Phonetic Documentation FIELD PHONETICS Phonetic Documentation • Describe the inventory of a language in a phonetically accurate way. • If the phonetic description is not accurate, it is difficult to compare to other languages (typologically, historically, synchronically), and to have the phonological patterns make sense. • To make an orthography that “sounds like the language when you read it out” Why phonetic fieldwork? • Transcriptions are influenced by the native language of the fieldworker (Strange 1995). Discovering Place of Articulation of consonants InField 2010 University of Oregon Dr. Amanda L. Miller Place of Articulation • Articulatory Photography: – Palatography / Linguography – Lip photography (still or video) • Acoustics of Stops: • Locus Equations (F2 Onset, F2 Mid) • Voice Onset Time Acoustics of Fricatives: • Spectral moments • Spectral peaks Phonetic features Palatography • By doing palatography, you can discover the exact place of articulation of a consonant, by seeing where black mixture on the tongue makes contact with the palate. Places of Articulation Palatography/ Linguography • Materials needed: – Dental mirror(s) – different sizes for different sized mouths, fog-proof mirror is quite helpful – Child’s paintbrushes or q-tips – Activated Charcoal (can substitute with jello mixture) – Olive oil (can substitute with other types of oil if olive oil is not available) – Dish soap to clean dental mirror between each production, and sink or plastic tub / dishpan to wash mirror Palatography / Linguography • Materials needed (continued) – Mixing bowl (disposable, as it is hard to remove charcoal mixture, a fresh one for each consultant) – Dish towel or apron to drape over consultant’s clothes to prevent from dirtying them – Lemonade or other citrus fruit for subjects to drink / eat to remove charcoal mixture, and fresh glass for each consultant (Fresh pineapple works wonders!, fresh oranges are not so good) Palatography / Linguography Materials needed (continued) – Digital camera, able to set to take good pictures in dark settings – Flashlight may be used to light up inside of mouth if camera cannot be adjusted for dark spaces – Plastic table cloth to avoid getting charcoal powder everywhere – Digital recorder to capture audio of the productions – Notebook to keep record of picture number / sound file number for each word Palatography / Linguography Other tips: • You can have the consultants paint their own tongue with a mirror, especially if there is sensitivity to having an unknown person , or person of different sex inside of someone’s mouth • It is best to do palatography / linguography outside, so that consultants can spit out charcoal mixture freely (alternatively you can be near a sink / tub, but this gets kind of gross). Palatography • Take a photo of the palate first without any charcoal mixture, so that you have a record of any dark spots on the teeth, etc. Palatography • Develop a wordlist – [a] vowel context on both sides is best because the tongue doesn’t approach the palate – bisyllabic words with the target consonant intervocalic, or bisyllabic with initial consonant – No other consonants, unless they are labial – Try to use real words. If the language has long words, you can ask the consultant to produce only the first syllable of a longer word. – Use minimal sets wherever possible – Keep prosodic positions the same (tone probably won’t effect articulation, though stress might). Palatography • Mix the activated charcoal and olive oil • If mixture is correct texture, it will cover the tongue well, and not slide off (except for some consultants who salivate a lot) • Paint the tongue with the mixture using a child’s paintbrush (Ladefoged 200X ) or q-tip – make sure it covers the tongue completely, nice and dark with no pink spots showing. • Paint as far back on the tongue as you can without gagging your consultant Palatography • Have the consultant produce the syllable/word in question, slip the dental mirror in their mouth, and take a picture of the palate. • Consultants can produce the word / syllable multiple times as long as they don’t speak other words or swallow or touch their palate with their tongue in between productions) • Have the consultant rinse their mouth with a citrus drink to remove the charcoal mixture. Palatography: A Recent Discovery • Fieldwork on Grootfontein !Xung in Namibia Claims for a retroflex click • Doke (1926) was the first to document a retroflex click • described !Xung spoken in Grootfontein area as having a 5th contrastive coronal click, the retroflex click • Snyman (1997) survey of !Xung lects • also described !Xung spoken in Grootfontein area as having a 5th retroflex click • König and Heine (2001) describe Ekoka !Xung as having four contrastive clicks: dental, alveolar, lateral, and retroflex (no palatal click) • admit that they do not know what the fourth click is phonetically 18 Claims against a retroflex click • Currently, the IPA recognizes only 4 coronal clicks: dental, alveolar, lateral, and palatal Dental Alveolar Lateral alveolar palatal | ! || ǂ c q x qc • no petition has ever been made to include the retroflex click in the IPA • Heikenen (1986) described the !Xung spoken in the Ekoka area as having a palatal, rather than retroflex, click • In Sounds of the World’s Languages (1996) Ladefoged and Maddieson state they do not believe a retroflex click to exist, based on the lack of evidence 19 !Xung is spoken in Namibia and Angola 20 !Xung Lects König and Heine (2001) Map 21 Phonemic Inventory: Ingressive Consonants /Palatal Doke (1926) 22 Grootfontein !Xung alveolar vs retroflex: palatography g!u ‘belly’ speaker KU g!!u ‘water’ speaker KU 23 Linguography • Linguography is when you paint the palate with the same charcoal / olive oil mixture, have the consultant produce a one syllable Ca word or bisyllabic aCa word, and see where the mixture rubs off on the tongue. Linguography • Linguography tells you whether a sound is apical (produced with the tip of the tongue), laminal (produced with the blade of the tongue), dorsal (produced with the tongue dorsum) or subapical (produced with the underside of the tongue touching the palate). Parts of the tongue Linguography: A Recent Discovery • Fieldwork in Grootfontein and Ekoka, Namibia Grootfontein !Xung alveolar vs retroflex: linguography g!u ‘belly’ speaker KU g!!u ‘water’ speaker KU 28 Ekoka reported retroflex vs Grootfontein palatal Grootfontein gǂa ‘to be old’ speaker XT Ekoka g!!a ‘to be old’ speaker DX 29 Lip photography • Tells you the degree of constriction • Provides information about the type of labial constriction: bilabial, labio-dental, linguolabial • Provides information about the gestural type involving the lips: protruded, or compressed Lip photography • It is good to get photographs from two angles (straight on, and from the side). • This can be done with one camera and a mirror, to get both images from the same production. • You can take still photographs (timing can be tricky), 30 fps video, high frame rate video Lip photography / video Acoustics of Stops: Place of articulation • You can also use acoustics to determine place of articulation of stops. • Acoustics is easier to collect, non-invasive, but you need multiple repetitions, and it takes longer to analyze. • Locus Equations can be used to determine place of articulation, and to uncover language specific C-V coarticulation patterns (Everett 2008). General tips for Acoustic Recordings • Head mounted microphones are quite good, because they keep the mike in the correct position (to the side of the mouth to avoid puffs of air), and keep it locked in the same position throughout the recording session (Shure SM10A with preamp like the USBPre, or Plantronics microphones). • You need a microphone with a good frequency response (80 Hz to 11,000 Hz). • Avoid background noise (use windscreen, away from other people, village) • Be sure to put an audio marking of what the recording is – the speaker’s name, location of recording, language, etc. • 22050 Hz sampling rate is sufficient (there is nothing measurable above 11,025 Hz) General tips for acoustic recordings • Record the whole wordlist, then repeat the whole thing (this way if the voice changes it changes for the whole list – not for some words on the list). • You need at least 10 repetitions of each consonant in each context to do statistics, and 15 gives you some room for recordings that aren’t usable – clipped, etc. • You can five repetitions of 3 words, 3 reps of 5 words, or 15 reps of 1 word, as long as it’s consistent across the types. General tips for acoustic recordings • Label your recordings in Praat, and write a script to analyze the data (or find one on the web). • You can record people for acoustics who may not be the best subjects for articulatory studies. • Spread the work out among the consultants, so more people get to work. Formant transitions in three synthetic stop consonant continua • Produced with the ‘pattern playback’ synthesiser. • The ‘steady-state’ F1 and F2 patterns determine the vowel. •The formant transitions constitute context sensitive cues to place of articulation of the stop. • Is there a common property that defines a given place of articulation in terms of formant transitions? The ‘locus’ of a formant transition • Figure shows steady state F2 for different vowels and their formant transitions for [d] (alveolar stop) • The formant transitions point back to a common ‘locus’ at 1.8 kHz. The ‘locus’ of a formant transition • The ‘locus’ of F2: – 3 kHz for velars – 1.8 kHz for alveolars – .6-.8 kHz for labials. • However, the locus is a somewhat idealized notion. • Velar stops vary in place of articulation with different vowels. Summary: cues to place of articulation in stop consonants. • Spectral energy distribution in the noise burst and formant transitions are the main cues. • Formant transitions are context-sensitive cues. • Context-sensitive cues require more complex signal processing. • A need for specialized phonetic feature detectors? Locus Equations • Record multiple repetitions of each stop in all vowel contexts. • Locus equations are linear regressions based on F2 formant transitions from vowel onsets to vowel midpoints. • The F2 value of the onset of a given vowel can be plotted on the y-axis, with the F2 value of the vowel’s midpoint plotted on the x-axis. • Has been used to document C-V coarticulation patterns of consonants in Karitiana (Amazonian language, Everett 2008). Locus Equations Linear regression equations derived from a two dimensional plot For any given consonant, the F2 onset for the consonant is plotted against the F2 target of the vowel on the x-axis (usually at the midpoint of the vowel). Y = mx + c where m is the slope of the line, and c is a constant. The linear behavior of locus equation data is best explained by the theory that the vowel gesture begins before the release of the consonant (Ohman 1966). What do they tell us? • Locus equation slopes correlate with the place of articulation of the consonant. • The slope allows us to assess the degree of coarticulation between the vowel and the consonant. • Different languages may also show differences in degree of coarticulation. (see Beddor et al.2002) Measuring Formants F2 SCATTER PLOTS Locus Equations: Kiritiana Slope correlates with POA POA: Cairene Arabic Slope and Y-Intercept Locus Equations: Crosslinguistic comparisons References Anderson, Victoria. (2008). Static Palatography for Language Fieldwork. Language Documentation and Conservation 2, 1, 127. Everett, Caleb. (2008). Locus Equation Analysis as a Tool for Linguistic Fieldwork. Language Documentation and Conservation 2, 2, 185-211. Ladefoged, Peter. (2003). Phonetic Data Analysis: An Introduction to Fieldwork and Instrumental Techniques. Malden, MA: Wiley-Blackwell. Ladefoged, Peter. (2007). Articulatory features for describing lexical distinctions. Language 83 (1). 161-180. Ladefoged, Peter and Maddieson, Ian. (1996). Sounds of the World’s Languages. Blackwell. References Ladefoged, Peter. (1957). “Use of Palatography”. Journal of Speech and Hearing Disorders. 22.5, 764-774. OSU Complete Denture Clinic Manual.