Assuming no prior knowledge, books in the series offer an accessible overview of the subject, with activities, study questions, sample analyses, commentaries and key readings — all in the same volume.
This edition has been completely reorganised and new features include: updated descriptions of the sounds of modern English and the adoption of the term General British GB ; considerable expansion of the treatment of intonation, including new recordings; and two new readings by David Crystal and John Wells. Written by authors who are experienced teachers and researchers, this best-selling textbook will appeal to all students of English language and linguistics and those training for a certificate in TEFL.
Intended for the absolute beginner, it requires no previous background in linguistics, phonetics or phonology. Starting with a grounding in phonetics and phonological theory, the book provides a base from which more advanced treatments may be approached. It begins with an examination of the foundations of articulatory and acoustic phonetics, moves on to the basic principles of phonology, and ends with an outline of some further issues within contemporary phonology.
Consider the words in [2. When two sounds can be used to distinguish word mean- ings in a particular language they are said to be separate phonemes. On the basis of [2. Supply examples of English words with different meanings which only differ in that where one word has [1] the other has [r], all the other sounds being exactly the same. Are [1] and [r] separate phonemes in English?
The pairs in [2. Sounds are classified as separate phonemes if they are responsible for a difference in meaning in a minimal pair. The forms in [2. Using this principle, sounds are isolated as belonging to separate phonemes if they occur in phonetically very similar, though not identical environments provided that the differences between them cannot be reasonably attrib- uted to the influence of neighbouring sounds see Chapter 5.
Admittedly, there are bound to be occasional differences of opinion between phonologists as to what constitutes adequate similarity to justify labelling environments as 'analogous'. Phonological analysis is not an exact science. Here, I shall avoid getting bogged d o w n in controversies and simply illustrate the principle of contrast in analogous environments with an example from E w e , a Ghanaian language.
The difference between nasalised [e] on the one hand and oral [e] and [o] on the other cannot be the reason for the difference between voiceless [f] and voiced [v] since nasal [e] as well as oral [e] and [o] are all voiced and should not affect in different ways the voicing of neighbouring sounds. A practising phonol- ogist can normally safely assume that sounds like [n] and [x] or [1] and [p] which show no phonetic resemblance are distinct phonemes.
A few examples should clarify this point. If you look back at the data in [2. Likewise, in the Ewe example, the contrast in analogous environments test is worth attempting because [fj and [v] are both labiodental fricatives, only distinguished by voicing: they could easily be allophones of the same phoneme. Hence sounds that are physically different may be regarded as functionally the same in a particular language. Furthermore, as we shall see presently, functionally it makes sense to think of the changes in the various phonetic variables which make up a sound as happening in an abrupt and synchronised fashion, even though physically they are gradual and they overlap each other.
That is to say, although in a word like inn, it is true that the soft palate will begin to be lowered before the tongue makes contact with the alveolar ridge, there is no functional significance in the fact that the latter part of [i] in inn will be progressively nasalised while the [i] of if is wholly oral. The English- speaking hearer will not perceive the difference between the two [i]'s.
To such a hearer inn consists of just two sharply- separated sounds: oral [i] followed by nasal [n]. In view of this, psychologically we are justified in thinking of speech as segmented along the time dimension, even though physi- cally the segments blend into one another - the situation is very much analogous to that of cursive handwriting.
I refer here to the way that alphabetic transcrip- tion traditional orthography and phonemic transcription alike treats successive segments of speech-sound as single atomic elements, transcribed with one symbol each, rather than as bundles of simultaneous elements, each element being a value of a particular phonetic variable. Clearly, speech sounds are not indivisible atoms. However, for convenience in reading and writing, it is advantageous to use a code which represents a whole bundle of such phonetic properties using a single symbol, rather than having to represent a word like inn in a two- dimensional notation, with a table showing articulatory parameters as in [2.
One reason for saying that the internal structure of phonemes is in all probability relevant to the way in which they function phonologically has to do with the tendency to SYMMETRY found in the phonological systems of numerous languages. Consider vowel systems, for instance.
They may be very basic, or more elaborate as you can see in [2. Arabic, e. Spanish, e. About the simplest system of vowel phonemes common- ly found it occurs in Aleut Eskimo and Arabic, for instance has just three members. Many other languages, such as Spanish, Czech, Burmese, and Swahili, distinguish three degrees of aperture and have a five-member vowel system.
Yet others, like Italian, Kikuyu and Efik have four degrees of aperture and a seven-member vowel system. All these systems are triangular: there is a lone low open vowel and at least one pair of non-low vowels. Azerbaijani, British Columbian French, Persian and English exemplify with different embellishments the other common type, namely a quadrangular vowel system which may be represented as [2. Thus, for instance, four-member vowel systems such as those in [2. If we consider just stop consonants, for simplicity, English has: [2.
In Siriono there are no voiced alveolar, velar and palatalised velar stops corre- sponding to voiceless stops at these places of articulation. Consequently phonological systems tend to be symmetrical. Hence, s y m m e t r y is an outward sign of the underlying dimension of contrast in phonology. Another c o m m o n factor encouraging phonological s y m m e t r y has to do with permissible combinations of phonetic properties in particular languages.
In the vowel systems surveyed above, all the vowels were either front and unrounded spread or back and rounded. The prohibition of the combination of rounded- ness with frontness and backness with unroundedness in vowels is by no means universal.
There exist languages such as French and Swedish in which it is permissible for front vowels to be rounded and other languages such as Ivatan, spoken in the Philippines, and Vietnamese in which back vowels may be unrounded. Thus, in French, the vowel in words like tu, vu and lu which is represented by the phonetic symbol [y] is both front and round.
And again, as you can see in [2. Before you proceed, rank the vowels on a scale indicating the relative height of the highest point of the tongue during the production of each one of these vowels. You may use [2.
This account might look plausible for Czech, but it would not do in those situations where asymmetry is unre- lieved, even by borrowing. Thus, while French permits front rounded vowels although they are unusual, Viet- namese forbids them but permits back spread unrounded ones which are no less unusual.
Interestingly, there are only two back unrounded vowels in Vietnamese where symmetry would lead one to expect three: [2. Such counter-examples notwithstanding, the import- ance of phonological symmetry as a tendency rather than an absolute rule is very well established. Yet, if phonemes were regarded as unitary, indivisible atoms, with their internal phonetic make-up irrelevant to their functioning, this tendency would seem quite inexplicable.
If a language, say English, were to possess twenty-four consonant phonemes, one would suppose that any selection of twenty- four consonant sounds would do as well as any other, and it would be a remarkable coincidence if the twenty-four actually used in the language manifested any particular symmetrical patterning. It would be even more remarkable if the types of symmetrical patterns accidentally stumbled upon by one language, say English, just happened to be the same as those stumbled upon by numerous unrelated languages.
As we have seen, this is indeed the case: the symmetrical patterns found in the phonological systems of the languages of the world appear to come from a very restricted range. Members of a language-community have to learn to produce and perceive the psychological 'target-value' of various phonetic dimensions which their language uses to distinguish word meanings: for instance, the Arabic speaker has to learn to make two distinctions on the dimension of vowel aperture and the Spanish speaker three.
Referring to the International Phonetic Alphabet in [1. What advan- tages has the symmetrical system got over the asymmetrical one? I hope you chose the system in [2. There, every stop has a corresponding fricative, every voiced oral stop has a corresponding nasal one, and every voiced stop and fricative has a corresponding voiceless one: [2. As [2. On the other hand, the consonant inven- tory in [2. Furthermore, you should note that while the asym- metrical inventory can be defined only by imposing complicated regulations to ensure that only segments with permissible combinations of phonetic properties are gener- ated e.
One probable consequence of this is that a symmetrical system is easier to learn than an asymmetrical one - even when, as in our example the latter contains fewer elements. Predictably, for the most part phonological systems tend to be symmetrical, using in an optimum way phonological properties such as voice, alveolar and fricative which speech- sounds are made up of. As we have already noted, it is unusual, but not unheard of, for a language to have an asymmetrical conso- nant system like the one in [2.
An example is given below in [2. It is these phonetic properties rather than phonemes that are the basic building blocks of phonology. Before we delve any deeper into phonology we shall explore the nature of these phonetic and phonological properties. That is the subject of the next chapter. Exercises 1. The pairs of consonants listed below are separate phonemes in English.
Provide minimal pairs showing how these sounds contrast word meaning. Study the data below and identify the phonetic context in which the voiced and voiceless member of each pair occurs. State the rule that governs the distri- bution of each sound. There exists a very rich literature on the phoneme. Standard textbooks like Hyman and Sommerstein also contain good introduction to the literature. To those who want a more detailed account of the devel- opment of phoneme theory and phonological theory in general , I recommend Fischer-J0rgensen and Anderson This view is held by those who regard the phoneme as a psychological entity.
Badouin de Courtenay is probably the best known holder of this view. He defined the phoneme as 'a mental reality, as the inten- tion of the speaker or the impression of the hearer or both' Twaddell 56 see page 67 below. We saw at the end of Chapter 2 that phonological systems tend to be symmetrical and that a limited number of phonetic parameters, taken from a fairly small universal set recur in a variety of combinations in different languages.
Besides introducing you to distinctive features, this chapter will also serve as a restatement in a slightly different form of the principles of phonetics which were outlined in the first chapter. There is nothing mysterious about the fact that there is a relatively small inventory of phonetic features from which languages select different combinations to construct their individual phoneme systems.
As all members of the human race are endowed with very similar articula- tory and auditory capabilities, it is only to be expected that they will only be able to produce and utilise speech sounds built up from the set which is pre-determined by their biological endowment.
The position presented in this chapter regarding distinctive features has not always been accepted by all students of phonology. At one time some scholars, like Bloomfield , would not agree with the claim that the phoneme is not the most basic phonological element.
Bloomfield thought that there were no phonologically relevant particles more elementary than the phoneme, although he was aware that on purely phonetic grounds the phoneme can be decomposed into more basic particles.
Firstly, even a cursory inspection of the phonology of any language will reveal that the phonological behaviour of phonemes is largely determined by the phonetic features which they are made up of.
It undergoes partial devoicing when i m m e d i - ately preceded by voiceless stops but not when it is preceded by voiced ones: [3. By highlighting each ar- ticulatory parameter and singling out the sub-phonemic particles i. Examine the American English data in columns A and B in [3. The SPREADING of nasality to the preceding vowel is due to the premature, anticipatory lowering of the velum, as the vowel is being produced, to let air escape through the nose during the articulation of the nasal consonant.
Distinctive features thus facilitate the statement of assimilation processes by high- lighting the various separate gestures involved in the production of speech. No equally natural way of stating assimilation processes is available if phonemes are treated as unanalysable units. An added advantage of the feature approach is that it enables us to highlight the internal structure of a sound. When we do that, it soon becomes obvious that phono- logical segments have internal structure. Sounds are not bundles of unordered, unstructured phonetic properties.
A simple example like [3. Similar evidence of internal structure is to be found in diphthongs. In words like way [wei] and why [wai] the vowel sound has two distinct vowel qualities, a fact which is reflected in this case in the way in which the two phases of the sound are transcribed.
The same point can be made about affricates like [ts], [dz], [tj] and [d3]. In many languages, affricates behave in part as though they were stops and in part as though they were fricatives.
Thus we can see that the phoneme is not an indivisible phonological unit. Thus, it would be bizarre for any language to have a rule nasalising vowels before the following assortment of consonants: [d s k t h], This is a ragbag of sounds which are phonetically very different from each other and which all lack the crucial property of being nasal which is passed on to an adjacent vowel during nasal- isation. The chances of such an arbitrary nasalisation rule existing in any language are extremely remote.
On the other hand, a rule which nasalises vowels in the neighbourhood of nasal consonants as in [3. The nasal consonants [m n rj] which condition the nasalisation form a natural class and they all contain the crucial feature of nasality which triggers off the nasalisation of the preceding vowel. Normally, sounds which are phonetically similar display similar phonological behaviour. In order to state the basis of the similarity between a group of phonemes, it is necessary to penetrate beyond the phoneme and scrutinise the phonetic features which they share.
When the Nazis occu- pied Czechoslovakia, Jakobson fled to New York and later moved to Harvard where he introduced some of the ideas of the 'Prague School' to the Americans. The key publications are Trubetzkoy and Jakobson, Fant and Halle But, in word-final position, [k] and [g] do not contrast.
There, the opposition between them is neutralised or suspended and only the voiceless [k] occurs. That is the situation in the singular form of the noun : 'horn' and 'juice' are pronounced as [rufe] and [sofe] respectively Kenstowicz and Kisseberth Neutralisation is one of the key concepts in phonology.
We shall come back to it in Chapter 8. The initial proposal was that just a dozen acoustically defined pertinent contrasts other than prosodic ones, involving such things as tone and stress which are usually associated with an entire syllable or word would be found in all languages. Jakobson and his co-workers further hypothesised that the presence of certain oppositions in a language precludes the existence of other oppositions.
For instance, they suggested that no language phonemically contrasts labialised consonants i. Although labialization and pharyngealisation are distinct as far as articulation is concerned, they are phonologically merely implementations of the same acoustic distinctive feature FLAT. Jakobson et al. For example, to discriminate between the words in [3.
Take the case of a language which distinguishes four degrees of vowel height and has the vowels [i e e a] or a language that distinguishes four places of articulation and has the bilabial, alveolar, palatal and velar nasal consonants [m n ji rj]. Nevertheless, the dominant view among phonologists, and the the view adopted in this book, is that although the case for multivalued features is not without merit, from a phonological point of view, distinctiveness is the paramount consideration and its inter- ests are best served by binary distinctive phonological features; for on every occasion, the hearer has to make a decision as to whether the speaker has said either this word or that word.
True, the phonetic features which specify the phonetic realisation of a given feature, say voicing of stops in English, could be put on a multivalued scale like: [3. In principle we could indicate the amount of voicing by using an infinite number of points on the voicing scale. Phonologically, however, this kind of low level detail is overlooked. Phonetic features may be multivalued but phonological features need not be.
As a rule, it is a dichotomous choice. It was discovered that the model was too parsimonious. The dozen or so features which it allowed were insufficient to account for all phono- logical contrasts found in the languages of the world. Furthermore, it was criticised for using the same phono- logical feature to characterise phonological oppositions which in some cases were manifested by different phonetic properties.
For instance, if a sound was described as GRAVE which means that most of the acoustic energy used in its production is concentrated in the lower part of the spectrum you would not be able to tell whether it was a labial like [p], or a velar like [g], since the acoustically defined phonological property GRAVE could be correlated with either labial or velar articulation. Because of these and other inadequacies, Chomsky and Halle in their book The Sound Pattern of English henceforth SPE proposed a major revision of the theory of distinctive features.
They replaced acoustically-defined phonological features with a set of features that have, in most cases, articulatory correlates. Furthermore, the number of features was also substantially increased. But, like their original Jakobsonian precursors SPE features remain binary. The list of distinctive features given below is based on SPE in the main, but it incorporates some of the modifi- cation that have been proposed since It is not important to 'master' all the details of distinc- tive feature theory at this stage.
They are described mainly in order to show how the system works. But you should come back to this chapter ot refresh your mind as the need to use features arises in later chapters. Obstruents, nasals and liquids are consonantal; vowels and glides" are nonconsonantal. Normally, syllabic sounds are auditorily more salient than adjacent nonsyllabic sounds. Vowels are syllabic and so are syllabic consonants such as [J] in bottle and candle or the nasal [n] in cotton and [m] in bottom.
In other words,the unmarked normally expected and natural state for sonorants is to be voiced, while for obstruents the unmarked state of affairs is to be voiceless. Vowels, nasals and liquids are sonorant; stops, fricatives and affricates are obstruents. They specify where in the vocal tract modifications of the airstream take place in the production of particular sounds.
Dental, alveolar, alveo-palatal, retroflex and palatal sounds are coronal; labial, velar, uvular and pharyngeal consonants are noncoronal. Labials, dentals and alveolars are anterior while all other sounds are not. In the literature the alternative feature R O U N D is often used to refer to many of the sounds which can also be described as labial. Rounded sounds are produced with a pursing or narrowing of the lip orifice. Rounded sounds like [o] [u] and [w] are a subset of labial sounds; consonants like [p b m] are labial but not round.
Labial sounds include bilabial and labiodental consonants as well as rounded vowels. All other sounds are nonlabial. In apical sounds the tip of the tongue makes contact with the front teeth or the alveolar or alveo- palatal regions of the roof of the m o u t h while in laminal sounds it is the blade of the tongue that makes contact with those same areas.
This feature also distinguishes labial from labiodental sounds. The following sound types are distributed: bilabial fricatives like [ j 6] lamino- alveolar fricatives like [s z] lamino- alveo-palatal fricatives like [J 3].
Other tongue configurations are regarded as departures from that norm. High sounds include vowels like [i u], the glides [w j ] , alveo-palatal, palatalized, palatal and velar consonants.
All other sounds are nonhigh. All other sounds are nonlow. Sounds produced with the body of the tongue either in neutral position or pushed forward are nonback. This feature distinguishes between back vowels like [u o a] and front vowels like [i e e]. Because the feature system is binary, there is no way of showing that certain vowels are neither front nor back; central vowels are grouped together with back vowels. This does not appear to be entirely satisfactory on purely phonetic grounds.
Of the consonants, velars, uvulars and pharyngeals are back while labials, dentals, palatals as well as glottals are nonback. While in English clicks are not fully-fledged speech sounds which can combine with other sounds to form words, in some languages which happen to be almost exclusively found in Southern Africa clicks are used as regular speech sounds. The production of a click involves two crucial phases.
Secondly there must be some constriction in the labial, dental, alveolar or alveo-palatal area. The resulting sound is a click. The soft palate and the back of the tongue are the ini- tiators of the movement of air and the direction of the airflow is inward. Clicks are velar suction sounds; other sounds are not.
Zulu has a dental click [j], an alveolar click [rj] and an alveo-palatal click [c]. The shape of this tube can be modified by rounding the lips and making them protrude - and thus elongating the tube. Alternatively, the tongue root position can be adjusted by pushing it forward or retracting it so that the vocal tract is either lengthened or shortened.
Either of these actions has the effect of modifying the shape of the resonating chamber in the vocal tract in much the same way as differences in size and shape of wind instruments affect the notes which they produce. Vowels like [i e o] in many West African languages are made with the tongue root pushed forward while [i c o] are made with the tongue root in neutral position. I quote Halle and Clements who give this cautious definition: 'Tense vowels are produced with a tongue body or tongue root configuration involving a greater degree of constriction than that found in their lax counterparts; this greater degree of constriction is usually accompanied by greater length.
Tense vowels vs. We note that this feature and the last [ATR] are not known to co-occur distinctively in any language and may be variant implementations of a single feature category.
This gesture, which is associated with voiceless- ness and aspiration, is absent in nonspread sounds. Spread sounds include aspirated stops; murmured and breathy voice sounds, voice- less vowels and voiceless glides.
All other sounds are nonspread. They are produced with a severe obstruction of the glottis which is made using the vocal cords. This inhibits or prevents the free vibration of the vocal cords. No such gesture occurs in the production of nonconstricted sounds. All other sounds are nonconstricted. In the opening chapter of this book we observed that English only uses speech sounds made with the pulmonic airstream mechanism, but there are languages which employ additional mechanisms.
Here I shall briefly describe the remaining airstream mechanism found in languages. To understand how it works, I suggest that you begin by making a big constriction in your throat - pretend that you are just about to cough.
Get your vocal cords firmly together and shut off the glottis. At this stage, an upward movement of the larynx will push out the air above the larynx while a downward movement of the larynx will suck in air and loosen the constriction; the vocal cords will vibrate vigorously as the air flows in. Ejectives are voiceless. An example of an ejecdve is the [t'] in Tlingit, a language of Alaska, in the word [t'ij] 'elbow'.
The air drawn into the larynx as it is pulled downward normally causes heavy voicing. An example of this is the bilabial stop, in the word [6a6a] 'father' in Shona Zimbabwe. Affricates, nasals and oral stops and laterals are noncontinuant.
All other sounds are continuant. In nonlateral sounds air flows out through the centre of the mouth. The term 'liquid' is conventionally used to refer to T and 'r'-like sounds.
Languages may have lateral sonorants, frica- tives and affricates made at various places of articulation. Oral sounds are produced with the velum raised so as to block access to the nasal cavity and to allow air to go out only through the mouth. Nasal sounds include nasal stops like [m n H, ji rj] which are made with complete blockage of air at the place where the articulators meet as well as nasalised consonants, glides and vowels. Start by pressing the button below!
An Introduction to Phonetics. Read more. Introducing Phonetics and Phonology. To browse Academia. Skip to main content. You're using an out-of-date version of Internet Explorer. By using our site, you agree to our collection of information through the use of cookies. To learn more, view our Privacy Policy. Part 3, Auditory Phonetics, covers the anatomy of the ear and the perception of loudness, pitch and quality.
The final part, Part 4, covers the articulatory production of speech, and shows how experimental techniques and tools can enhance our understanding of the complexities of speech production. Though the audience for this book is mainly students and professors in the Speech Sciences, it will also be valuable to any students studying hearing science and acoustics.
The book is well supported with figures, tables, and practice boxes with experiments. The Sounds of Language provides an accessible, general introduction to phonetics with a special emphasis on English. Focusing on the phonetics of English, the first section allows students to get an overall view of the subject. The discussion is arranged so that students can read only the RP or GA portions, if desired. Sixteen additional accents of English spoken around the world are also covered to provide students with wider international coverage.
0コメント