Phonetics

Phonetics

Janice FonGraduate Institute of Linguistics

National Taiwan University

2

Overview

Speech sounds and phonetic transcriptionArticulatory phoneticsPhonological categories and pronunciation variationAcoustic phonetics and signalsPhonetic resourcesAdvanced: articulatory and gestural phonology

3

Speech sounds and phonetic transcription (1)

PhoneticsThe study of speech sounds used in the languages of the world

PhoneA speech sound Represented with phonetic symbolsTwo types:

ConsonantsVowels

4

Speech sounds and phonetic transcription (2)

Phonetic alphabetsIPA

Standard developed by International Phonetic AssociationAlphabet + principles of transcription

ARPAbetDesigned for American English in ASCII symbolsComputer-friendly

5

Speech sounds and phonetic transcription (3)

[ɹ]

6

Speech sounds and phonetic transcription (4)

7

Articulatory phonetics (1)

Definition:The study of how phones are produced

The vocal organ

8

Articulatory phonetics (2)

Sounds are formed by the motion of air through the mouthConsonants:

Made by restricting or blocking the airflow in some wayMay be voiced or voiceless

Vowels:Made with less obstructionUsually voicedGenerally louder and longer than consonants

9

Articulatory phonetics (3)

Consonants are defined by Place of articulation

The point of maximum constriction

Manner of articulationHow the restriction of airflow is made

Voicing State of the glottis

10

Articulatory phonetics (4)

Place of articulation

coronal

11

Articulatory phonetics (5)

LabialSounds whose main restriction is formed at the lips Two subtypes:

Bilabial: two lipsLabiodental: upper teeth + lower lip

Examples [國 ]: [p] 爸 , [pʰ] 怕 , [m] 媽 , [f] 法[閩 ]: [b] 肉 , [p] 爸 , [pʰ] 打 , [m] 媽[E]: [p] spy, [pʰ] pie, [m] my, [f] four, [v] very

12

Articulatory phonetics (6)

DentalSounds made by placing the tongue against the teethTwo subtypes:

Dental: behind the teethInterdental: between the teeth

Examples [國 ]: [t̪] 大 , [t̪ʰ] 踏 , [s̪] 撒 , [t̪s ̪] 資 , [t ̪s ̪ʰ] 刺 , [n ̪] 那 , [l ̪] 辣 [閩 ]: [t̪] 大 , [t ̪ʰ] 拖 , [s ̪] 沙 , [z ̪] 熱 , [t ̪s ̪] 十 , [t ̪s ̪ʰ] 柴 , [n̪] 藍 , [l ̪] 賴 , [ɾ̪] 賊仔[E]: [θ] thing, [ð] the

13

Articulatory phonetics (7)

AlveolarSounds made by placing the tongue against the alveolar ridgeTwo subtypes:

Alveolar: b/t teeth and alveolar ridgePostalveolar/palato-alveolar: after the alveolar ridge

Examples [E]: [t] ten, [d] duck, [n] new, [s] sing, [z] zoo, [l] love, [ɹ] red, [ɾ] butter, [ʃ] she, [ʒ] garage, [tʃ] China, [dʒ] joy

14

Articulatory phonetics (8)

15

Articulatory phonetics (9)

PalatalSounds made by placing the tongue against the (hard) palateTwo subtypes:

Prepalatal/alveolopalatal: the arc towards the roof of the mouthPalatal: the roof of the mouth

Examples [國 ]: [ɕ] 削 , [tɕ] 交 , [tɕʰ] 敲 , [j] 妖[閩 ]: [ɕ] 是 , [tɕ] 摺 , [tɕʰ] 試 , [ʑ] 皺 , [j] 妖[E]: [j] yes

16

Articulatory phonetics (10)

Velar Sounds made by placing the tongue against the velum/soft palateExamples

[國 ]: [k] 該 , [kʰ] 開 , [x] 害 , [ŋ] 忙[閩 ]: [k] 該 , [kʰ] 開 , [g] 牛 , [ŋ] 黃[E]: [k] ski, [kʰ] key, [g] good, [ŋ] sing

17

Articulatory phonetics (11)

Glottal Sounds made at the glottisExamples

[國 ]: [ʔ] 唉[閩 ]: [h] 海 , [ʔ] 食[E]: [h] high, [ʔ] uh-oh

18

Articulatory phonetics (12)

Retroflex Sounds made with the tongue tip curled back

Examples [國 ]: [ʂ] 扇 , [tʂ] 詹 , [tʂʰ] 蟬 , [ʐ] 然

19

Articulatory phonetics (13)

Manner of articulationStop NasalFricative Affricates Approximant Flap/tap

20

Articulatory phonetics (14)Stop

= plosive Two phases:

Closure: airflow is completely blocked for a short timeRelease: an explosive sound as the air is released

Three types:Aspirated: closure + release + big puff of airUnaspirated: closure + release + small puff of airUnreleased stops: closure

Examples [國 ]: [p] 爸 , [pʰ] 怕 , [t ̪] 打 , [t̪ʰ] 踏 , [k] 乾 , [kʰ] 看[閩 ]: [p] 爸 , [pʰ] 打 , [b] 肉 , [t̪] 踏 , [t ̪ʰ] 桃 , [k] 菇 , [kʰ] 苦 , [g] 牛 , [p ̚] 十 , [t ̚] 結 , [k ̚] 角[E]: [p] spy, [pʰ] pie, [b] buy, [t] story, [tʰ] tie, [d] die, [k] sky, [kʰ] key, [g] guy

21

Articulatory phonetics (15)

voiceless

voiced

spy pie

buy

鼻 皮

抹

narrow glottal opening wide glottal opening

to buy

22

Articulatory phonetics (16)

NasalMade by lowering the velum and allowing air to pass into the nasal cavityExamples

[國 ]: [m] 媽 , [n ̪] 拿 , [ŋ] 紅[閩 ]: [m] 媽 , [n ̪] 貓 , [ŋ] 紅[E]: [m] my, [n] new, [ŋ] sing

23

Articulatory phonetics (17)

Fricatives Process

Airflow is constricted but not cutoff completelyTurbulence is created

Two typesSibilants: fricatives with high-pitched hissing noiseNonsibilants: fricatives without high-pitched hissing noise

Examples [國 ]: [f] 法 , [s ̪] 素 , [ʂ] 刪 , [ʐ] 然 , [ɕ] 先 , [x] 何[閩 ]: [s̪] 素 , [z̪] 熱 , [ɕ] 先 , [ʑ] 任 , [h] 何[E]: [f] few, [v] view, [θ] they, [ð] the, [s] sing, [z] zoo, [ʃ] she, [ʒ] garage, [h] high

24

Articulatory phonetics (18)

Affricates Homorganic stop + fricative Examples

[國 ]: [t ̪s ̪] 租 , [t ̪s̪ʰ] 粗 , [tʂ] 朱 , [tʂʰ] 出 , [tɕ] 家 , [tɕʰ] 恰[閩 ]: [t ̪s ̪] 租 , [t ̪s̪ʰ] 粗 , [tɕ] 一 , [tɕʰ] 七[E]: [f] few, [v] view, [θ] they, [ð] the, [s] sing, [z] zoo, [tʃ] choice, [dʒ] job

25

Articulatory phonetics (19)

ApproximantTwo articulators are close together but not close enough to cause turbulent airflowTwo types:

Approximant: air flows from the center of the tongue outwardLateral approximant: air flows from the side(s) of the tongue outward

Examples [國 ]: [j] 牙 , [w] 娃 , [ɥ] 圓 , [l ̪] 來[閩 ]: [j] 厭 , [w] 歪 , [l̪] 來[E]: [j] yes, [w] we, [l] like, [ɹ] read

26

Articulatory phonetics (20)

Tap/flapA quick motion of the tongue against a hard surfaceExamples

[閩 ]: [ɾ̪] 賊仔 [E]: [ɾ] butter

27

Articulatory phonetics (21)

28

Articulatory phonetics (22)

Vowels are defined byTongue heightTongue frontness/backnessRoundedness

29

Articulatory phonetics (23)

意 欲 思 屋

矮 黑可

辦 棒

about

it

bet

batbut bought

30

Articulatory phonetics (24)

Three types of vowelsMonophthongs

A vowel in which the tongue position does not changeDiphthongs

A vowel in which the tongue position changes once= VG/GV(ex) [國 ]: [aɪ] 埃 , [eɪ] 黑 , [aʊ] 好 , [oʊ] 候 , [ja] 牙 , [wa] 娃 , [ɥɛ] 約(ex) [閩 ]: [ju] 油(ex) [E]: [ɔɪ] boy

Triphthongs A vowel in which the tongue position changes twice= GVG(ex) [國 ]: [jaʊ] 妖 , [jaɪ] 崖 , [waɪ] 歪

31

Articulatory phonetics (25)

SyllableConsonants and vowels combine to make a syllable

32

Articulatory phonetics (26)

PhonotacticsConstraints on which phones can follow each other in a language[國 ] CCVC: 片[閩 ] CCVC: 接 [E] CCCVCCC: strengths[E] CCVCCCC: twelfths

33

Articulatory phonetics (27)

Lexical stressA prominent syllable in a word that is determined by the lexicon

(ex) [國 ] 東西 [ˈtoŋɕi] vs. [ˈtoŋˈɕi]Three levels

Primary stress: dictionarySecondary stress: dictionaryUnstressed: dictionary

In unstressed syllables, vowels can be Reduced: about [ə]Unreduced: carry

34

Articulatory phonetics (28)

Sentence stress= pitch accentA prominent syllable in an utterance that is determined by pragmatics

(ex) 連站都站不好！ 連戰都站不好！

35

Phonological categories and pronunciation variation (1)

Pronunciations vary!

36

Phonological categories and pronunciation variation (2)

Phoneme allophone/t/ [t, tʰ, ʔ, ʔt, ɾ, t ̚, t]̪

37

Phonological categories and pronunciation variation (3)

(assimilation)

The degree of variation increases with more natural and colloquial speech

38

Phonological categories and pronunciation variation (4)

Coarticulation The movement of articulators to anticipate the next sound or persevering movement from the last sound(ex) nasalization, palatalization, voicing, etc.

39

Phonological categories and pronunciation variation (5)

Distinctive features Binary variables which express some generalizations about groups of phonemesUsed to represent each phoneme as a matrix of feature valuesPlace features

[labial], [coronal], [dorsal]Manner features

[consonantal], [vocalic], [continuant], [sonorant]Voicing features

[voice]Vowel features

[high], [low], [back], [round]

40

Phonological categories and pronunciation variation (6)

41

Phonological categories and pronunciation variation (7)

Predicting phonetic variationPhonological rule

Caveat:Variation is a stochastic processMany non-phonetic factors are important to this prediction task

42

Phonological categories and pronunciation variation (8)

Factors influencing phonetic variationNon-phonetic

Speech rateWord frequencySpeaker’s state of mindMorphological boundariesSociolinguistic factors

Phonetic Intrinsic sound qualityProsodic boundariesCoarticulation

43

Acoustic phonetics and signals (1)

Waves

44

Acoustic phonetics and signals (2)

45

Acoustic phonetics and signals (3)

For computers to record and manipulate speech, it is necessary to perform analog-to-digital conversionA two-step process

Sampling—limit the number of places after the decimal point on the time axisQuantization—limit the number of places after the decimal point on the amplitude axis

46

Acoustic phonetics and signals (4)

Sampling rateNumber of sample taken per secondAt least two samples per cycleNyquist frequency

the highest frequency component that can be captured with a given sampling rate= ½ sampling rate

47

Acoustic phonetics and signals (5)

Frequency = 100 HzSampling rate = 200 HzNyquist frequency = 100 Hz

48

Acoustic phonetics and signals (6)

signal not in the original signal aliasing

49

Acoustic phonetics and signals (7)

Quantization Analogous to measuring the amplitudes in the waveform with a rulerThe main issue is the accuracy of the amplitude measurementsThe most popular choices for the number of bits used to encode speech samples are 8, 12, and 16 bits

50

Acoustic phonetics and signals (8)

51

Acoustic phonetics and signals (9)

Recording formatChannels: mono or stereoFormat: linear or compressed

Linear: linear PCMCompressed: -law log compression

Common format.wav, .aiff, .au

52

Acoustic phonetics and signals (10)

53

Acoustic phonetics and signals (11)

Fundamental frequency = F0

Measures speed of vocal fold vibrationHertz

AmplitudeMeasures amount of air pressure variationPascal (Pa)RMS

IntensitySound power per unit area measuring at a listener’s location Decibels (dB)

54

Acoustic phonetics and signals (12)

55

Acoustic phonetics and signals (13)

Perceptual propertiesPitch

Mental sensation or perceptual correlate of F0

Mel scale

Loudness Perceptual correlated of power Human ear has greater resolution in the low power rangeLoudness interacts with frequency

56

Acoustic phonetics and signals (14)

57

Acoustic phonetics and signals (15)

58

Acoustic phonetics and signals (16)

most sensitive

quick sensitivity drop

59

Acoustic phonetics and signals (17)

1000 Hz 100 Hz

1x 2x 1x 2x

60

Acoustic phonetics and signals (18)

Autocorrelation methodA pitch extraction algorithm that correlates the signal with itself at various offsets

The offset that gives the highest correlation gives the period of the signal

61

Acoustic phonetics and signals (19)

62

Acoustic phonetics and signals (20)

Interpreting phones from a waveform

63

Acoustic phonetics and signals (21)

64

Acoustic phonetics and signals (22)

FFT

65

Acoustic phonetics and signals (23)

66

Acoustic phonetics and signals (24)

The source-filter model

67

Phonetic resources

Pronunciation dictionariesCELEXCMUdictPRONLEX

Phonetically annotated corpusEnglish: TIMIT, Switchboard, BuckeyeGerman: Kiel corpus Japanese: CSJMandarin: AS, NTU (Taiwan), CASS (China)

Phonetic softwardsPRAAT

68

Advanced: articulatory and gestural phonology (1)

Articulatory phonologyRepresents a speech utterance as a sequence of potentially overlapping articulatory gestures

Advantages:Gesture scores are likely to be much better hidden states at capturing the continuous nature of speech than a discrete sequence of phones

Help model the fine-grained effects of coarticulation of neighboring gestures

69

Advanced: articulatory and gestural phonology (2)