Background: the production of sound for speech Adam Albright - PowerPoint PPT Presentation

Background: the production of sound for speech Adam Albright (albright@mit.edu) LSA 2017 Phonology University of Kentucky

How do humans use sound to communicate? ▶ Speech production ▶ Intended meaning, calculate syntactic form, determine phonological form, translate into physical instructions ▶ Speech perception ▶ Sound waves enter ear, auditory analysis, parse for linguistically significant units, determine phonological and syntactic form, reconstruct intended meaning The nature of speech The production of voicing Vowels Consonants 1/54

Properties of speech sounds What makes a speech sound speech? ▶ Speech is noise created by air flowing from the lungs, modified as it exits through the mouth/nose ▶ Speech is just one of many modes of linguistic communication ▶ Hand signs, writing, semaphore, morse code, … ▶ But all except sign language are derivative from spoken speech ▶ Speech is continuous ▶ A fluid stream of sounds, without necessarily any pauses between individual sounds, words, or even sentences ▶ Speech is segmentable ▶ Composed of smaller units: words, morphemes, syllables, and individual sounds The nature of speech The production of voicing Vowels Consonants 2/54

Although it’s not obvious from the waveform, the speech stream is composed of smaller units, which we can refer to discretely Phrases, words, phones ( individual speech sounds) explain e k s p l ai n Speech: continuous, yet segmentable “If you can’t see that, then I don’t know if I can explain it to you” ▷ The nature of speech The production of voicing Vowels Consonants 3/54

explain e k s p l ai n Although it’s not obvious from the waveform, the speech stream is composed of smaller units, which we can refer to discretely Phrases, words, phones ( individual speech sounds) Speech: continuous, yet segmentable “If you can’t see that, then I don’t know if I can explain it to you” ▷ The nature of speech The production of voicing Vowels Consonants 3/54

Although it’s not obvious from the waveform, the speech stream is composed of smaller units, which we can refer to discretely Phrases, words, phones ( individual speech sounds) Speech: continuous, yet segmentable “If you can’t see that, then I don’t know if I can explain it to you” ▷ ▶ explain = e k s p l ai n The nature of speech The production of voicing Vowels Consonants 3/54

Speech: continuous, yet segmentable “If you can’t see that, then I don’t know if I can explain it to you” ▷ ▶ explain = e k s p l ai n Although it’s not obvious from the waveform, the speech stream is composed of smaller units, which we can refer to discretely ▶ Phrases, words, phones ( = individual speech sounds) The nature of speech The production of voicing Vowels Consonants 3/54

Evidence for discrete phones ▶ Phonemic awareness tasks (typically used for testing development in children) ▶ What is the first sound in cat ? ▶ Which of these words start with the same sound: cat, pat, coat ▶ Say cat without the [k]; say stand without the [t] ▶ Say cat. What word do you get if you change the [k] to [m]? ▶ Language games ▶ pig latin → ig-pay atin-lay ▶ Speech errors ▶ “cad you rean the small print?” ▶ (More common: ‘spoonerisms’ of word-initial consonants) ▶ Alphabetic writing The nature of speech The production of voicing Vowels Consonants 4/54

Producing speech T wo components: ▶ Aerodynamic component ▶ Air is forced out of lungs, though throat, mouth, and nose ▶ Articulatory component ▶ Air stream is modified in various ways on its way out The nature of speech The production of voicing Vowels Consonants 5/54

The vocal tract nasal passages palate alveolar ridge tongue dorsum brain lips velum teeth tongue uvula tongue blade larynx (glottis) to the lungs (my vocal tract, for illustrative purposes) The nature of speech The production of voicing Vowels Consonants 6/54

The larynx ( the glottis) From http://www.oto-hns.northwestern.edu/Voice/education.html (broken link) The vocal tract The first linguistically significant airstream obstruction: The nature of speech The production of voicing Vowels Consonants 7/54

The vocal tract The first linguistically significant airstream obstruction: The larynx ( = the glottis) ☞ From http://www.oto-hns.northwestern.edu/Voice/education.html (broken link) The nature of speech The production of voicing Vowels Consonants 7/54

Manipulations of the vocal folds Vocal folds are held at varying degrees of stiffness/closure ▶ Completely open: for respiration ( = breathing) ▶ Close enough to obstruct airflow: whispering, sighing, [h] (aspiration) ▶ Close together, intermediate stiffness: normal (modal) voice ▶ Tightly closed: glottal stop (as in [ʔ]uh-[ʔ]oh) The nature of speech The production of voicing Vowels Consonants 8/54

Voicing is the first major source of noise for speech Creates a complex periodic wave, which resonates in the mouth and nose, and is otherwise modified on its way into the atmosphere Voicing ▶ When the vocal folds are held somewhat close together, and with the right amount of stiffness, air rushing through causes them to “vibrate” ▶ Vocal folds are forced apart and rapidly close again, repeatedly ▶ Periodic vibrations = voicing ▶ Can feel externally as a “buzzing” on your throat; try [sssszzzzsssszzz] The nature of speech The production of voicing Vowels Consonants 9/54

Voicing ▶ When the vocal folds are held somewhat close together, and with the right amount of stiffness, air rushing through causes them to “vibrate” ▶ Vocal folds are forced apart and rapidly close again, repeatedly ▶ Periodic vibrations = voicing ▶ Can feel externally as a “buzzing” on your throat; try [sssszzzzsssszzz] Voicing is the first major source of noise for speech ▶ Creates a complex periodic wave, which resonates in the mouth and nose, and is otherwise modified on its way into the atmosphere The nature of speech The production of voicing Vowels Consonants 9/54

Source properties: voice quality How would you characterize the difference between these pairs of sounds? ▶ Mpi (Tibeto-Burman, spoken by ≈ 900 people in Thailand) T one Word Gloss Word Gloss Low rising si ‘to be putrid’ si ‘to be dried up’ ▷ ▷ Low level si ‘blood’ si ‘seven’ ▷ ▷ Mid rising si ‘to roll rope’ si ‘to smoke’ ▷ ▷ Mid level si (a color) si (classifier) ▷ ▷ High falling si ‘to die’ si (name) ▷ ▷ High level si ‘four’ si (name) ▷ ▷ ▶ Regular vs. ‘tense’ voice Hear them here. The nature of speech The production of voicing Vowels Consonants 10/54

Source properties: voice quality !Xóõ (Khoisan, Botswana/Namibia) Plain (voiced) Pharyngealized Strident Breathy qá ʕ a kǁáa k!ào k!a̤o ▷ ▷ ▷ ▷ ‘camelthorn tree’ ‘long ago’ ‘base’ ‘slope’ The nature of speech The production of voicing Vowels Consonants 11/54

Emotive use: frustration, surprise, etc. Individual style: Louis Armstrong, Julie Kavner, etc. description Source properties: voice quality Voice quality contrasts ▶ Bai (Tibeto-Burman, China) (Esling 2002) tɕi 31 ‘field’ tɕi 21 ‘flag’ ▷ ▷ (breathy) (harsh) ▶ English: what determines the voice quality that a word is spoken with? The nature of speech The production of voicing Vowels Consonants 12/54

Source properties: voice quality Voice quality contrasts ▶ Bai (Tibeto-Burman, China) (Esling 2002) tɕi 31 ‘field’ tɕi 21 ‘flag’ ▷ ▷ (breathy) (harsh) ▶ English: what determines the voice quality that a word is spoken with? ▶ Emotive use: frustration, surprise, etc. ▶ Individual style: Louis Armstrong, Julie Kavner, etc. description ▷ The nature of speech The production of voicing Vowels Consonants 12/54

The vocal tract as filter nasal passages palate alveolar ridge tongue dorsum brain lips velum teeth tongue uvula tongue blade larynx (glottis) to the lungs The nature of speech The production of voicing Vowels Consonants 13/54

The simplest effect The vocal tract as a tube, closed at one end From: Ladefoged (1996) Elements of Acoustic Phonetics , p. 117 The nature of speech The production of voicing Vowels Consonants 14/54

Resonances ▶ Formants: F n = 2 n − 1 c × L 4 ▶ F n = formant n ▶ c = speed of sound ≈ 35000cm/sec in warm moist air ▶ L = length of the tube The nature of speech The production of voicing Vowels Consonants 15/54

Calculating tube length ▶ Solving for L L = 2 n − 1 c × F n 4 ▶ Example: F1 = 500Hz, F2 = 1500Hz, F3 = 2500Hz ▶ L = 17.5cm The nature of speech The production of voicing Vowels Consonants 16/54

Possible filtering actions ▶ Can change the size and shape of the oral cavity ▶ Can obstruct the flow of air so much that it must flow through a very narrow passageway ▶ Can completely block airflow through the mouth ▶ Can open or close velum to allow allow air to flow through nose The nature of speech The production of voicing Vowels Consonants 17/54