Title | Introduction to Phonetics Ch. 3 (Ogden): Representing the sounds of speech Reading Notes |
---|---|
Course | English Phonetics |
Institution | Lancaster University |
Pages | 8 |
File Size | 113.2 KB |
File Type | |
Total Downloads | 36 |
Total Views | 134 |
Download Introduction to Phonetics Ch. 3 (Ogden): Representing the sounds of speech Reading Notes PDF
Introduction to Phonetics: Chapter 3 – Representing the sounds of speech Week 20 LING223 Phonetics Reading Notes Phonetic transcriptions built on the apparently simple alphabetic principle of 1 symbol for each sound There are representations based on acoustic analysis (waveforms & spectrograms). o Help us see that despite our impressions, reinforced by an alphabetic writing system, sounds of speech are constantly changing o More commonly used in phonetics, make it possible to see individual aspects of sounds separately Phonetic Transcription o Represent an analysis of the sounds we can hear (using written letters) o Transcriptions often have a linguistic status Some transcriptions are more impressionistic, try capture what we hear rather than make claims about the significance of what we hear for making meaning o Phonetic values of letters are variable across languages IPA o Most common tool for phonetic transcription is the alphabet of the International Phonetic Association Both the Association and Alphabet commonly known as ‘the IPA’ o IPA is a set of tables containing symbols organised into rows and columns labelled with terms that have agreed meanings Unlike the random order of the Alphabet o Rows of the Consonant chart groups sounds according to manner 1st row – Plosives [p b t d ʈ ɖ c ɟ k g q ɢ ʔ] Rows below have sounds with progressively more open stricture o Columns organise symbols by place of articulation Leftmost column – Bilabial Following columns contain symbols for sounds made progressively further down the vocal tract Rightmost column – Glottal o Symbols of the IPA presented in a number of tables Main = Pulmonic egressive consonants and vowels Other tables have non-pulmonic consonants, diacritics and suprasegmentals (aspects of sound which relate to things like length, phrasing, intonation, etc)
Principles of IPA o Makes some assumptions about the nature of speech o According to IPA, ‘Some aspects of speech are linguistically relevant whilst others… are not’. Phonetic transcriptions should only have info that is linguistically meaningful o Physical differences, caused by things e.g. gender, age, physical state (out of breath) mean people sound different But these are physical not linguistic differences, phonetic transcription doesn’t capture them o ‘Speech can be represented partly as a sequence of discrete sounds or segments’ Segment A piece of something that has been chopped o IPA claims that speech isn’t made of segments but that we can represent it as segments o IPA establishes 2 major types of segment = Consonant & Vowel Consonant: Sounds produced with some kind of constriction in the vocal tract. Can feel, see and hear where the constrictions are made and what kind of constriction they are Vowel: Produced without a constriction in the vocal tract, harder to sense how they’re articulated. o Suprasegmentals Aspects of speech which persist over several segments E.g. duration, loudness, tempo (speed), pitch characteristics and voice quality. May include lip rounding Often thought as the ‘musical’ aspects of speech Important for marking all kinds of meanings, in particular speakers’ attitudes/stances to what they’re saying (or the person they’re saying it to) Marks how 1 utterance relates to another (e.g. continuation or disjunction) Different situations make different demands of a transcription One important point is the amount of detail that a transcription has o Transcriptions can have as many details we can observe = Narrow o Transcriptions can use a restricted set of symbols, therefore glossing over many phonetic details, as they’re predictable from the context, not important in distinguishing word meanings = Broad Typical in dictionaries
o Using familiar Roman letter shapes = Simple Transcriptions sometimes used to compare sounds Transcribing different varieties of a single sound when we hear them produces a comparative (narrower) transcription Systematic transcriptions limit the number of symbols used to a given set o In some circumstances, there are choices about how to represent sounds o Phonemic transcriptions are by definition, systematic o In a systematic transcription, set of available symbols is restricted Phonemic transcriptions embrace the concept that 1 linguistically meaningful sound should map on to 1 symbol, necessarily broad Allophonic transcriptions capture such details, predictable Narrower than phonemic transcriptions Impressionistic transcriptions use the full potential of the IPA, necessarily narrow For vowels its more difficult to provide a systematic transcription system o As vowels are extremely variable across varieties of English Citation form = Word spoken slowly in isolation, found in dictionaries o Concatenates the citation forms for each word in the sentences Broad transcription – All the symbols used represent sounds use to distinguish word meanings o Uses a small, limited set of transcription o Can add allophonic details making it ‘narrower’
Vowels before nasals in the same syllable – e.g. ‘think’ are often nasalised o Velum is lowered at the same time as a vowel is produced, allowing air to escape through both the nose and mouth o Marked by diacritic over the relevant symbol Voiced final plosives and fricatives (e.g. ‘need’, ‘shoes’) often produced without vocal fold vibration all through the consonant articulation when they occur finally and before voiceless consonants Phonetic details are sometimes predictable in English Since labiodental articulations don’t involve the same articulators as alveolar ones, 2 articulations can overlap o Represent this with diacritic [ ] meaning 2 articulations occur simultaneously
Sounds of speech made by changes in air pressure caused by airflow through the vocal tract o As air moves it causes perturbations which the ear picks up o Ear converts physical movements in the air into electrical signals sent to the brain, where processing of other kinds occurs Technology makes it possible to convert these changes of air pressure into pictures, static and unchanging, these pictures allow us to examine more detail of talk as it happened 2 main kinds of acoustic representations: o Waveforms o Spectrograms Wave forms = Graph o X-axis = Time, usually scaled in seconds or milli-seconds o Y-axis = Amplitude, representation of loudness The bigger the displacement from the baseline, the louder the sound is Spectrograms = Pictures of speech o Provide more complex information than waveforms o Times, as in waveforms marked on the x-axis o Frequency on the y-axis o Amplitude reflected in darkness Louder a given component in the speech signal is, the dark is appears 3 main kinds of sound easily distinguishable on a spectrogram o Periodic = Regularly repeating In speech, periodicity associated with vocal fold vibration, so periodic waveforms are associated with voicing Each major peak in a periodic waveform corresponds to 1 opening of the vocal folds One complete repetition called a cycle or period o Aperiodic = Random Continuous = like fricatives ([s f]) Transient = (short and momentary) like [p t k] The number of complete cycles the vocal folds make in 1 second = Fundamental frequency (f0) o Measured in Hertz (Hz)
In spectrograms, periodic signals have 2 important visual properties o Vertical striations Correspond to the opening of the vocal folds; each time the vocal folds open and air escapes, there is a sudden increase in amplitude Shows in the striations in the spectrogram which line with the peaks in the waveform Voicing regularly spikes in a waveform, corresponding regular striations in a spectrogram o Formants Darker horizontal bands running across the spectrogram Named counting upwards 1st = F1 2nd = F2 etc They’re natural resonances Each configuration of the vocal tract has its own natural resonance Aperiodic sounds have no repetition, random noise o Friction noise generated when the airflow between 2 articulators is turbulent o Correlate of this in a waveform is more irregular/random pattern than periodic Lacks the regular ups and downs of a periodic waveform Transient Sounds are aperiodic sounds which come and go quickly o E.g. knock on a door, cutlery rattling against eachother, firework exploding o Main source is the explosive release of a closure, e.g. releasing a closure for [p] or [k] o Others are the tongue/lips coming apart as someone starts to speak, bubbles of saliva bursting in the mouth, the velum being raised, sides of the tongue making contact with the teeth or cheeks o Show as spikes in a waveform Appear as dark vertical lines in spectrograms which only last a short time Acoustic representations are rarely static like transcriptions Transcriptions and acoustic representations capture different kinds of truth about speech o Transcriptions – Distinct sounds
More general o Acoustic = In speaking, articulators are rarely static Capture details and facts about 1 utterance on 1 occasion by 1 speaker Less useful in representing facts about language, easily subject to change When articulators move, these movements have acoustic consequences, this fluidity helps make everyday speech easier to perceive o Both are subjective Approximants Most profound class of consonants. Share many characteristics of vowels including characteristic formant patterns Palatal Approximant o [j] – Low F1 due to tongue body being close (raised) High F2 due to tongue body being fronted o No real clear beginning or end Labiovelar Approximants o Lips are rounded and tongue back is raised towards the velum o First formant moves downwards, while F2 gets much lower due to both tongue backing and lip-rounding o As approximations are made, vocal tract narrows, amplitude diminishes o [w] = Double Articulation Open approximation at the lips Lips are closely rounded, not close enough to produce friction noise Tongue back at the velum Tongue back raised up to the soft palate Velum is raised, air can’t escape through the nose Vocal folds vibrate, so there is voicing o No real clear beginning or end Lateral Approximants o Syllable-initial vs Syllable-final lateral Syllable-initial is clearer Has a higher F2 than syllable-final lateral Has a more abrupt ending than the beginning of the syllable-final
Transition out of the lateral portion is rapid, unlike syllable-final which has a slow transition Rhotic Approximants o Common property of having a low F3 (around 1800 Hz) o As most rhotic versions involve tongue body movement, which is relatively massive and slow to move, their acoustic properties tend to be extensive in the time domain o Plosives 3 phases of a typical plosive can be seen in a waveform and spectrogram Closing Phase o Formants move slights – Movements called ‘transitions’ o Transitions correspond to movements of the articulators into the closure o Although there’s some voicing during closure, its quiet and its amplitude diminishes until it eventually stops Hold Phase o Amplitude drops significantly as there is no airflow in the vocal tract o On release there’s a transient burst, seen as a sudden increase in energy, in the waveform and spectrogram Release o Voicing starts again very soon after the release of the closure Fricatives Voiced o Voicing and friction don’t overlap o Friction is loud and turbulent compared to voiceless fricatives o Friction is long in duration compared to voiceless fricatives Voiceless o Voicing and friction may (but do not necessarily) overlap o Friction is quiet (low amplitude) compared to voiced fricatives o Friction is short in duration compared to voiced fricatives Nasals Coming out of a nasal to a vowel, oral closure must be released, velum must be raised to block airflow through the nasal cavity
2 articulations are generally simultaneous, when the oral closure is released, airflow becomes oral o Producing a discontinuity in volume, can be seen in waveforms and spectrograms as a rapid change in amplitude Sometimes movements of the velum can be seen in a spectrogram Abrupt change in amplitude as nasality ends Some areas of low amplitude, such as around 1600Hz o Due to the nasal cavity absorbing some of the acoustic energy, gaps (called zeroes) often evident on spectrograms Although formants are visible during the nasal portion, they’re less distinct than in the vocalic portion o Because the formant peaks are wider, making them quieter and less visually prominent o Particularly noticeable in F1...