symbol
- ex. Eng speaker: presses lips together
- ex. russian speaker: does same thing but slightly inward
(ex)
but compare t and p sound
- sounds prod'ed sufficiently distinct from each other s.t. they are transcribed w/ sep. symbols (and not just one
symbol to rep. both)
You're Reading a Preview

Unlock to view full version