Download Chapter 1-Phonetics and Phonology:Understanding the Sounds of Speech and more Exercises Linguistic Philosophy in PDF only on Docsity!
The Scream (detail) , by Edvard Munch Chapter 1 – Phonetics and phonology: understanding the sounds of speech Robert Kirchner, University of Alberta Introduction In most fields of study, language is thought of principally in terms of the written word, for it is in this form that we usually make permanent records of important ideas. Rela- tively little attention is spared for something as fleeting and unremarkable as spoken con- versation. In linguistics, however, speech, rather than writing, is regarded as more central to human language, for several reasons. First, humans have probably used spoken lan- guages for 100,000 years, perhaps longer. Writing is a relatively recent development, only a few thousand years old. Even today, most of the world's 5,000 or so languages have no established writing system. But there is no society which communicates just by writing, without a spoken language. Furthermore, children learn to speak long before they learn to read and write; indeed, learning of spoken language takes place without formal instruction. But does ordinary speech really warrant scientific atten- tion? Although we generally take the processes of speech pro- duction and recognition for granted, they involve a range of surprisingly intricate mental abilities – part of the knowledge we have of the language(s) we speak. The words that we wish to express seem to emerge inexplicably from our mouths, as soundwaves. These soundwaves then hit the hearer's ear, send- ing auditory signals to the brain, which are interpreted – again, seemingly magic – as the words intended by the speaker. What kind of mental system might underlie this capacity to produce and recognize speech? Which aspects of this system appear to be common to all humans, and which aspects vary from lan- guage to language? And what exactly goes on in the mouth and throat to produce speech? These sorts of questions are the domain of phonetics and phonology (both from the Greek root phon - 'sound'), the two subfields of linguistics concerned with speech sounds. In the remainder of this chapter, we examine some basic observations, terminology, and techniques of analysis used by phoneticians and phonologists to address these questions. Phonetics, phonology – what's the difference? Traditionally, phonetics deals with measurable, physical properties of speech sounds themselves, i.e. precisely how the mouth produces certain sounds, and the characteris- tics of the resulting soundwaves; while phonology investigates the mental system for representing and processing speech sounds within particular languages. In recent years, however, the two fields have increasingly overlapped in scope. For our purposes, the im- portant point is that linguists (whether they're called phoneticians or phonologists) have accumulated some basic observations about how the speech systems of human language 'work,' and these principles have a good deal to do with the physical properties of the speech sounds in question.
I. Forget letters, we're talking sounds Sound energy is disturbance of air molecules: the disturbance radiates outward from its source, in waves of fluctuating air pressure ('soundwaves'), like ripples from a stone dropped in a pond. When we speak of an individual sound of speech, however, we mean something more specific: a portion of the speech in which the sound energy (and the configuration of the mouth to produce that sound energy) remains relatively stable. In the word so , for example, the sound energy changes, from a hissing sound at the be- ginning (with the mouth relatively closed) to a more open, singable sound at the end. But within each of these two portions of the word – the hissing sound of the s, and the singa- ble sound of the o , there is relatively stability. We can therefore say that so is composed of two distinct sounds. Indeed, this decomposition of words into individual speech sounds is reflected to some extent in our writing system, for we spell so with two letters. Nevertheless, it is important to bear in mind, throughout this chapter, that we are interested in the sounds which make up words, not the letters with which they are spelled. The word fought , for example, has six letters, but only three sounds: the f, followed by a single vowel sound (written with two letters, ou, in this word), and the final t. The gh is, of course, 'silent'; it is not part of the word's sounds, so we disregard it. In fact, for the purpose of representing sounds, the English spelling system is quite unreliable – as generations of schoolchildren, struggling to memorize English spellings, can appreciate. The letter c , for instance, is pronounced like s in some words (e.g. cell ), and k in others (e.g. call ). Similarly, o corresponds to one vowel sound in Rob- ert and a different one in robe. The inadequacies become even more obvious if we try to transcribe (write down) the words of other languages – as linguists must do. The language might have no estab- lished writing system, or it may have sounds which don't occur in English. We might in- vent our own way of transcribing such sounds, using the closest-sounding letters of English. But how is a Russian linguist going to understand our English-based transcrip- tions, if she is not fluent in English? And how are we to understand this Russian's tran- scriptions of an unusual Kurdish dialect, written in the Russian (Cyrillic) alphabet, if we are not fluent in Russian? Linguists need an internationally agreed-upon system of tran- scription, in which the symbols correspond straightforwardly to sounds, and in which there are enough symbols to represent all the sounds of the world's languages. This system is called the International Phonetic Alphabet (IPA) , first devel- oped in 1886 and since modified in light of subsequent linguistic discoveries. For your interest, the full chart of IPA symbols appears at the end of this chapter. For present pur- poses however, we'll focus on the symbols needed for the basic sounds of North Ameri- can English, adding other symbols as needed. A. Consonants. If your first language is not English, and you are not sure how to pronounce any of the example words in Table 1, check with a native English-speaker.
[ g ]. This symbol is always pronounced as a 'hard' g , as in get or bag, never as in gem or age. By the same token, a number of letters of the alphabet are not needed as IPA symbols for transcribing English consonant sounds. For example, the qu in quick is the same as [kw], and the end of tax is simply a [ks] sequence_._ As we already noted, either [s] or [k] can replace c , depending on the word. These extra letters are used in IPA to denote different sounds, not found among the basic sounds of English. B. Vowels. The vowels require more careful study, as the symbols are less famil- iar; and even the familiar symbols generally do not have the phonetic values we would expect from English spelling. They're more like the spelling-pronunciation correspon- dences of Spanish or Italian. Table 2 : IPA symbols for the basic vowel sounds of North American English IPA symbol Example words IPA symbol Example words i s ee , funn y , b ea d U p u ll, g oo d, w ou ld I b i t, s i ng, r i b o g o , b oa t, p o le, s ew e h a ze, gr ea t, ob ey O c au ght, d aw n, b o ss E (^) b e t, s e nd, aff e ct A (^) c o t, D o n, f a ther œ (^) st a mp, p a ck, y ea h å (^) sh u t, c o me, b u g^1 u l oo n, fl u te, wh o ´ a bout, Albert a , el e ment Note that, for many of these vowel sounds, a number of different spellings are used in English. The [U] sound, for example is spelled oo in good, but ou in would ; nevertheless, the vowel sound is the same in both words: would and good rhyme, which tells us that the vowel sounds (as well as the final consonants) in these two words are identical. Examine the example words for the other vowel symbols as well, to satisfy yourself that the sounds corresponding to each symbol really are the same. The point of this mental exercise is to develop an awareness of the distinct vowel SOUNDS, independent of their spelling in particular words. [ O,A ]. Except in certain regions, most younger North American English speakers nowadays make no distinction between [O] (as in caught ) and [A] (as in cot ), instead using [O] for both cases; or [A] for both cases; or a vowel somewhere between the two ([Å]). If you pronounce cot / caught and Don / Dawn the same, you're in this group of cutting-edge English speakers. Dialect variation. More generally, bear in mind that the symbols and examples in Table 2 hold true for most dialects of North American English. But if you speak a dialect distinct from the North American mainstream, your vowels may vary significantly, as English dialects differ mainly in the vowels. Remember: the 'right' way to transcribe a word depends on its pronunciation in the speech you are transcribing, not on any exter- nal standard of correctness. (^1) The symbol [ø] is widely used for this vowel of North American English (although it represents a differ- ent vowel according to the IPA chart). Either symbol is acceptable for our purposes.
Diphthongs. English also has a few 'vow- els' that are really a sequence of two vowels. These are called diphthongs (from Greek di- 'two' + phthongos 'sound'). The most common diphthong is the sound in hide or eye. It begins something like [A], and moves smoothly into [I]. If you say eye slowly, you can hear the one vowel change into the other. Because the sounds of a diphthong change from beginning to end, they are transcribed in IPA with two vowel symbols, as shown in Table 3. Table 3 : Diphthongs IPA symbol Examples AI (^) h i de, eye , s i gh AU h ow , r ou nd OI b oy , av oi d Vowel + [ ® ] sequences. When a vowel appears before [®] in North American English, the [®] has a strong effect on the vowel's sound, making identification of the vowel tricky, in some cases, for beginners at phonetic transcription. Here, then, is a list of examples. Table 4 : Vowel + [®] sequences IPA symbol Examples e® h air , c are d, wh ere , b ear i® h ere , w eir d, ear , b eer ([I®] in many dialects) A® b arr ed, f ar , ar m O® (^) b or n, st ore , p our , sh ore ([o®] in a few dialects) U® (^) t our , p oor , s ure ([u®] or [O®] in many dialects) Check this list carefully, thinking about how you pronounce these words. Are the vowel + [®] sequences in the examples on each row the same for you? Are the sounds of each row different from those of the other rows? For example, do you pronounce pour and poor differently, or the same? A generation ago, many dialects of North American English had even more distinct vowel + [®] sequences. The author's father, for example, pronounces Mary , merry , and The Queen's English? We beg your Majesty's pardon, but there is nothing inher- ently superior about any particular dialect of English – or any other language. The populations of Alabama, Manitoba, and Oxfordshire are equally 'good' English speakers, from a lin- guist's perspective. Each distinct dialect presents an equally valid object of study. The belief that some dialects are bet- ter than others is just another form of the attitude that some ethnic groups or social classes are better than others (more simply, 'prejudice'). For we tend (often unconsciously) to attach prestige to the dialects of groups we admire, and to stigmatize the dialects of groups we look down upon. Dude, that attitude is, like, s0oo last millenium!
Finally, a warning: beginners often approach IPA transcription by trying to translate directly from English spellings into IPA symbols: e.g. ea (as in treat ) = [i] in IPA. But the ea in great , for example, is not [i], it's [e]. As we've noted above, English spellings of sounds are notori- ously inconsistent, making this strategy fundamentally unworkable. Rather, al- ways be aware of how the word sounds in any exercise involving phonetic tran- scription. Here's the paragraph to the left, in IPA: [fAIn´li, ´ wA®nIN: bigIn®z Of´n ´p®otS AI pi e t®œnsk®IpS´n bAI t®AIN t´ t®œnzlet dAI®Ektli fr´m INglIS spElINz Int´ AI pi e sImb´lz. båt D´ i e In g®et, f® ´gzœmpl, Iz nAt i, Its e. œz wiv not´d ´båv, INglIS spElINz ´v sAUndz ® notO®I´sli INk´nsIst´nt, mekIN DIs st®œt´dZi fånd´mEnt´li ånw®k´bl. ®œD®, Olwiz bi ´we® ´v hAU D´ w®d sAUndz In Eni Eks®sAIz InvAlvIN f´nEtIk t®œnsk®IpS´n.] Now here is the opening paragraph of this chapter in IPA. See if you can read it without referring back to page one: In most fildz ´v stådi, lœNgw´dZ Iz TOt ´v prIns´pli In t®mz ´v D´ ®It´n w®d, f® It Iz In DIs fO®m Dœt wi juZ´li mek p®m´n´nt ®Ek®dz ´v ImpO®t´nt AIdi´z. ®El´tIvli lIt´l ´tEnS´n Iz spe®d f® såmpTIN ´z flitIN ´nd ´n®imA®k´b´l ´z spok´n kAnv®seS´n. In lINgwIstIks, hAwEv®, spitS, ®œD® D´n ®AItIN, Iz ®igA®d´d ´z mO® sEnt®´l t´ hjum´n lœNgw´dZ, f® sEv®´l ®iz´nz. f®st, hjum´nz h´v p®Ab´bli spok´n lœNgw´dZ´z f® fIfti TAuz´nd ji®z, p®hœps måtS lONg®. ®AItIN Iz a ®El´tIvli ®is´nt d´vEl´pm´nt, onli ´ fju TAuz´nd ji®z old. iv´n t´de, De® ® s´sAI´tiz witS spik lœNgw´dZ´z wIT no ®AItIN sist´m, O® we® ®AItIN Iz sEld´m juzd. båt De® Iz no s´sAI´ti wItS k´mjunIkets dZåst bAI ®AItIN, wITAUt ´ spok´n lœNgw´dZ. f®D®mO®, tSIld®´n l®n t´ spik lON bifO® De l®n t´ ®id ´nd ®AIt; Indid, l®nIN ´v spok´n lœNgw´dZ teks ples wIDAUt fO®m´l Inst®åkS´n. båt d´z O®dIne®i spitS ®ili wO®´nt sAI´ntIfIk ´t´nS´n? OlDo wi dZEn®´li tek D´ p®AsEs´z ´v spitS p®´dåkS´n ´nd ®Ek´gnIS´n f® g®œnt´d, De InvAlv ´ ®endZ ´v s®p®AIzINli Int®Ik´t mEnt´l ´bIl´tiz – pA®t ´v D´ nAl´dZ wi hœv ´v D´ lœNgw´dZ´z wi spik. D´ w®dz D´t wi wIS tu ´ksp®Es sim tu im®dZ Ot´mœtIkli f®´m AU® mAUDz, ´z sAUnd wevz. Dise sAUnd wevz DEn hIt D´ hi®®z i®, sEndIN Od´tO®i sIgn´lz t´ D´ b®en, wItS ® Int®p®´t´d – ´gEn, simINli Ot´mœt´kli – ´z D´ w®dz IntEnd´d bAi D´ spik®. wåt kAInd ´v mEnt´l sist´m mAIt ånd®lAI DIs k´pœs´ti t´ ®Ek´gnAIz ´nd p®´dus spitS? wItS œspEkts ´v DIs sist´m ´pi® t´ bi kAm´n tu Ol hjum´nz, ´nd wItS œspEkts ve®i f®´m lœNgw´dZ t´ lœNgw´dZ? ´nd wåt ´gzœktli goz An In D´ mAUT ´nd T®ot In O®d® t´ p®´dus Dis sAUndz? Diz sO®ts ´v kwEstS´ns ® D´ domen ´v f´nEtIks ´nd f´nAl´dZi (boT f®´m D´ g®ik ®ut fon- 'sAUnd'), D´ tu såbfildz ´v lINgwIstIks k´ns®nd wIT spitS sAUndz. In D´ ®imend® ´v DIs tSœpt®, wi ´gzœm´n s´m besIk Abz®veS´nz, t®m´nAl´dZi, ´nd tEkniks ´v ´nœl´sIs juzd bAI fon´tIS´nz ´nd f´nAl´dZIsts tu ´d®Es Dis kwEstS´nz. Note that the pronunciation of particular words in a phrase may vary from their pronun- ciation in isolation, e.g. and as [´nd] rather than [œnd]. For your convenient reference, we repeat, in consolidated form, the IPA symbols discussed above:
Table 5 : IPA symbols for the basic sounds of North American English Consonants Vowels p p at, hi pp y, tri p i s ee , funn y , b ea d t t op, re t urn, pa t I b i t, s i ng, r i b k (^) c at, bi k er, sti ck e (^) h a ze, gr ea t, ob ey b (^) b at, ru bb er, sno b E (^) b e t, s e nd, aff e ct d d ay, a d ore, ba d œ st a mp, p a ck, y ea h g g uts, ba gg y, ri g u l oo n, fl u te, s ou p, wh o f ph oto, co ff ee, lau gh U p u ll, g oo d, b oo k v v oice, ri v er, li v e o g o , b oa t, p o le, s ew T (^) th ink, au th or, tee th O (^) c au ght, d aw n, b o ss D (^) th is, wea th er, tee th e A (^) c o t, D o n, f a ther s (^) s it, re c eive, ba ss å (^) sh u t, c o me, b u g z z oom, fu zz y, ma ze ´ a round, Albert a , el e m e nt S (^) sh ip, pre ss ure, ra sh tS ch ip, fu t ure, sti tch Diphthongs Z J acques, lei s ure, rou g e AI h i de, eye , I , s i gh dZ j erk, proce d ure, e dge AU h ow , r ou nd m m ice, le m on, hi m OI b oy , av oi d n n ick, fu nn y, gai n N si ng er, ba ng, ba n k Vowels + ® l l ight, ye ll ow, fee l A® b arr ed, f ar, ar m ® r ice, ve r y, b ir d, h er , f ur e® h air , c are d, wh ere w w inter, a w ay i® h ere , w eir d, b eer j y ell, on i on O® b or n, st ore h (^) h ill, a h ead U® (^) t our , m oor Exercises
1. Each word below (as pronounced by a native speaker of any dialect of English) has
one clear mistake in its transcription. Circle the specific part of the transcription where the mistake occurs, and show what the correct symbol(s) (if any) should be.
(Ex: honest, [hAn´st], should be Ø (i.e. nothing); rain, [®AIn], should be e .)
Written: IPA: Should be: Written: IPA: Should be: a. shine [shAIn] ___________ b. beauty [bjuty] ___________ c. wrench [w®EntS] ___________ d. paper [pap®] ___________ e. jumping [jåmpIN] ___________ f. savage [sœvœdZ] ___________ g. user [uz®] ___________ h. shed [Sed] ___________ i. teacher [tich®] ___________ j. his [hIs] ___________
A. Anatomy. The alveolar ridge refers to the gums just behind the upper teeth. The palate refers to the 'hard palate,' i.e. the roof of the mouth. The 'soft palate' is called the velum , and ends in the uvula (this is the fleshy appendage you can see hanging down in the back of your throat). If the velum is raised, this closes the velo-pharyngeal port , preventing airflow between the nasal passages and the rest of the vocal tract. The tongue is a mass of muscle, which we can divide into tip (the only part you usually see), body, and root. The epiglottis is a flap be- low the pharynx (the back of the throat): it covers the trachea (or 'windpipe') when you swallow, so that food goes down the esophagus in- stead. Lastly, the larynx is a sort of valve, encased in cartilage (the 'Adam's apple,' more prominent in males, but present in all humans), at the top of the trachea. It opens wide during breathing (Fig. 2a); closes when you swallow (b); and when you say a vowel, the two sides draw to- gether, so that they vibrate as air passes through (c). This voicing (pulsing of air in the glottis as it passes through the vibrating larynx) is what creates the sound of your voice. B. Consonants. Speech sounds are the result of movements of parts of the vocal tract, particularly the lips, tongue tip, tongue body, and larynx (the major articulators ) which affect the flow of air as you exhale. Consonants are formed with significant ob- struction of this airflow by one or more of the articulators; whereas in vowels , the mouth remains relatively open. We can describe particular types of consonants in terms of how much obstruction is involved ( manner of articulation ). ! S tops ([p,t,k,b,d,g]) involve a complete blockage of airflow, due to full closure at some point in the mouth. ! Nasals ([m,n,N]) involve complete closure in the mouth, but the back of the velum is lowered, allowing the airflow to pass through the velo-pharyngeal port, and out the nose. ! Fricatives ([f,v,T,D,s,z,S,Z,Ł h]) involve a partial constriction in the mouth, such that airflow is forced through a narrow channel, creating a hissing sound. ! Affricates is a term sometimes used for stop + fricative sequences made with the same articulator, including ([tS,dZ]). ! Approximants ([l,®,j,w]) involve less obstruction than a fricative, but more than a vowel. In an [l], the tip of the tongue often makes full contact with the alveolar ridge, Figure 2: Schematic representations of the larynx (not to scale) neck (cross- section from above) trachea a. Adam's apple spine glottis larynx b. c.
but one side of the tongue is lowered: [l] is therefore called a lateral approximant ; the others are central. We can also classify consonants in terms of the state of the larynx ( phonation ) during their pronunciation. ! Voiced consonants ([b,d,g,v,D,z,Z,m,n,N,l,®,j,w]) are accompanied by voicing (Fig 2c). ! In voiceless consonants ([p,t,k,f,T,s,S,h]), the glottis is more open, as in Fig. 2a, so that air passes through without vibrating. Finally, consonants can be described in terms of where the obstruction occurs in the vocal tract ( place of articulation ). ! Bilabials ([p,b,m,w]) involve closure or constriction of the two lips. ! Labiodentals ([f,v]) involve constriction of the upper teeth and lower lip. ! Dentals ([T,D]) involve constriction of the tongue tip and the upper teeth. ! Alveolars ([t,d,n,s,z,®,l]) involve constriction of the tongue tip and the alveolar ridge. ! Post-alveolars (or palato-alveolars) ([S,Z]) involve constriction of the tongue tip and the palate, just behind the alveolar ridge. ! Palatals ([j]) involve constriction of the tongue body and the palate. ! Velars ([k,g,N,(w)]) involve constriction of the tongue body and the velum. ([w] is considered a velar as well as a bilabial because it involves constrictions both at the lips and velum.) ! Glottals ([h]) involve constriction of the glottis (in this case, sufficient constriction to create a fricative, but not enough to cause voicing). These classifications of consonants are summarized in the following chart: Table 6 : Classification of English consonants by manner, place and phonation type bi- labial labio- dental den- tal alve- olar post- alveolar pala- tal velar glot- tal voiceless p t k stops voiced b d g voiceless f T s S fricatives voiced v D z Z h voiceless tS affricates voiced dZ nasals m n N central w^ ®^ j^ (w) approximants lateral l We can thus articulatorily describe [s] as a voiceless alveolar fricative; [N] as a (voiced) velar stop; etc. Likewise, we can refer to the set [b,d,g] as the class of voiced stops. A Feel the buzz! With two fingers on your Adam's apple, say [sssszzzzsssszzzz... ]. You should be able to feel a vibration under your fingers, during the [zzzz] parts only. That 'buzz' is voicing.
Exercises
- Give the IPA symbols for the sounds corresponding the articulations shown in the following diagrams. (Voicing is indicated with a zig-zag line at the larynx.)
- Give the IPA symbols for the sounds with the following articulatory descriptions: a. voiceless glottal fricative _______ b. voiced bilabial nasal _______ c. open-high back rounded vowel _______ d. voiced palatal approx. _______ e. voiced post-alveolar fricative _______
- Give the articulatory description for the following sounds: a. [N] ________________________ b. [j] ________________________ c. [T] ________________________ d. [v] ________________________ e. [e] ________________________
- The following sets of sounds are natural classes, characterized by shared articulatory properties. For each of the sets, identify these properties. Examples: [t,d] are the set of alveolar stops. [m,n,N] are the set of nasals; they are also voiced , but the voiced set includes other sounds as well, so only nasals is correct. a. [i,I,e,E,œ] _____________________________ b. [p,b] _____________________________ c. [®,l,j,w] _____________________________ d. [v,D,z,Z] _____________________________ e. [i,I,u,U] _____________________________ III. Acoustic phonetics A. Fundamentals of sound. Speech sounds can also be understood in terms of their acoustic properties, i.e. properties of the soundwaves. Soundwaves are simply waves of fluctuating air pressure, radiating out from their source. It is the structure of these waves which distinguishes one sound from another. In a pure tone (approximated by the sound of a tuning fork) these ripples of air pressure correspond to a simple sine function, where the x-axis is time , and the y-axis is pressure. Such a wave has a particular frequency , measured in Herz ( cycles per sec- ond ): the higher the frequency, the higher the sound is in pitch. The sine wave in Fig. 4a a. b. c. d.
has a little over 3 cycles per 10 milliseconds, or 300 cycles per second, i.e. 300 Hz.^4 Moreover, the more extreme the fluctuations in pressure, the greater the amplitude of the wave (measured in decibels ), and the louder the sound. In comparison to Fig. 4a, the wave in Fig. 4b is higher and quieter. If we sum the two waves above, the result is a complex waveform (4c). The more individual sine waves we combine, the more com- plex the resulting waveform. Figure 4 a. A sine function: b. A higher frequency, lower amplitude wave c. A complex waveform, the sum of (a) and (b) d. Actual waveform of the author's voice, saying [A] The sound signals of speech are always complex waveforms (see Fig. 4d). But just as we can sum simple sine waves to yield the complex wave in Fig. 4c, we can also take a complex waveform and break it down into simple waves, each with its own fre- quency and amplitude (a mathematical technique called Fourier analysis ). The lowest- frequency component of the waveform is called the fundamental frequency ( F0 ), which (^4) That's roughly D below concert A, for you musicians. 1 cycle time (in milliseconds)! pre ssure amplitude (in decibels)
B. Vowel cues. Vowels are acoustically distinguished principally by the frequen- cies of the formants. ! The higher the vowel articulatorily, the lower the F1 frequency. ! The backer the vowel, the lower the F2 frequency. ! Lip rounding further lowers F. ! The formants smoothly change in frequency during a diphthong , from the values of the first vowel to those of the second. C. Approximant cues. Approximants are similar in cues to vowels. ! [w,j] are very similar in their formant frequencies to the high vowels [u,i] respec- tively, but a bit shorter in duration, with a slightly lower F1 , and a slight weakening of the higher formants, particularly in [w]. ! [®] is similar in formant frequencies to [´], but with low F. ! [l] is similar to [®], but with high F. D. Fricative cues. Up to this point, we have focussed on periodic (humming) sounds. Periodic sounds, such as the vowel shown in Fig. 4d, have a repeating pattern to the waveform. Fricatives, however, involve aperiodic ( hissing or crackling) noise. Note in Fig. 7 the fricatives [s] and [S], which look like charcoal smudges, vs. the vowels [u] and [i], which have clear vertical striations and clear formants. Although they have no fundamental frequency, aperiodic signals can be stronger in some frequencies and weaker in others. ! The alveolar [s] has almost all of its noise above 4000 Hz ([z] too), whereas the post- alveolar fricative's noise extends down to 2000 Hz. ! Voiced fricatives are gener- ally shorter than the voiceless ones, and may have a band of voicing striations along the bottom of the spectrogram. ! T h e o t h e r f r i c a t i v e s ([f,v,T,D,h]) are all much qui- eter than [s , S , z , Z ]. The labiodentals ([f,v]) are typi- cally slightly louder than the interdentals, with more noise below 4000 Hz. [h] has bands of aperiodic energy in the same frequency regions as the formants of adjacent vowels. E. Stop cues. The complete articulatory closure in a stop results in an interval of silence, which shows up as a blank column on a spectrogram, followed by a brief burst of aperiodic noise when the closure is released (see the [k] and [t] in Fig. 6). The stops are Figure 7 : Spectrogram of [suSi] [ s u S i ]
distinguished from each other by movement of the formants before and after closure (formant transitions) and by properties of the burst. ! In bilabial stops, all formants move down- ward heading into the closure, and upward coming out of the closure. ! In alveolar stops, F2 heads towards a fre- quency of around 1800 Hz moving into clo- sure, and originates from the same frequency coming out of closure. The release burst has considerable energy above 4000 Hz (note the burst after the [t] in Fig. 5). ! In velar stops, F2 and F3 move toward each other heading into closure, and split apart coming out of closure. Velars also frequently have a double burst note the two verti- cal 'blips' of noise after the [k] in Fig. 5). ! Voiced stops are shorter than voiceless stops, and they may have a narrow band of dark striations (a voicing bar) at the very bottom of the spectrogram. Voiceless stops, particularly in English, have a delay between the release burst and the start-up of full voicing in the following vowel. F. Nasal cues. Nasals are acoustically somewhat like approximants, and some- what like stops. ! They resemble approximants in that one can see formants and voicing striations dur- ing the whole consonant. Nasals have a low F1, and a marked weakening of the higher formants. ! Like stops, identification of the nasal's place of articulation depends on formant tran- sitions, into and out of closure. The transitions in a bilabial nasal are similar to those of a bilabial stop; likewise for nasals and stops at other places of articulation. Exercises
- Fill in the blanks: a. A soundwave without a repeating pattern is _____. b. A graphic display of sound showing changes in formants over time is a _____. c. The basic pitch of a voice is its _____ frequency. d. Voicing appears on a spectrogram as _____. e. Stops appear on a spectrogram as _____.
- Fill in the blanks: a. The differences between [s] and [Z], as they appear on a spectrogram, are _____. b. The differences between [m] and [N], as they appear on a spectrogram, are _____. c. The differences between [e] and [o] , as they appear on a spectrogram, are _____. d. The vowel [i] has _____ F1 frequency and _____ F2 frequency (high or low). e. The vowel [A] has _____ F1 frequency and _____ F2 frequency (high or low).
- The following spectrogram contains two single-digit numbers of English. What are they (in order)? To help you, dotted lines are drawn between the sounds, a pause be- tween the words is marked, and F1-F3 are highlighted. Figure 7 : Formant transitions bilabial alveolar velar 1800 Hz
around us (not necessarily those of our biological ancestors), our ability to zero in on the particular set of sound properties which are relevant for our language can't be attributed to our genes, like hair colour. Important aspects of this mental speech system have to be learned_._ Indeed, this is a crucial part of learning a language fluently. But in the case of a first language, we seem to pick up this knowledge within a few years of birth, without any formal instruction – in fact, without much conscious thought at all. In the remainder of this chapter, we focus on the observation that lan- guages obey phonological rules – rules concerning what sounds may occur in the language, and how these sounds may be put together to form words of the language. Have you ever overheard someone speaking a language you don't understand – never- theless you've been able to recognize the language as French, Spanish, Chinese, etc.? How can you identify a language without being able to recognize any of the words? The answer is that you're recognizing the phonological rules which characterize the language. A plausible hypothesis is that phonological rules arise as particular languages' re- sponses to this problem of maintaining recognizable words despite variation.^5 Consider the fact that language sound systems (henceforth ' phonological systems ') must be able to convey a broad range of information, with a minimum of confusion, for a broad range of speakers and hearers, across a broad range of situations. This practical consideration in- troduces two important constraints on phonological systems: Ease of perception : recovery of meaning must not depend on cues which are ! subtle, i.e. difficult to hear, nor ! unstable, i.e. not always present in the signal, nor ! singular, i.e. differences in meaning are not supported by multiple cues; misper- ception of just one cue could result in confusion of meaning. Ease of articulation : recovery of meaning must not depend on cues that require highly effortful or precise articulations. (^5) Although this hypothesis is accepted, in some form, by most phoneticians and phonologists, the question of how directly phonetic pressures constrain phonological systems, and whether some aspects of phonological systems are independent of these phonetic constraints, are topics of debate in current linguis- tic theory. So what do you know? If you speak English fluently, you must already 'know' the phonological rules of English. But how can you 'know' something that you've never even thought about before? Actually you know a great many things, without being at all conscious of that knowledge. For example, you proba- bly know how to pick up a carton of milk, a complex task requiring nearly instantaneous assessment of the weight of the milk vs. the strength of the container, so that you neither drop nor crush it (robots are terrible at this task). But humans do this without conscious thought; and it is difficult to put this knowledge into precise words. Psychologists call this 'implicit knowledge.' Speakers' knowledge of the phonological rules of their language is likewise implicit. We're generally unaware of these rules (outside of linguistics courses). But we instantly detect violations of these rules, e.g. in speech with a foreign accent, or in computer-synthesized speech.
Each language develops its own particular set of rules, as strategies for satisfying these constraints. This is not to say that anyone ever sat down and consciously designed a phonological system. Rather, these systems continually evolve, through the back-and- forth of communication, and miscommunication, between speakers and hearers – includ- ing young children learning the language. Moreover, these rules are not prescriptive rules, which speakers are explicitly taught that they should obey (e.g. don't say 'ain't' ): speakers follow these rules without even thinking about them. Indeed, it requires careful analysis, and some understanding of phonetics, to be able to figure out what the rules are
- even for one's own language. V. Phonemes and allophones A. Allophonic variation. Because the organs of the vocal tract generally move in smooth trajectories rather than abrupt jerks, sounds are inevitably influenced by the sounds around them. As a case in point, consider the influence of nasal consonants on preceding vowels in English. In words such as ran , doom , or sing , the velum begins to lower, opening the nasal passages, well before the oral closure in the nasal consonant be- gins. This results in a significant part of the vowel being nasalized. This sort of overlap in movements of the articulators is called coarticulation. In a narrow transcription, these examples should therefore be transcribed as [®œ)n], [du)m], [sI)N] ([~] is the IPA diacritic , or supplementary symbol, for nasalization). We thus have two different sets of vowels in English: ! nasal ([i),u),I),U),e),´)),o),E),O),œ),å),A)]), and ! oral ([i,u,I,U,e,´,o,E,O,œ,å,A]). Weird phonology: A language might consist of nothing but sequences of [f] and [T], where [fTTfT] means 'dog', [TffTTf] means 'cat', etc. It might be 'spoken' by singing particular sequences of exact pitches. More imaginatively, Kurt Vonnegut's novel Slaughterhouse Five presents a race of extraterrestrials who communicate by tapdancing and making other bodily noises. Clearly, none of these is remotely like a real human language. But what's the dif- ference; and why hasn't any human society ever developed anything like them? A plau- sible answer is that these imaginary systems seriously violate Ease of Perception or Ease of Articulation. The [fTfT] language depends upon accurate perception of quiet fricatives, which are easily confused with each other, and easily masked by background noise. Vonnegut's alien language would be considerably more strenuous (for earthlings, at any rate) than speech. And the singing language would require all speaker/hearers to have perfect pitch (in perception) and flawless intonation (in production). On the other hand, language doesn't need sound at all. Sign languages (princi- pally used by deaf communities) are complete human languages, independent of the sound-based languages of the societies around them; and they are sight- rather than sound-based. Nevertheless, sign languages are subject to similar functional constraints: they avoid signs which involve extreme physical exertion or dexterity (e.g. walking on one's hands), or which require perception of extremely subtle gestures (e.g. a twitch of the calf muscles).
