Qual Study Sessions: Phonetics / Phonology 02/04/08 - Phonetics Why do voiced stops go to voiced fricatives? - Because the pressure eventually equalizes in the oral cavity in voiced stops, but in fricatives the pressure never equalizes because the air escapes out one end Why do pharyngealized sounds get reinterpreted as labialized sounds in borrowed words? - Because both pharyngealization and labialization have low f3, the one is interpreted as the other Spectrogram reading - velar pinch, f2 & f3 meeting before and after a stop - bilabials, f2 & f3 rise coming out of stop, dip going into stop - alveolars stay level - anti-resonances for nasals, laterals, liquids - r has super low f3, low everything - l also has low f1, f2, f3 - voicing bar - stridents: esh energy in 2000-3000 range, ess much higher (5000) - weak fricatives (v, th, f etc.) have lower energy weak formant structure Preparing for studying: Jasmin - articulatory phonetics Alice - perturbation theory Ryan - speech pathologies, ipa flashcards Peet - stress, intonation, pitch accent etc. 02/14/2008 Ryan's Notes on Speech Pathologies: Speech pathologies Cleft palate -hard palate and nasal septum not joined -constant passage b/w oral and nasal cavities (i.e., like if velo-pharyngeal port is always open) -results in nasal voice quality Laryngectomy (larynx surgically removed) -no sound source for generating speech -can use voice prosthesis (artificial larynx), but no control over pitch (i.e., vocal fold length) -esophageal speech - air swallowed into esophagus, expelled as belch -tracheoesophageal valve - blocks stoma; air channelled through valve into esophagus, expelled as belch; pitch changes due to changed in rate of airflow Aglossia/glossectomy (no tongue, or part of tongue missing) -since there are multiple ways to produce sounds, people can usually communicate fairly well Paralysis of pulmonic system -use ventilator to assist in breathing -can produce sounds during expiration cycle Emphysema -loss of elasticity of lungs -more difficult to maintain subglottal pressure -makes producing pulmonic sounds more difficult; also decrease in volume, pitch 02/21/2008 Source-Filter Theory - source is voice-box, filter is vocal tract Filter function - peaks in the filter function of the vocal tract = formants - for 17 cm unrestricted vocal tract normal formants are at 500 Hz, 1500 Hz, 2500 Hz, 3500 Hz etc. - A resonant frequency of a physical body is a frequency whose standing wave optimally conforms to the boundary conditions of the body. Perturbation Theory - how constrictions at locations affect formants - nodes and antinodes Rules of Thumb - constriction at a node raises resonant frequency - constriction at an antinode lowers resonant frequency Constriction Fraction of Lvt (Length of vocal tract) glottis 0 pharynx 1/4 uvula 1/2 velum 2/3 palate 3/4 alveolum 7/8 lips 4/4 Lvt = ( 2n-1 ) / 4 lambda_n length of vocal tract = ( 2n-1 ) / 4 times resonance frequency f_n = c/(4/(2n-1)*L) resonance frequency = speed of sound / (( 4 / 2n-1 ) * L) speed of sound = 34,000 cm/sec Formants *ARE* resonances Formants *ARE NOT* the same as harmonics Harmonics are multiples of the fundamental frequency - formants are high energy bands (?) - value given for formant is center of band (?), or peak of band (?) 02/25/2008 Vowels Around the World (from 10/22/2007) Humans can distinguish about 40 vowel heights and between 5 and 30 degrees of backness but the differences in vowels may not be contrastive. ~20% of languages have 5 vowel systems the realization of the "same" vowels cross-linguistically might not be the same - [e] in Danish might not be the same as [e] in English cross linguistic tendency: more consonants means fewer vowels & vis versa - Zulu Counter-examples: - British English: 20 vowels, 23 consonants - Hawaiian: 5 vowels, 8 consonants Vowels described with auditory impressionistic description - back, front, high, low - different from alveolar, etc. - main distinctions: front, central, back high, mid, low rounded, unrounded - additional distinctions: tense, lax Primary Cardinal vowels - auditory equidistant points in the vowel space - front unrounded, back rounded, no central vowels - 1 - 8: i - u - extreme points of vowel space, not necessary corresponding to actual vowel space of any language - standards of reference Secondary Cardinal vowels (opposite rounding) Contrastive vs. intrinsic length - relative duration can be distinctive, doesn't have to be Nasalization can be contrastive, doesn't have to be Nasal vs. Nasalized vowels - nasal means contrastive - nasalized means non-contrastive (from 10/24/2007) More on Vowels can also talk about vowels in terms of phonation type modal voice: regular voicing creaky: waveforms are far apart harmonics have more energy than in modal voice correlated with low pitch also called laryngealization low airflow, hence vocal folds flapping slowly vocal folds slightly open at the open breathy: relaxation of the vocal folds pharyngealized: constricted pharynx epiglottalized: tightening the lower part of the pharynx near the glottis ATR (Advanced Tongue Root) - when the root of the tongue is drawn forward and/or lowering of the larynx - the point is to increase the size of the pharynx - this often results in a breathy quality - vowels that are +ATR tend to have lower F1 and higher F2 - tend to sound more raised Diphthongs - distinction between diphthongs and vowel clusters depends on phonology and morphology of language - diphthongs can't be split up - diphthongs pattern with monophthongs - second element of diphthong represents the direction of glide, NOT target to be achieved 03/03/2008 Perception McGurk Effect ba ba + ga ga = da da categorical perception when we hear sounds we put them in specifc categories all audition, not just humans natural boundaries general property of sound perception within-category compression, between-category compression connected speech, dynamic speech co-articulation inertia/sluggishness of articulators leads to sound/gestural overlap perception of clear speech is poorer than sloppy speech can affect more than the immediate neighbors progressive - preservatory regressive - anticipatory in 'late calls' the [t] isn't heard because the [k] overlaps it but the vowel in 'late' shows effects of being next to a [t] 03/10/2008 Acoustics - power spectrum is 2 dimensional, showing sound at one time, one dimension is harmonic, other dimension is hertz - spectrogram is 3 dimensional, third dimension is time, loudness becomes darkness, hertz is on y axis