Chapter IV: Phonology and Sematology (2)

Other vowel-sounds which may be noticed are the _e_ of the French _prêtre_, German _väter_, whose natural pitch is made G″ or D‴, the closely related open _e_ (_ä_) of the English _pair_, the short _a_ of English closed syllables like _hat_ or _happy_, the short _e_ of the English _men_, and the short _i_ of the English _hit_, _pill_. These short vowels are in great measure due to the little use made of the lips in articulation, and the compensatory exercise of the tongue, which characterize modern English. It is small wonder that we experience so much difficulty in pronouncing _ö_ and _ü_, when even our _u_ is uttered with lips scarcely at all rounded. On the other hand, whenever we find these sounds in a language, we may conclude that we have to do with a speech which gives the lips their full share in articulation. Sievers would call those vowels _passive_ in which all the organs of speech needed for their clear pronunciation are not brought into play, fully pronounced vowels being termed _active_.[153]

The same lazy pronunciation of cultivated English which has almost dispensed with the service of the lips is the cause of the increasing preponderance of the so-called neutral vowel heard in such words as _but_, _virtue_, _dove_, _bird_, _oven_. Except in affected pronunciation we may detect it in most unaccented syllables, especially if they happen to be final; thus we have _diligĕnce_, _muttŏn_, _ăgainst_, _finăl_, _evĭl_, _valuăblĕ_. So, too, as Professor Max Müller remarks, “_town_ sinks to _Paddingtŏn_, _ford_ to _Oxfŏrd_.” He believes it to be pronounced with non-sonant or whispered breath.[154] Mr. A. J. Ellis would make it voice in its least modified form; and Mr. Sweet regards it as a mere voice-glide. The “indistinct” vowel heard in Arabic words by travellers seems to be identical with it. Its existence in a language is a sign of age and decay; meaning has become more important than outward form, and the educated intelligence no longer demands a clear pronunciation in order to understand what is said. The participation of all the organs of speech in the creation of vowel-sounds is, on the contrary, a mark of linguistic freshness and youth. When we find both tongue and lips equally active in the formation of _u_ and _i_, we may feel pretty sure that we are in the presence of an uncultivated dialect. Vowels formed by combining the position of the tongue required for _u_ with that of the lips required for _i_ are extremely rare in Aryan speech; an exceptional instance is to be met with in the Russian _jery_ (_y_).

But we must never forget the infinite capability of modification possessed by a vowel. The same vowel-sound of the same word is not only apt to be pronounced differently by two natives of the same country, but even by the speaker himself at different times, particularly if his attention has been directed to his pronunciation of the sound in question. It is true that the shades of difference between the sounds may be so fine as to escape all but the specially trained ear; but this does not prove them to be any the less real. Putting aside quantity, accent, emphasis, or accidental alteration in the vocal organs, it is difficult to pronounce the same word twice over in exactly the same way, so far, at least, as its vowels are concerned. It is not wonderful, therefore, that it is in their vowels that dialects soonest and most easily alter, and that the vowel-system is the best guide in mapping out the several stages in the history of a language. Of course the character of a vowel-sound is materially affected by its position in a word, or by the consonants with which it is associated; the pronunciation of the same vowel varies in a closed or an open syllable. Long and short vowels, too, differ not only quantitatively, but qualitatively also. Every vowel has both its own peculiar pitch and a pitch dependent on the length of the vocal chords. The peculiar pitch is the result of the resonance-chamber in which the vowel is formed. The high pitch of _i_ is due to the narrow air-passage in the front of the mouth in which it is produced, while the lowered pitch of _a_ and _u_ is caused in the one case by the greater size of the resonance-chamber, and in the other by the narrow opening of the lips. The same pitch may be produced by different modifications of the same resonance-chamber. Thus the French _eu_ in _fleur_, produced by slightly raising the front part of the tongue and rounding the lips, has the same pitch as the English _e_ in _err_, produced without any rounding of the lips at all.

But we have not yet finished with the vowels. The mouth is not the only agent concerned with their production. Brücke[155] asserts that the bones of the skull itself participate in the vibration caused by the utterance of the high-pitched vowels. However this may be, the larynx, the posterior wall of the pharynx, and the _velum pendulum_, or soft palate, with the uvula attaching to it, have all to do with the creation of vowel-sounds. Czermak has proved by experiment that the _velum pendulum_ changes its place with each vowel that is uttered, rising successively for the pronunciation of _a_, _e_, _o_, _u_, and _i_. The nasal orifice, too, is closed during the pronunciation of some vowels, and more or less open during that of others. _A_ and _e_ were the only two vowels which a young man named Leblanc, whose larynx was completely closed, was able to utter; while, on the other hand, experiment has shown that with _i_, _o_, and _u_ the passage to the nose is shut, slightly open with _e_, and considerably open with _a_. From this it will be seen that the term “nasal vowel” is a misnomer. Nasal vowels, in fact, are produced by dropping the uvula, and so allowing the air to vibrate freely through the cavities which connect the nose with the pharynx. So far from a passage of the air through the nose being necessary, we may even increase the nasal twang by stopping the nostrils. The strength of the nasalization depends on the distance of the _velum pendulum_, or soft palate, from the tongue; and in languages like French, in which much use is made of nasalized vowels, the vowel is frequently followed by a true guttural nasal. It has often been noticed that French, in spite of its strong tendency to nasalize the vowels, has no nasalized _i_ or _u_. The cause of this deficiency is very simple. A nasalized vowel requires a free passage for the air from the pharynx to the nose; but this is rendered almost impossible in the formation of _i_, where the tongue is raised so high as to send most of the air through the mouth however much depressed the _velum_ may be, as well as in the formation of _u_, where the tongue is pushed backward towards the soft palate itself. A nasal _i_, however, occurs in Portuguese, and probably also in the Sanskrit _simha_, “lion.”

Every vowel-sound, then, demands three main conditions for its production—the exspiration of air from the lungs, the vibration of the vocal chords, and the formation of a chamber of resonance by the organs of speech. The three conditions must co-exist if we are to have a simple vowel of definite quality, though the exspiration of air need not last beyond the moment at which the vowel-sound is formed. But the position of the organs of articulation both before and after its formation occasions important differences in the manner in which it is introduced or ceases to be heard. In quick and lively utterance, the energy with which the stream of air is emitted makes it difficult for each exspiration to be exactly simultaneous with the corresponding vibration of the vocal chords, while if the exspiration is weak, the vocal chords are apt for a moment not to vibrate. In order to give the chords on the one side the resisting power requisite in energetic exspiration, and on the other side to make them vibrate without delay in weak exspiration, the windpipe must be contracted for a second, thus checking the outflow of breath and causing the chords to vibrate in unison. The sonant breath so produced is the _spiritus lenis_ of our old school-grammars, the slight noise produced by the check given in the throat to the uprush of air from the lungs. The noise may easily be detected in whispering, or in the pronunciation of a word like _’ear_, when a special effort is made to prevent it from degenerating into _year_, and the fact that it is a noise will explain the dislike felt by the sensitive Greek to what the grammarians term a _hiatus_. The _spiritus lenis_ varies according as it is the result of a compression of the _chordæ vocales_ alone, or of the false _chordæ vocales_ as well; but it is doubtful whether we can treat it as a distinct consonant and not rather as the pure tone of the voice. Perhaps it should most strictly be called a glide. It readily passes into the non-sonant aspirate or _spiritus asper_, by allowing the breath to pass through the throat without check or hindrance. The glottis, indeed, is in the latter case slightly narrowed and the larynx stiffened, but the difference between the rough and soft aspirates is that the one is a continuous sound, the other a checked breath. The vocal chords are brought together while the breath is passing through the throat, and since their movement may be either quick or gradual the hard aspirate or _h_ may correspondingly vary in character. As Czermak first pointed out, the more usual hard aspirate is that produced by the gradual compression of the vocal chords when they remain for a moment in a given contracted position.[156]

The same causes which produce the _spiritus lenis_ or the _spiritus asper_ at the beginning of the vowel-sound produce similar results at its end. It may terminate with a weak breathing, a firm breathing, or a non-sonant aspirate. In the case of a weak breathing the exspiration either ceases before the vocal chords have begun to vibrate, thus resulting in a long vowel, or at the very moment at which the windpipe is opened to admit the passage of air, the result being a short vowel. The weak breathing answers to what may be called the neutral vocalic utterance, so rarely heard in language, when the vowel-sound is introduced without either the soft or hard aspirate, the windpipe being merely narrowed sufficiently to set the vocal chords in motion at the same moment that the exspiration takes place. The firm breathing corresponds with the _spiritus lenis_, and is due to a sudden check given to the vibrating voice. Examples of it occur in words like _no!_ _bah!_ uttered abruptly, or where we wish to divide two similar vowels one from the other. The non-sonant aspirate is produced by continuing the exspiration for a while after the opening of the windpipe, and may be heard in final vowels which are at once short and strongly accented. The non-sonant aspirate is sometimes combined with the firm breathing, especially in Danish, where such words as _ti_, _nei_, are pronounced with a double exspiratory effort, the second consisting of a non-sonant breath of more or less strength, jerked up, as it were, after the vowel.

Now, let us stop for a moment to remind ourselves of the distinction between sonant and non-sonant. Non-sonant or surd sounds (also called “hard” and “breathed”) are breath as modified by the organs of speech; sonants, “soft” or “voiced” sounds, are voice similarly modified, voice being breath when played upon by the vibrating _chordæ vocales_ in its passage through the partially closed glottis. Voice, therefore, continues to be heard without interruption as long as we have a succession of sonants following one upon the other; the transition or “glide” from one sonant to another consisting simply in the change of position assumed by the organs of speech. In pronouncing the sound _al_, all that happens in passing from _a_ to _l_ is a transference of the tongue from the position required for forming _a_ to the position required for forming _l_; voice continues without interruption. Now it is clear that while voice is passing from _a_ to _l_, neither pure _a_ nor pure _l_ can be sounded, though the time occupied by its passage (that is, by the change in the position of the tongue) is so infinitesimally small that the sound or sounds actually produced cannot be heard, and all we can be conscious of is a modification of _a_ at its end or of _l_ at its beginning. If we have two successive vowels, each belonging to a different syllable, a separate effort of exspiration is needed for both, and the transition-sounds are apt to escape notice from the weakening of the exspiration during the interval between the two efforts; but if the vowels do not belong to distinct syllables, the result is wholly different. Diphthongs, as we term them, consist in the combination of two simple vowels, usually short, into a single syllable pronounced, therefore, with a single exspiratory effort, and with the stronger accent on the first vowel. The sound we hear is produced while the organs of speech are being changed from the position required for the one vowel to the position required for the other. We have only to sing the diphthongs _ai_ or _au_ on a long note to hear a distinct _i_ and _u_ at the end of each, and the Sanskrit grammarians discovered more than two thousand years ago that the diphthongs _ê_ and _ô_ were really combinations of _a_ + _i_ and _a_ + _u_. The primary condition of the existence of a diphthong is the rapid transition from one of the component vowels to the other, and this renders the true resolution of a diphthongal sound so extremely difficult except to the specially trained ear. Once acquainted with the two component vowels, we can easily determine the intermediate or transition sounds in which the diphthong really consists; but written documents rarely do acquaint us accurately with them. Diphthongs whose second element is _e_ or _o_ have sometimes been termed “imperfect” and considered of younger origin than those whose second element is _i_ or _u_, because of their greater fulness of tone and consequent inappropriateness to the unaccented place in the compound; but such a view does not seem to be correct. It appears certain, however, that languages show a tendency to form diphthongs the longer they live and the greater the extent to which they have been affected by phonetic decay. English is a prominent example of this tendency; our vowels are all becoming diphthongs; even the first personal pronoun _I_ (_ai_) has become one, and already we hear _aither_ and _naither_ more frequently than _either_ (_eether_) and _neither_. The so-called long vowels which occur in such words as _say_, _no_, _he_, are all diphthongal, and some of the local dialects have carried the tendency even further than the literary language.

The existence of triphthongs has been disputed, and no doubt most of the alleged cases, such as _iei_ or _ieu_ in the Romance idioms, are either dissyllables or consist of a semi-vowel followed by a diphthong. But, as Sievers remarks:[157] “the transition from the first to the second component element of a diphthong may be so prolonged that even the transition sounds themselves may be distinctly heard.” As for semi-vowels, they differ from the first element of a diphthong only in having lost the accent and being followed by a strongly accented vowel. Hence they come to assume the function of sonant consonants. Hence, too, the necessity that the vowels in which they originate should possess less fulness of tone than the vowels by which they are immediately followed. We may have _yá_ and _wá_, but hardly _ᵃᵢ_ and _ᵃᵤ_. Naturally _i_ and _u_ most readily pass into semi-vowels, partly from their comparatively weak tone, partly from the compression of the air-passage needed to produce them, partly from the similar position of the organs of speech in forming the spirants _y_ and _w_. These spirants, as we shall see, are not to be confounded with the semi-vowels _y_ and _w_.

A vowel, then, is the quality or timbre of voice as modified by the tongue and lips, and consists of the forms assumed by the vibrating air as it passes through the windpipe and vocal chords. But the tongue and lips naturally tend towards the same position whatever be the vowel sounded. A man who has been accustomed to give his tongue a particular position in pronouncing _i_ will give it much the same position in pronouncing _e_, for we must never forget that there is an almost infinite number of _i_’s or _e_’s varying with the slight changes of position of the tongue and lips when placed for enunciating those vowels. According to the greater or less use made of the lips in speaking will be the character of all the vowel-sounds of a language. The vowels, consequently, fall into systems, and in investigating the phonology of a dialect, we have to inquire not only what vowels it possesses, but more particularly what system these fall into. The basis of English vowel pronunciation is the passive position of the lips, just as in the Holstein dialect it is the withdrawal and flattening of the tongue. Sievers states, that in speaking the dialect of Lower Hesse the tongue must be relaxed and in a position of the slightest possible tension; while, on the contrary, in the Saxon dialects the whole tongue must be tense, the throat stiffened and the exspiration energetic. “Hence the hard, somewhat screaming impression made by this dialect in contrast with the dull, almost heavy and negative character of the Hessian.”[158]

But it is time to turn from the vowels to the consonants, the skeleton, as it were, of articulate utterance. A language could consist wholly of vowels; indeed, a Polynesian dictionary contains numbers of words which have not a single consonant in them, and children frequently mark the differences between words rather by the vowels than by the consonants they contain. The earliest systems of writing other than ideographic are syllabaries and not alphabets, while alphabets like the Sanskrit ascribe an “inherent” vowel to each of their consonants. But though vowels are indispensable to an organized language, it by no means follows that they were equally indispensable to the first attempts at speech. As a matter of fact, a preponderance of vowels such as characterizes the Polynesian dialects is a sign of phonetic decay and linguistic old age. “Consonants,” says Professor Max Müller, “are much more apt to be dropped than to sprout up between two vowels.” If we had only the Greek μέρμερος or the Latin _memor_ before us, we should have no idea that they have lost an initial sibilant; in fact, this only becomes apparent when we compare the Sanskrit _smar_, “to remember.” The endeavour sometimes made to reduce the Parent-Aryan alphabet to a small number of simple and easily pronounced consonants, is founded on the fallacy that the results of a phonetic analysis of the words we utter and a reduction of the sounds they contain into their leading types, is identical with the primitive alphabet of the Aryan race. On the contrary, the sounds of a language become more simplified and clearly marked the longer it continues to be spoken, and the primitive Aryan alphabet, instead of being a simple list of primary sounds, from which all that are harsh or indistinct have been carefully eliminated, must really have resembled the existing alphabets of barbarous or semi-barbarous tribes, and included a large variety of consonants, many of which we should find it extremely difficult to reproduce.

Consonants may be divided, in the first place, into hard and soft, or, as they are more usually termed, surd and sonant. A surd consonant consists of checked breath, a sonant consonant of checked voice. If, in the second place, either breath or voice is completely checked in its passage through the organs of speech, an explosive or momentary (also called a stopped or mute) consonant is heard at the moment the check is removed; if the check is not complete, and the organs of speech only approximate so that the breath cannot escape without friction, a fricative (spirant, “unstopped”) or continuous consonant is the result. Where a spirant or fricative is immediately preceded by an explosive, a double sound or affricative is the result (_e.g._ German _pf_, Armenian _t’š_); where the spirant follows the explosive we have the aspirated letters, which will be spoken of hereafter. Among the continuous consonants must be ranked the nasals, produced by dropping the uvula and so allowing some of the breath to make its way to the nostrils through the pharynx, and the trills produced by the vibration of the uvula, the lips, or more commonly the tongue. Distinct from the nasals and the trills are the central continuous consonants (_h_, _ch_, _y_, English _r_, _w_, _wh_, and the sibilants) formed by lifting the centre and point of the tongue to the centre and front of the palate, and the lateral continuous consonants (_l_, and, according to Bell, English _th_, _f_, _v_), in forming which the breath is allowed to escape along the edges of the tongue. A further cross division will be into liquids, gutturals, dentals, palatals, labio-dentals, and labials, to which may be added the linguals or cacuminals (cerebrals) of Sanskrit.

_The Liquids._—Among the liquids should properly be reckoned only those kinds of _r_ and _l_ which stand to the spirant _r_ and _l_ in the same relation that the vowel _i_ stands to the spirant _y_. In forming the vowels, as we have seen, the tongue assumes a dorsal position, that is, some part of its back is raised towards the palate; in forming the liquids, on the other hand, the tongue has either an oral (central) or a lateral position, the liquid _r_ requiring the articulation of the centre and tip, the liquid _l_ that of the sides. But there are several kinds of _r_, which may be classed as cacuminal, spirant, alveolar or dental, uvular or guttural, and laryngeal. The cacuminal _r_ is the purest liquid _r_ that we hear, inasmuch as it is wholly untrilled, and is especially common in cultivated English. In order to produce it, the front surface of the tongue is hollowed out into a spoon-like shape and raised towards the hard palate behind the alveolar teeth-roots of the upper jaw, while the edge of the tongue is stiffened and kept free from any sort of vibration. It will be clear from this how closely allied this cacuminal _r_ is to the vowels, and we can easily understand the readiness with which it combines with a vowel-sound when we remember that it may be formed in almost any part of the hard palate, while the lips have free play during its creation. Corresponding to the cacuminal _r_ is the spirant (or “buzzed”) _r_, which also occurs plentifully in English as in such words as _try_ or _dry_. The mouth is completely closed by the tongue when sounding _t_ or _d_, and if in passing to the position needed for _r_ the tongue is not removed from the palate quickly enough, or the exspiration is not sufficiently strong, a slight fricative sound like that of _sh_ is produced which results in the spirant _r_. As for the dental or alveolar _r_, all that is requisite to produce it is to raise the front part of the tongue, at the same time slightly arching its extreme edges, and so obtaining a constricted or “squeezed” chamber of resonance between the side of the tongue and the alveolars. This _r_ may be untrilled, but in German it is more frequently a trilled one. The trill is caused by the force of the exspiration which strikes the thin hollowed edge of the tongue in an outward direction, the tongue the moment after returning to its former position like a piece of india-rubber. If the two edges of the front part of the tongue be pressed against the teeth, the tip of the tongue between them being alone allowed free play, and accordingly vibrating in a very small and narrow space, a sound is heard approaching that of _s_ or _sh_. The stronger the uprush of breath and the vibration it occasions, the plainer will be the sibilated sound; indeed, a genuine sibilant can even attach itself to the liquid, as in the Polish _rz_. The uvular or guttural _r_ is supposed by Sievers to be a modern substitution for the trilled alveolar _r_. At any rate it is produced by lifting the back of the tongue to the soft palate and forming a deep groove along the middle of it, in which the uvula can vibrate freely. The groove, however, is frequently left wholly or nearly unformed, the consequence being a very grating character acquired by the _r_, which then passes over into the sonant guttural spirant heard in sounding the modern Greek γ. The laryngeal _r_ was first observed and described by Brücke, who makes it arise from sinking the voice so that the vocal chords cease to vibrate audibly, and merely produce intermittent and explosive sounds.

Each kind of _l_ is formed in the same way, by raising the tip of the tongue and so closing the orifice of the mouth, at the same time allowing the breath to pass along the two sides of the tongue in successive oscillations produced by the vibrations of the elastic edges of the tongue. We may distinguish the cacuminal _l_ in which the tip of the tongue is bent backwards as in the cacuminal _r_; the alveolar _l_ with the edge of the tongue laid against the alveolars; the dental or interdental _l_ in which the flattened surface of the tongue fills up the space between the two sides of the mouth; and the dorsal _l_ (as in the Spanish _llano_) in which the tip of the tongue presses against the lower incisors, while the centre of the tongue is raised towards the alveolars of the upper teeth. The best-known variety of the cacuminal _l_ is that of the Welsh _ll_ formed by pressing the flattened tip of the tongue against the gums of the upper teeth and allowing the breath to escape on its right side. The same sound is heard in the Icelandic _hl_ and _l_ before a _t_, and also in Cheroki,[159] though in Icelandic the tongue is pressed against both sides of the mouth. A half-sonant, spirant _l_ may be heard when the exspiration is strong; a surd _l_ often occurs at the end of a word or after surd consonants (particularly _t_ and _s_). The sound of the _l_ may be made clearer or obscurer by raising or depressing the front part of the tongue, and so narrowing or enlarging the space between its edges and the teeth, and since the vowels may be pronounced with the tip of the tongue on the palate, they may readily pass into _l_ by simply broadening the surface of the tongue.

We have already seen that the tongue is not the only organ of speech which may be “trilled.” In the Arabic _grhain_ (غ), the Northumberland _burr_ and the French Provençal _r_, _grasseyé_, the uvula which lies along the back of the tongue towards the teeth is very distinctly made to vibrate. “If,” Mr. A. J. Ellis says, “the tongue is more raised and the vibration indistinct or very slight, the result is the English _r_ in _more_, _poor_, while a still greater elevation of the tongue produces the _r_ heard after palatal vowels, as _hear_, _mere_, _fire_. These trills are so vocal that they form distinct syllables, as _surf_, _serf_, _fur_, _fir_, _virtue_, _honour_, and are with difficulty separable from the vowels.” The lips, too, may be trilled, the result being _brh_, a sound constantly heard from children.

_The Nasals._—The characteristic of a nasal is, as the name declares, the participation of the nose in producing the sound. The breath passes through the nose rather than through the mouth. Sometimes, however, all that happens is the removal of the membrane which separates the nasal orifice from the pharynx; this alone is indispensable to the formation of a nasal letter. Hence its resemblance to a vowel, the buccal tube being alike silent in both cases. If we try to converse when walking uphill we shall find that the nasals are longest heard. These nasals must be classified as labial, dental, palatal, and guttural, according to the part of the speaking apparatus in which the current of air is checked in its exit, and it will be best to treat them along with the other sounds formed in the same part. It should be noted, however, that the so-called surd nasal which we hear in _hm!_ has really, as Sievers remarks, not the slightest similarity to a nasal, but approximates to the aspirates or breathings.

The traditional division of the consonants into labial, dental, palatal, cerebral (cacuminal) and guttural, though not scientifically precise, is yet too familiar to be disregarded, and we shall therefore follow it so far as is possible. We must, however, remember at starting the primary distinction between the two classes of letters, called variously hard and soft, tenues and mediæ, surds and sonants, as well as between those called momentary (explosive) and continuous or checks and fricatives. What this distinction consists in has already been explained.

_The Labials._—The labials may be subdivided into pure labials, with the formation of which the lips only have to do, and the labio-dentals, in the formation of which the teeth also participate. In pronouncing the surd _p_, the sonant _b_, the nasalized _m_, or the middle German _w_, the lips are either wholly or (as in _wh_) almost wholly closed. _B_ only differs from _p_ in being pronounced with voice instead of breath, the voice partly preceding, partly following the check occasioned by the closure of the lips. As in all sonant letters, the exspiration is less forcible than in the case of surd letters. The labio-dentals _f_ and _v_ are merely modifications of the rough and soft aspirates by pressing the lower lip against the upper teeth. When the lips are brought together without any interference of the teeth the _spiritus lenis_ becomes the German _w_ as heard in a word like _Quelle_. Our _wh_, or rather _hw_, and _w_ are continuous sounds, the lips being slightly opened, the back of the tongue raised, and the breath passing over its central part.

_The Dentals._—The articulation needed for the dentals is partly oral, partly alveolar, partly dorsal. The common principle, however, involved in the formation of them all is the same; the tongue must be brought against the teeth. The so-called cerebral or cacuminal dentals of Sanskrit and the Dravidian tongues (_ṭ_, _ḍ_, _ṭh_, _ḍh_) are due to oral articulation, the tongue being made convex and the lower surface raised towards the palate. The English _t_ and _d_ are also said to be cerebral, though the tip of the tongue is not bent very sharply backwards in forming them. Alveolar articulation is needed for the dentals when they have to be pronounced with the edge of the flattened tongue pressed against the alveolars of the upper teeth, while in dorsal articulation the point of the tongue is simply turned back against the lower teeth, its convex being at the same time lifted to the palate. It is in this way that the Bohemian dorsal _t_ is formed. The dorsal dentals may be varied by raising the back of the tongue nearer to the mouth or the throat, the tip either resting behind the lower teeth or being raised to the upper alveolars. Besides the surd dental _t_ and sonant dental _d_, we have also a series of dental spirants which bear the same relation to _t_ and _d_ that _f_ and _v_ bear to _p_ and _b_. By slightly opening the teeth and stopping the aperture with the extended edges of the tongue we produce the interdental sounds heard in _breath_ or _think_ and _breathe_ or _then_. The first _th_ (or _thorn_ þ) differs from the second (ð)[160] in being pronounced with the rough breathing instead of the soft breathing. They stand midway between an oral and a dorsal articulation. How readily they may pass into the labio-dentals _f_ and _v_ is clear at a glance; we have only to raise the lower lip a little and curl back the tongue, and our _th_ becomes an _f_. Equally readily, as we shall see, is the passage from them to a sibilant. We seldom meet with an interdental consonant; Sievers, however, states that they exist in Servian and Armenian, where they regularly represent the whole class of dentals.

_The Palatals._—The palatals come next. They stand between the dentals and gutturals, and are formed by throwing the middle of the tongue, raised as it were into a hump, against that part of the roof of the mouth where the hard palate begins. The sound (_ch_) heard in the English _church_ or the Italian _cielo_ is now held to be, not a palatal, but a dental (_t_ followed by _sh_), and we must go to the Sanskrit (_ch_) as still pronounced to find a type of the whole palatal series. It “is formed most easily,” says Professor Max Müller, “if we place the tongue and teeth in the position for the formation of _sh_ in _sharp_, and then stop the breath by complete contact between the tongue and the back of the teeth.” It will be seen from this that the true _ch_ is not a double letter, a compound of _t_ and _sh_ or _s_, but a single consonant which ought to be denoted by a single character. The Sanskrit palatal _ch_ may have had the same pronunciation as the Armenian _t‘ sh_,[161] as Sievers thinks, or it may have been equivalent to _ky_. However this may be, it is plain from the great extent of the “chamber of resonance” in which the palatals are formed—the whole of the hard palate being available for the purpose—that a large number of palatal sounds is possible. They may range, in fact, from _ky_ to _tsh_. The guttural _k_ passes easily enough into the palatalized _ky_, as may be seen from the pronunciation of _kind_ and _cow_ as _kyind_ and _kyow_, not unfrequently heard in English; indeed, all that is requisite for the transition is for the front part of the tongue to assume the position needed for _y_, while the back part is in that needed for _k_. In the northern dialects of Jutland _j_ is heard after _k_ and _g_ when followed by _œ_, _e_, _o_, and _ö_. The German “soft” guttural aspirate or palatal spirant in words like _ich_, _licht_, is the result of the _spiritus asper_ passing the middle of the tongue when raised against the hard palate, _y_ in _you_ or _yet_ being due to a softening of the breath, the organs of speech remaining unchanged. The palatal sibilants will have to be considered separately.

_The Gutturals._—Putting aside the cerebrals, which have been treated under the head of the dentals, we now come to the gutturals, usually an important class of sounds in savage idioms. First of all we have the tenuis _k_, produced by bringing the root of the tongue against the soft palate, together with the deeper _k_ heard in the Semitic _koph_ or Georgian _q_. Next is the media _g_, to create which breath has to be changed into voice. Then will come the guttural nasal _ng_ (as in _sing_), and the continuous _ch_ and _g_ heard in the German _nach_ and _Tage_. The sound heard in _nach_ or the Scotch _loch_ is formed by raising the tongue against the soft palate or uvula, and so checking the uprush of breath, its sonant representative being the _g_ of _Tage_. The result of only slightly checking the uprush of breath in the latter case is the passage of the guttural into a semi-vowel. This sonant _g_ is the γ of modern Greek; it sometimes takes the place of the uvular _r_, though this office more properly belongs to the sonant _g_ of Armenian pronounced further back in the mouth. The surd _ch_ may be similarly modified by a posterior pronunciation, and so become the Armenian _xe_, the Russian _x_, the Polish _ch_, and the deep _ch_ of the Swiss.

_The Sibilants._—The main division of sibilated sounds is into the surd _s_ and _sh_, and the sonant _z_ and _j_. When the centre and tip of the tongue are raised to the centre and front of the palate, the breath or _spiritus asper_ is modified into _s_ (as in _sin_), the voice or _spiritus lenis_ into _z_ (as in _zeal_ or _rise_). When the tongue is turned back with its lower surface against the alveolars of the upper teeth, less of the palate being covered than is required for _s_ and _z_, breath becomes _sh_ (as in _sharp_), voice _j_ (as in _azure_, _pleasure_, French _jamais_). The ordinary German _s_ is a dorsal one, the current of air being allowed to pass between the upper alveolars and the lower surface of the uplifted tongue; in North German dialects, however, we frequently meet with an alveolar _s_, formed in much the same way as the alveolar _r_. The same _s_ also occurs in English, as well as a cacuminal _s_ distinguished by a more pronounced retraction of the tip of the tongue and narrower space between it and the palate. The palatal _ś_, found in Russian, for instance, before the weak vowels (_e_, _i_, &c.), only differs from the dorsal _s_ in the more retracted position of the tongue. _Sh_ (_j_) can be modified in three ways. The channel formed in the tongue when pronouncing _s_ may be so diminished as to allow the breath to strike against the lips, or the lips may form with it an approximately rectangular aperture, or, thirdly, the left (or more rarely the right) side of the tongue may be pressed against the palate, causing the breath to strike against the lips, which are generally raised a little on the side. Sievers declares that he has sometimes heard this unilateral _sh_ in England. However this may be, all three modifications of _sh_ may combine with the dorsal, alveolar, cacuminal, and palatal positions of the tongue to produce the cacuminal _sh_ of English (identical, probably, with the Sanskrit _ś_), the palatal _mouillé_ _ś_ and _ć_ of Polish and Russian, the alveolar _sh_ of the North German dialects, and the dorsal _sh_ of the Middle and Southern German dialects. It is one of the many evils of our defective and misleading mode of spelling that the surd _sh_, though a single sound, is represented by two letters, and so cannot be distinguished from the aspirated _sh_ (as in _gas-hole_), which is really a double sound.

These aspirated sounds consist, as we have seen, of an explosive followed by a spirant, and they occupied an important place in the older languages of our Aryan family of speech. A large number of roots contain them, and the Brahmans still pronounce each part of the compound sound distinctly, _ph_ and _th_, for instance, being pronounced as in our _up-hill_ and _ant-hill_. The compound nature of the sound caused sometimes the one element in it, sometimes the other, to fall away. Thus, to a Sanskrit _tubhy(am)_ corresponds a Latin _tibi_, and the Latin _mihi_ and Sanskrit _mahyam_ presuppose an earlier _mabhyam_, _mabhi_. The Athenian tendency to false aspiration which has produced the initial aspirate of ὑδώρ (Latin _unda_, _udus_) or ἵππος (Latin _equus_) has also occasionally affected the labial tenuis. φῦσα and its kindred, for instance, answer to the Latin _pustula_, the Lithuanian _pústi_, “to blow;” ἄφνος is the Sanskrit _apnas_, the Latin _ops_, and κεφαλὴ is the Sanskrit _kapâla_, the Latin _caput_. A curious metathesis of the aspiration may take place in both Sanskrit and Greek. In Sanskrit a final aspirated media before a following tenuis loses its aspirate, which is transferred to the initial of the root, provided that be _g_, _d_, or _b_ (as _bhut-karoti_, “he who knows acts,” for _budh-karoti_); and in Greek we find θρίξ becoming τριχὸς, τρέχω becoming θρέξω.

But it must be remembered that it is only the surd explosives (or tenues) that properly can thus be combined with the rough breathing (_h_). A difficulty occurs in the case of the sonant explosives (or mediæ); and it is a grave question whether we ought to transcribe _gha_, _dha_, and _bha_ by the side of _kha_, _tha_, and _pha_. In Greek, at any rate, we have only aspirated tenues, and while τ’ followed by an aspirate is written θ, this is never the case with δ’. At the same time, the existence of aspirated mediæ was recognized by the Prâtiśâkhyas by the side of the aspirated tenues, and the accuracy of the Prâtiśâkhyas is confirmed by the requirements of etymology.

Closely connected with the sibilants are the palatal and guttural sounds, already noticed, heard in the German _ich_, _tage_, and _acht_. The palatal _ch_, written χ by Sievers, _jh_ by Sweet; is of two kinds. What Sievers calls χ,[162] heard in the German _ich_, Icelandic _hjarta_, and sometimes in such words as our _hue_, is formed on the hard palate near the soft palate by the front part of the tongue. On the other hand, χ,[163] as in the Dutch _g_ before _e_ and _i_, is formed in the hollow of the arch. The guttural sonant heard in the North German _tage_, or the modern Greek γ, is formed between the back of the tongue and the middle of the soft palate, the tongue being lifted up towards the front of the mouth. As already remarked, it sometimes represents the uvular _r_; thus, Mr. Sweet says, “when the passage (of the voice) is widened so as to remove all buzzing, the sound of (_gh_)[162] no longer suggests (_kh_)[163] or (_g_), but rather a weak (_r_) sound.” Further back in the mouth is formed the Armenian sonant _g_, corresponding to χ.[163] The _ch_ of _acht_, again, may be divided into two varieties. _Ch_,[162] formed, as stated above, between the back of the tongue and the middle of the soft palate, is the guttural spirant usual in German after _a_, _o_, and _u_, and heard in Scotch _loch_. Further back is formed _ch_,[163] common in Swiss and other South German dialects. We have also _ch_,[163] noted by Mr. Sweet in Scotch after _e_ and _i_, formed between the back of the tongue and the place where the hard palate begins. It thus comes very near χ.[164]

Distinct from the exspiratory sounds, whether vowels or consonants, which have now been passed in review, are sounds formed either by inspiration or simply by the air in the mouth itself. Winteler[165] describes certain Swiss dialects which make use of inspiratory sounds to disguise the voice, and the clicks characteristic of the South African languages are examples of sounds produced without either taking in or emitting breath. The Kafirs have borrowed the three easiest clicks (the dental, the cerebral, and the lateral) from their Hottentot neighbours,[166] and there are reasons for thinking that the Hottentots themselves borrowed in turn from the more primitive Bushmen. At all events, the labial and compound dental clicks are wanting in Hottentot, and the Bushman fables put what Dr. Bleck calls “a most unpronounceable click,” which does not occur otherwise in any of the dialects, into the mouth of the hare, the anteater, and the moon.[167] These inarticulate clicks, thus adapted to the purposes of articulate speech, bridge over the gulf between the latter and the cries of animals, and we may see in them a survival of those primæval utterances out of which language was born. Traces of what may thus be termed the germs of language on its phonetic side are met with here and there all over the globe. Thus Haldeman describes at least three clicks heard in Texan, Chinook, and other North American languages, _t_ in the Anadahhas of Texas, for instance, being followed by “an effect as loud as spitting.”[168] According to Klaproth, clicks occur in Circassian; and Bleek states that two clicks are distinguished in the ǀikhe language of Guatemala—one somewhat resembling the Hottentot dental click, and the other the Hottentot palatal combined with some guttural. Mr. Whitmee has heard a click in certain dialects spoken by the Negritos of Melanesia. Clicks are also known among the Gallas; and Miss Lloyd has found a little boy from Lake Ngami using clicks resembling those of Nama Hottentot. Clicks are formed by placing the tongue or lips in the position required by an explosive, and then sucking out the air between the organs thus brought into play, the result being the “cluck” or “smack” with which grooms are accustomed to encourage a horse, but in combination with the explosive for which the organs of speech were set. According to Mr. Sweet, the labial click is an ordinary kiss; the dental click, “the interjection of impatience ordinarily written ‘tut.’”[169] In Káfir the clicks are not pure, as in Bushman—that is to say, they are always accompanied by an exspiratory consonant, which is formed at the same moment as the click. This affords an additional reason for thinking that the Káfir clicks are not survivals from the original condition of speech, but loans from another people, which have been attached by way of ornament to the existing exspiratory sounds of the language. Of the same nature as the clicks are the implosives peculiar to Saxon German, where no distinction is made between _d_ and _t_, or _b_ and _p_. Similar sounds are heard in Georgian and the Armenian of Tiflis, and they must have characterized ancient Accadian, since no distinction is made in writing between final _d_ and _t_, _g_ and _k_, or _b_ and _p_. These implosives are due to compression of the air between the closed glottis and the organs of speech when in position for an explosive, by forcing the glottis upwards. No sound is emitted until the sound is fully formed, when the final or transition sound is curiously modified.

We have hitherto dealt with the individual sounds in the same fashion as the lexicographer deals with individual words. But just as a word is really but one of the elements of a sentence, and to be thoroughly understood must be treated as such, individual sounds are but the elements of which syllables are composed. Whatever may be the nature of a sound when regarded apart and by itself, it is necessarily much modified when combined in actual speech with other sounds. The syllable, and not the single sound, is the starting-point of phonetic utterance.

A syllable must contain either a vowel or a semi-vowel, by which are meant such inspiratory utterances as that heard in the interjection _’m_, or the vocalic _r_ and _l_ of Slavonic and other tongues. One of the first achievements of the phonograph has been to show that an open syllable like _ga_ can be pronounced either backwards or forwards indifferently when once the organs of speech are in position; and not only so, but that when the waves of air set in motion by the pronunciation of a word are reversed, the word will be reproduced backwards—_əsoshiéshun_ (_association_), for instance, becoming _nushéshiosə_.

Mr. Sweet has pointed out that syllables are divided by the stress. Speech has to be carried on by a succession of exspirations or puffs of breath, and naturally the force with which the breath is emitted gradually diminishes during the continuance of the exspiration. Only in special cases—the interjections, for example—the force increases instead of diminishing. When the exspiration is spent, and a new breath is taken, a new syllable begins. Wherever, therefore, the stress is laid we must place the beginning of a new syllable. In “a name” the stress is on the nasal, where accordingly the syllable begins; in “an aim” it is, on the contrary, on the diphthong.

The passage from one sound to another, as has already been noticed, consists of a series of infinitesimal intermediate sounds, corresponding with the series of positions assumed by the vocal organs in passing from one position to another. These intermediate sounds have been conveniently termed “glides” by Mr. Ellis, and they play an important part in the formation of syllables. Glides are of two kinds, as the organs of speech may either be moved from one position to another in the shortest possible time, or be shifted, on the way, towards another position needed for the production of a third sound. Thus, in the syllable _ki_ we have the immediate glide required for the transition from _k_ to _i_; in the syllable _qui_, the indirect glide from _k_ to _i_ through the position needed for _u_. A glide may, of course, be described as either initial or final; in _ki_, the glide of _k_ is being final, that of _i_ initial. Some of the so-called consonants and vowels are really glides. The neutral vowel (_ə_) is termed the “voice-glide” by Mr. Sweet, as “produced by emitting voice during the passage to or from a consonant.” It may begin a word, as in “against,” and in English is very frequently replaced by a liquid, as in the words “little,” “possible.” It is also found plentifully in the Semitic languages, the Hebrew _sh’wa_, for instance, being simply the neutral vowel or voice-glide. In words like “follow,” when pronounced rapidly, we may hear it labialized. A diphthong, again, is a combination of a full vowel with a glide-vowel either before or after it, though the glide-vowels may be prolonged into full vowels without destroying the diphthong, by equalizing the stress upon the two elements of which it is made up. These glide-vowels (like the consonantal glides) are produced by putting the vocal organs into position for pronouncing a particular vowel, but not letting voice sound until this position is being shifted to that required by the full vowel which forms the second part of the compound, and reversing the process when the full vowel forms the first part. Consonantal glides (_y_, _w_, _r_, _l_, _m_, _n_) are illustrated by the sound of _y_ in _you_, and of _r_ in _here_, and in a common South-country pronunciation of words like _red_.[170] According to Mr. Sweet, the aspirate _h_ is a consonant in the glottis, but “a voiceless glide-vowel in the mouth.”[171]

At all events it is often difficult to distinguish the rough breathing from the glide which easily develops into it by the help of a little additional stress. This glide may be detected after mediæ, tenues, and _s_, whether initial or final, as in our _cold_ (when pronounced emphatically), _pack_, and _big_. The Irish and Danish aspirated consonants are formed by laying a separate stress on the glide apart from the stress laid upon the preceding consonant. The aspirated letters of Greek and Sanskrit, described above, are of course different, as here we have a combination of two independent sounds, though the latter of these (_h_) is in Mr. Sweet’s eyes a mere glide-vowel in the mouth.

Glides may be absent where two consonants formed in the same part of the vocal organs are united together (e.g. _and_, _its_), or even where they are formed in different parts. This is especially the case with English. Wherever homorganic sounds are produced, the vocal organs pass at once from the position required for the first to that required for the second, without first falling back into the “position of indifference.” Where an explosive is followed by a nasal, a sudden opening of the _velum pendulum_ is substituted for the usual “explosion,” as first pointed out by Kudelka.