Chapter IV: Phonology and Sematology (3)

Syllables may differ one from the other in respect of _pitch_ or _tone_, of _stress_, and of _quantity_. Pitch or tone is but little noticed by Englishmen, since with us it serves merely a logical or emotional purpose, such as the expression of surprise or the asking of a question, but in some languages, Chinese or Swedish or Lithuanian, for example, every word has its own separate tone, which helps to distinguish it from other words. This, too, was the case in Vedic Sanskrit, and in ancient Greek and Latin, what we call the Greek accents being really the marks of the pitch at which words were pronounced. Pitch or tone depends on the rapidity of the vibrations of sound, and may be either rising, level, or falling. The rising tone is that indicated by the acute accent. Tone may also be compound, marked in Greek by the circumflex. The compound or circumflex is heard when the tone of a vowel is again raised after it has already passed the moment of its greatest intensity, and it may therefore be described as composed of the acute and the grave, or of the rising and the falling. It may be noticed in Lithuanian as well as in several German dialects, such as the Thuringian, which have a singing character, and when it falls upon a diphthong the second element of the diphthong is distinctly raised in pitch. Naturally it is usually found with diphthongs and long vowels, but short vowels combined with a liquid may also carry the circumflex. In Greek it commonly implies a contraction, the circumflex resulting from the coalescence of a vowel which has the acute accent with one which has the grave.[172]

The Vedic system of accentuation best exhibits the fundamental character of accent of pitch. The _udâtta_ or acute denotes the highest pitch reached by the voice in a group of syllables or words. In the syllable immediately preceding the voice naturally sinks to its lowest, thus producing the _anudâtta_, or grave tone. After the _udâtta_, however, the voice falls gradually; consequently the syllable which follows has the _swarita_ or circumflex accent, and it is only the next syllable to that which is again _anudâtta_.

But the tone is regulated by three different conditions, which sometimes act antagonistically. It may be either a syllable-tone, determined by the relative force with which the syllables of a word can be uttered, dependent on the nature of the sounds of which they are composed; or a word-tone, determined in great measure by the meaning, and serving to distinguish words from one another; or a sentence-tone, mostly determined by logic or the feelings. The Greek accents, like the Vedic ones, were used to denote all three varieties of tone; while the acute and the circumflex sometimes represent the syllable-accent (as in θῖνα, ἔτυπον), sometimes the word-accent (as in νυμφή, νύμφα, ποδῶν), the grave, as Sievers remarks, “is a concession to the requirements of the sentence-tone.” Similarly in Vedic Sanskrit, the _udâtta_ which ordinarily indicates the word-accent, falling as it does upon the syllable (commonly the flection) to which the signification caused the attention to be chiefly directed, seems also to have indicated the sentence-tone, since the verb of the principal clause has no accent whatever attached to it. Previously, however, both in Greek and Sanskrit the accents denoted the word-tone, and the remarkable agreement between the accentuation of the two languages enables us to restore in great measure the accentuation of the undivided parent-speech. It cannot be an accident, for instance, which makes the numeral seven (_saptán_, ἑπτά) oxyton in both languages, and the numeral five (_pánchan_, πέντε) paroxyton, or places the acute accent on the last syllable of adjectives in -_us_; the accentuation in each instance must have been that of the Parent-Aryan. Where the accentuation of the two languages differs, it can generally be explained by the disturbing influence of analogy. Thus while there is so remarkable an agreement between the accentuation of Vedic and Greek nouns, there is next to none between that of the verbs. But an explanation of this is forthcoming. The verb of the principal clause in the Veda loses its accent, as has just been remarked, unless it stand at the beginning of the sentence; in fact, it is regarded as an enclitic, and throws its tone back upon the preceding word however many syllables it may contain. Now in Greek a rule gradually grew up forbidding the accent to be placed further back than the antepenultimate; the accent, accordingly, which in the case of verbal forms of more than two syllables would have been on the last syllable of the preceding word in the Veda fell on the penultima of the corresponding verbal form itself in Greek. The accentuation which thus fixed itself in the verb of the principal clause was extended by analogy to the verb of the subordinate clause, and eventually to verbal forms of less than three syllables; φημι, εἰμι, and ἐστι, however, remained unaccented to bear witness to the process whereby the Greek language had changed the original accentuation of the Aryan verb.[173] This, like the accentuation of the noun, was mostly (and probably at the outset altogether) on the flection-suffix to which it called attention, and thus marked out the symbols that expressed the grammatical relations of the sentence. In the Semitic languages, on the contrary, the primitive accentuation was on the penultima, though there may possibly have been an earlier time when it was upon the ultima.[174] The tendency to throw back the accent set in early in Aryan speech; in Latin, as in the Æolic dialect of Greece, it was uniformly as near the beginning of a word as possible, and the preservation of the original pitch-accent in Lithuanian is one of the most curious marks of archaism in that most conservative of West-Aryan tongues.

In Aryan the word-tone, we have seen, was primarily used in the service of grammar. In Chinese, Siamese, and other Taic languages, however, its use is lexical rather than grammatical; here it serves to distinguish the senses of words which would otherwise be pronounced in the same way. Dr. Edkins has shown that modern Mandarin Chinese is an exceedingly decayed speech; its initial consonants have been worn away; and all its final consonants reduced to the same monotonous nasal. To prevent the confusion that would thus have been occasioned in a monosyllabic language, where the possible number of different syllables denoting words was limited even before the corroding action of phonetic decay, tones were adapted to the expression of meaning, and as old letters disappeared new tones came into existence. To create a new tone, says Dr. Edkins, requires about 1,200 years.

The sentence-tone is inseparable from speech even of the most lifeless character. Each sentence has its own key, and the several parts of it their own pitch. The tone rises when we ask a question, it falls when we answer it, it reaches the “level” point of neutrality when we speak in monotone. But there are dialects and languages in which monotone is either acute or grave. “Thus in Scotch the rising tone is often employed monotonously, not only in questions but also in answers and statements of facts. In Glasgow Scotch the falling tone predominates.”[175] In French, too, the rising tone is often used in making statements of fact.

Quite distinct from accent of pitch is accent of stress, though the close connection between the two may be gathered from the fact that in modern Greek the stress accent regularly answers to the acute and circumflex of the ancient language. Much of this regularity, however, may be due to the same pedantic revival which has resuscitated the dialect of Plato and Thucydides and substituted it for the “modern Greek” spoken half a century ago. Stress is the force with which the different syllables of words are uttered, and increased force is naturally accompanied by increased pitch. Stress, in fact, corresponds to syllable-tone and word-tone, emphasis—the stress of a sentence—corresponding to sentence-tone. Like pitch, it may be regarded as either rising, level, or falling. Stress, however, differs from pitch in its variability; there is no gradual fall, but a tendency “to sway to and fro,” as Mr. Sweet expresses it. Rising stress may consequently be of varying degrees of force and falling stress of weakness, level stress, even in French, being practically unknown. Stress and pitch together give to speech its rhythmic character, and make it the lyric utterance in which man expresses his thoughts and his emotions. Where the rhythm is regular we have poetry and song, where it is irregular the language of ordinary prose. Stress is the great conservator of language; the chief counterpoise to the action of phonetic decay. The accented syllable will be preserved though all the other syllables by which it is surrounded may disappear in pronunciation, just as the idea upon which emphasis is laid will hold out successfully against the attacks of age and forgetfulness. Winteler[176] has laid down the law that in accented syllables, liquids, nasals, and spirants are always long after a short vowel if followed by a consonant (e.g. _man̄ly_, Germ. _al̄t._)

The loss of the accent of pitch in modern English and the consequent extension of the accent of stress have made us less observant of _quantity_ than the grammarians of India or the poets of ancient Greece. All syllables, however, may be classed as long, half-long, or short, due to the duration of the force with which they are uttered. According to Brücke, the duration needed for the production of a long vowel is to that needed for the production of short vowels in the proportion of five to three, but Sievers remarks that this only applies to the oratorical pronunciation of modern literary German. In any case, the length of the same vowel may vary according to circumstances; it is long, for instance, in the English _sīz_ (_seize_), short in _sĭs_ (_cease_). Several of the Scotch dialects possess no long vowels at all, while in French most vowels are half-long, distinctly short accented vowels being final, as in _oui_.[177] Like vowels, consonants, too, may be long or short. In our own language final consonants are long after short vowels (as _hill_), short after long vowels (as _heel_), and _l_ and the nasals are lengthened before sonants (as _build_), shortened before surds (as _built_). Short final consonants after short vowels make the pronunciation appear clipped, as in German words like _mann_.

Accent has considerable influence upon quantity. On the one side short vowels may be lengthened and pure vowels converted into diphthongs by the accent falling upon them. This is partly the origin of the Sanskrit _guṇa_ and _vṛiddhi_, according to which a simple _ă_ is raised to _â_, an _ĭ_ to _ê_ (_ai_) and _ai_ (_âi_), and an _u_ to _o_ (_au_) and _au_ (_âu_).[178] The lengthening of short vowels in Hebrew in a “pause,” that is at the end of a sentence, is another example. In the German dialects monosyllables which end in a consonant frequently have their vowel changed into a diphthong by the accent, the original vowel appearing again as soon as an additional syllable is added. In our own English the short vowel of a monosyllable which ends in a sonant frequently becomes half-long when accented (compare _fog_ with _fóggy_, _god_ with _góddess_). On the other side, the absence of the accent may bring with it a diminution of quantity. Thus a diphthong may be shortened by being pronounced in the same period of time as is required for the pronunciation of a short vowel, or may even be reduced to the short vowel which lies midway between the two elements of which the diphthong consists. A short vowel, again, may be reduced to a vocalic consonant like the Slavonic _r_. Since much movement of the lips in speaking implies an energetic enunciation, shortened syllables are naturally pronounced with passive lips. To this fact we must ascribe the numerous short syllables of modern cultivated English.

There is but little difference between a long or “strong” consonant and a doubled one. In the first case, the position of the vocal organs for pronouncing the consonant is retained with gradually decreasing force, until it is suddenly shifted to the position needed for the following vowel; in the second case it is shifted back again, when the force required to produce it is half spent. Strictly speaking, therefore, the consonant cannot be said to be doubled; there is simply a break or pause in the utterance of it, the force necessary to produce it being renewed before it has been fully exhausted. In English, French, German, or Slavonic the double consonants have become long ones; to find them still pronounced we must turn to Italian, Swedish, Finnic, or Magyàr. Analogous to a double consonant is the combination of a sonant with a surd, when assimilation does not take place, as in _has to do_ or _has seen_. In Sanskrit and Greek aspirated letters could not be doubled, Sanskrit permitting only _kkh_, _tth_, and _pph_, and Greek only κχ, τθ, and πφ; hence it seems plain that there was either no glide or a glide practically inaudible.

It is obvious that the combination of a consonant and a vowel admits of an almost infinite series of variations according as the formation of the one or other sound is made prominent in pronunciation. The consonant may, as it were, swallow up the vowel; on the other hand, the vocal organs may be shifted to form the vowel while they are still in the act of forming the consonant. Hence arise _mouillé_ and labialized letters. If the front part of the tongue be raised and the lips opened while a consonant is being uttered, a palatalized or _mouillé_ letter is the result, of which the Italian _gl_ and _gn_, the Spanish _ll_ and _ñ_, or the Portuguese _lh_ and _nh_, may be regarded as examples. Still better examples, according to Sievers, are combinations of consonants with an original _i_ in many Slavonic languages (_e.g._ Russian _nikto_). Certain consonants are incapable of being _mouillé_; gutturals, for instance, in whose formation the back part of the tongue plays so prominent a part can only be so by becoming palatals. Labialized sounds are those in which the lips are rounded while the pronunciation of a consonant is in process. Labials and gutturals show the same fondness for this labialization or “rounding,” that the palatals and dentals do for mouillation; and a comparison of the derived languages proves that the primitive Aryan speech must have possessed a row of labialized or “velar” gutturals—_kw_, _gw_, _ghw_—of which the Latin _qu_ and our own _cw_, _qu_ are descendants. There is nothing to show that these velar gutturals were ever developed out of the simple gutturals; so far back as we can go in the history of Indo-European speech the two classes of guttural exist side by side, and the groups of words containing them remain unallied and unmixed. Γυνή and _queen_ (_quean_) must be separated from γένος, _genitrix_, _kinder_, and other derivations of the root which we have in the Sanskrit _janâmi_, the Greek γίγνομαι, γείνομαι, and the Latin _gigno_; and the labialized _quies_ can have nothing to do with the Greek κεῖμαι and κώμη (κύμη), our own _home_ and _ham-let_.[179] Both rounding and mouillation may be combined, as in the Danish _kyst_, _pynte_, and when occurring at the end of a word may frequently be explained from the analogy of cases in which the word is followed by a syllable beginning with _u_ and _i_. Such an explanation, however, is more likely to be true of mouillation than of rounding; indeed, an _i_ or _y_ sound is very apt to develop itself after consonants in affected pronunciation, as in the English _kyind_, _duke_ (for _dook_), or the Greek ζορκάς (δyορκας) for δορκάς and the Magyàr _ágy_, “bed.” Conversely a palatal _i_ or _y_ may develop a dental sonant before it: thus the Italian _diacere_ comes from the Latin _jacere_, the Low Latin _madius_ from _majus_,[180] and the Greek ζειά (δyειά) and ζυγόν (δyυγόν) from _yava_ and _jugum_ (Sansk. _yugam_). In these instances we may trace the influence of emphasis; the parasitic letter is due to the attempt to speak with greater distinctness and solemnity.

But whether it be emphasis or the other two causes of change described in an earlier chapter, the pronunciation of sounds, like the meaning they convey, is in a constant state of flux. Nowhere is the dogma of Herakleitus, πάντα ῥεῖ, truer than in the history of speech. No two people pronounce exactly alike, nor does the same person always pronounce the same word or group of words in exactly the same way. Apart from the changes undergone by the pronunciation of words according to the sounds of the other words with which they may be associated, it is difficult to pronounce the same word when uttered singly twice in precisely the same way. The very effort to do so produces modification of the sound. Such shades of difference in utterance, however, are imperceptible to any but an unusually sensitive ear; it is only when the difference becomes considerable that it attracts notice. It then constitutes what we may term a variety, and such varieties we may hear sometimes from the lips of a single individual, sometimes from the members of a family, sometimes from those who live in daily contact and under the same conditions of life. The faculty of imitation is strong within us, and a particular pronunciation once started soon spreads, as it were instinctively, amongst those who are much together. It has often been observed how like the members of a family are to each other, not only in general appearance and manner, but still more in the use of similar expressions and idioms and the pronunciation of sounds. It is the same with schools, and to a less degree with universities to which the students come with their habits of phonetic utterance more or less formed: it has been said that the handwriting betrays the school at which the man has been educated; it may be said with equal justice that the mode of speaking does so too. In a savage state of existence, where tribe-life and village-life are on the one hand strict and intense, and the husband on the other hand sees but little of his wife and children, the conditions favourable to the growth of varieties in pronunciation are more numerous than among civilized men. The language of the nursery becomes in time the language of the tribe.

This phonetic variety may be broadly stated as mainly due to differences in the structure of the vocal organs. Putting aside imitation and analogy, putting aside, too, all wilful and conscious changes of pronunciation such as those enumerated on page 205, a particular sound or a particular way of pronouncing a sound may be easier to one speaker than to another. Very slight differences in the physical formation of the organs of speech may produce the most important consequences. And when a habit of pronunciation has once been fixed, it is difficult to alter it. The child who is learning to speak will as readily learn Chinese as English, the Japanese _r_ as the Northumberland _burr_; it is quite another matter when the attempt to catch the sounds of a new language has to be made in adult years.

Climate and food have, doubtless, an important effect in producing changes in the formation of the vocal organs; but at present we have no means of knowing the nature and extent of their influence. Professor March remarks of the change of _i_ to _g_ in Anglo-Saxon,[181] that “the movement (of consonants to vowels) is sometimes reversed, as when a nation moves northward, or northern peoples mix with a vowel-speaking race.” The Rev. W. Webster has drawn attention to the nasal twang which distinguishes not only American English, but American Spanish, Portuguese, and French as well; and which seems to be due to the dryness and the extremes of the American climate, while he further suggests climatic influences for the origin of the loss of the aspirate in Spanish words like _hijo_, pronounced _ijo_, the Latin _filius_, which in the fourteenth century still had _f_, and for the intensification of the aspirate in the corresponding Gascon words. We are all well acquainted with the hoarseness and roughness that exposure to the atmosphere lends to the voice, and the exercise and strength that a mountainous region gives to the lungs produce their effect in the vigour with which sounds are uttered. In cold countries the respiration is accelerated, while the air being denser contains a larger volume of oxygen.[182] The prognathism of the lower and older races of men, again, must have considerably modified their powers of utterance. “The lower jaw,” says Dr. Rolleston, “which in every well-marked variety of the human species contributes very importantly towards the making up of its distinctive character, was in the brachycephalous Briton usually a very different bone from the lower jaw of his Silurian predecessor.”[183] The strange fashions, too, which lead the savage to mutilate and deform his person, have frequently a very direct bearing upon phonology. Thus the loss and confusion of the labials and the excessive nasalization in the languages of the natives of the Pacific coast of America must be traced to the rings that are worn through the nostrils and lips of the people.[184] The Otyi-herero of South Africa is lisping in consequence of the custom of knocking out the four lower teeth, and partly filing off the upper front ones, to which also Professor Max Müller suggests the occurrence of the English, _th_ and _dh_ in the language may be due, and the Dinkas, who, like all the negroes of the White River, extract the front teeth of the lower jaw, have no sibilants.[185]

Whatever may be the causes which bring about varieties in pronunciation, certain it is that they are as continually making their appearance as varieties in the realm of natural history. Where they are unrestrained by the conservative tendencies of literature and education, they soon spread from the individual and the household and become species or dialects. The dialect itself may in course of time assume so marked a character of its own, and be so widely spoken as to be accounted a separate language; and will stand to the varieties and species destined to grow out of it in the relation of a genus to its species. But with this further development phonology has little to do.

It is otherwise with the changes which result in the rise of a new dialect. Comparative philology is based on the recognition that the same word will be represented by different combinations of sounds in a group of allied dialects or languages, and that each combination will be governed by a fixed phonetic law. An English _h_, for example, will answer to a Greek and Latin _k_, an English _t_ to a German _z_ and a Sanskrit _d_. When once a sound is given in a language, we may know the sounds which must correspond to it in the cognate languages. Now and then, of course, subordinate laws will interfere with the working of the general law; but unless such an interference can be proved, we must never disregard the general law for the sake of an etymological comparison, however tempting. To compare the Greek θεός with the Latin _deus_ and the Sanskrit _devas_, rests upon almost as unstable a foundation as the old derivation of _whole_ from ὅλος, and _call_ from καλέω.[186] We must never forget that the laws of phonology are as undeviating in their action as the laws of physical science, and where the spelling does not mislead us will display themselves in every word of genuine growth. Even the vowels cannot be changed and shifted arbitrarily; they, too, follow definite laws of development, and though it is not yet possible to state their equivalence in the several languages of a single family with the same precision as in the case of the consonants, we may feel quite sure that this is the fault of our ignorance and not of the facts themselves.

It was the great Grimm who, following in the wake of Rask, first formulated the empiric law of that regular _Lautverschiebung_, or shifting of sounds, in our Indo-European family of speech which has since gone under his name. Since his time the law has been the subject of much discussion and examination;[187] his statements have been amended and amplified, and an endeavour made to apply the same law to the vowels that has been applied to the consonants. The following table[188] exhibits the equivalence of sounds in the Aryan family of speech:—

+---------+---------------+-------------+-----------+--------------+
| | Sanskrit. | Zend. | Greek. | Latin. |
+---------+---------------+-------------+-----------+--------------+
| K | ś (ç) | ç | κ | c |
| Kw }| k, ch, p | k, ch, p | κ, π, τ | qu, c, v |
| (K²) }| | | | |
| G | j, sh | z, sh | γ | g |
| Gw }| g, j, k | g, j, zh, k | β, γ[189] | [g]v, b |
| (G²) }| | | | |
| G H | h | z | χ | h, g |
| G Hw }| [g] h | g, j, zh | χ, φ | v, gv, g |
| (G H²) }| | | | |
| T | t | t | τ | t |
| D | d | d | δ | d, l |
| D H | [d] h | d | θ | f, d, b[190] |
| P | p | p | π | p |
| B | b | b? | β | b |
| B H | [b] h | b | φ | f, b |
| N G | ṅ | ñ | γ | ng |
| N | n | n | ν | n |
| M | m | m | μ | m |
| R | r, l | r | ρ, λ | r, l |
| Y | y | y | y, ζ, δ | j |
| V | v | v | ϝ, υ, ῾ | v |
| S | s | h, s | σ, ῾ | s, r |
| A | a | a | ε | ĕ |
| A² | â, a | — | ο | ŏ, ĕ |
| A³ | a, i, u, î, û | — | α, ο | a, o |
| Â | â | â | ᾱ, ω | ā, ō |
| I | i | i | ῑ | i |
| U | u | u | ῠ | u, o, i |
+---------+---------------+-------------+-----------+--------------+

+---------+---------------+-------------+-----------+--------------+
| | Oscan and | Gothic. | English. | Modern High |
| | Umbrian. | | | German. |
+---------+---------------+-------------+-----------+--------------+
| K | k | h, g | h, g | h, g |
| Kw }| p | hv, f(p), h | wh, f | w, f |
| (K²) }| | | | |
| G | g | k | k, ch | k, ch |
| Gw }| b | kv | qu, c | qu, k |
| (G²) }| | | | |
| G H | h | g | g, y | g |
| G Hw }| ? | g, v? | g, w? | g, w? |
| (G H²) }| | | | |
| T | t | th, d | d, th | d, t |
| D | d | t | t | z, ss, sz |
| D H | f | d | d | t, th |
| P | p | f, b | f, b | f, b |
| B | b | p? | p? | pf?, f? |
| B H | f | b | b | b |
| N G | ng | ng | ng | ng |
| N | n | n | n | n |
| M | m | m | m | m |
| R | r, l | r, l | r, l | r, l |
| Y | j | j | y | j |
| V | v | v | w | w |
| S | s, r, z | s, z | s, r | s, r |
| A | e | e | — | — |
| A² | — | — | — | — |
| A³ | — | a | — | — |
| Â | — | — | — | — |
| I | i | i | — | — |
| U | u | u, au | — | — |
+---------+---------------+-------------+-----------+--------------+

+---------+-------------+-----------+----------+--------+------------+
| | Lithuanian. | Church | Gaulish. | Old | Old |
| | | Slavonic. | | Irish. | Welsh. |
+---------+-------------+-----------+----------+--------+------------+
| K | sz | s | c | c, ch | c |
| Kw }| k, p | k, p | p | c, ch | p |
| (K²) }| | | | | |
| G | ż | z | g | g | g |
| Gw }| g | g | b? | b, m? | b, m? |
| (G²) }| | | | | |
| G H | z | z | g | g | g |
| G Hw }| g | g | b? | b | b |
| (G H²) }| | | | | |
| T | t | t | t | t, th | t |
| D | d | d | d | d | d |
| D H | d | d | d | d | d |
| P | p | p | ... | ... | ...[191] |
| B | b | b | b | b, m | b, m |
| B H | b | b | b | b, m | b, m |
| N G | ng | -n | ng | ng | ng |
| N | n | n, -n | n | n | n |
| M | m | m, -n | m | m, b | m, b |
| R | r, l | r, l | r, l | r, l | r, l |
| Y | j | j | j | ... | j, ddj, dd |
| V | v | v | v | f, b | gu, u |
| S | s | s, ch | s | s | h |
| A | e | e | — | — | — |
| A² | o, à | o | — | — | — |
| A³ | a | a | — | — | — |
| Â | — | — | — | — | — |
| I | i | i | — | — | — |
| U | u | u, o, ŭ | — | — | — |
+---------+-------------+-----------+----------+--------+------------+

Some of the changes of sound recorded in the above table are as old as the undivided Aryan speech itself. They go back to the dialects that existed in the earliest period of which our materials allow us to know. Instead of clinging, with Fick, to a genealogical tree, and deriving the Aryan languages of Europe and Asia from two parent-stems, Western and Eastern Aryan, and these again from a single _Ursprache_ or primitive speech, it is better to follow J. Schmidt in tracing the later languages to co-existent dialects, which by the loss or absorption of intermediate dialects and the migration of the speakers became more and more distinct and divergent one from the other. It is, of course, quite possible that the speakers of the most western of these dialects moved across the Ural range into Europe in a compact body, and there settled for a while in a district westward of a line drawn from Königsberg to the Crimea, where the beech grew, and that it was from this second home of the Aryan race that the waves of European emigrants successively broke off. Certainly Professor Fick seems to have shown the common possession of certain phonetic peculiarities, such as the vowel _e_, by the Western as distinguished from the Eastern Aryans, and the Eastern or Indic branch of the family clearly once formed a single whole which subsequently divided into Iranian and Hindu. Unfortunately the position of Armenian and the allied dialects is still a matter of doubt; and there are scholars who would regard them as a link between the European and the Asiatic sections of the Aryan group. But Fick labours hard, and apparently with success, to prove that the Aryan dialects of Asia Minor, such as we know them from glosses and inscriptions, belonged to the European, not the Asiatic section, while Armenian, on the other side, is an Iranian tongue. Fick’s conclusion is confirmed by the evidence of the cuneiform inscriptions. Up to the eighth century B.C. Armenia was still inhabited by tribes who spoke non-Aryan languages, and it was only a century previously that the Medes had first forced their way into the country regarded by the agglutinative Accadians as the cradle of their race, but which was afterwards to be the seat of the Aryan Medes. Eastward of the Halys there was nothing Aryan until long after the occupation of Armenia by the new-comers.

We have certain proof that the series of changes which resulted in the formation of High German took place subsequently to the overthrow of the Roman Empire. Latin words for instance like (_via_) _strata_ or _campus_, adopted by the Teutons during the era of their wars with Rome, are found in both Low and High German in the very forms which the application of Grimm’s law would require them to have were they native words. Thus _strata_, Low German _strata_, our _street_, becomes _straza_ in Old High German, _campus_, our _camp_, similarly becomes _kamph_, _kampf_. The Hessians were called Catti in Roman times, and though now High Germans, had the same ancestors as the Batavi, from whom the modern Dutch draw their descent, while the Malbergian glosses show the language of the Franks to have been Low German, although the Franconians of to-day, who are descended from the same stock as the Suabians and Ripuarians, speak High German. Here, at any rate, we have an instance of a series of varieties finally resulting in a new language in historical times.

It must not be supposed that all the changes of pronunciation that serve to distinguish one branch of the Aryan stock from another took place simultaneously. On the contrary, they were slow and gradual; first one and then another new fashion in sounding words sprang up and became general: when once the new pronunciation had, from any cause, taken a firm hold of the community, analogy caused every word to be submitted to its influence, unless special reasons, such as accent, stood in the way, until in course of time the process of shifting the sounds was completed. An instructive illustration of this shifting of sounds has lately been going on almost under our eyes. In the Samoan Islands of the Pacific only fifteen years ago _k_ was an unknown sound except in one small island of the group, where it replaced _t_. Since then it has practically disappeared from all of them, and _t_ has taken its place. What makes the rapidity of the change the more extraordinary is that the speakers of the language live on separate islands, and that intercourse between them is less intimate now, according to Mr. Whitmee, than it was in the days of heathenism. And yet in spite of books and schools, in spite of education and every effort to check it, the change has come about. The natives will ridicule the foreigner who pronounces in the new fashion, they will themselves take pains to sound the _k_ when reading aloud or making a set speech, but in conversation it has ceased to be heard. The tendency to put _k_ for _t_ seems to be irresistible; it is in the air, like an epidemic, and the spelling, so recently introduced, no longer represents the common pronunciation of the people.[192]

We must be on our guard against thinking that the sounds represented by the same letter of the alphabet in different languages are really identical. We have seen of what numberless variations each sound that we utter is capable, and it does not follow that because the Sanskrit _cha_ and the English _church_ are written with the same palatal _ch_, that therefore they are to be pronounced alike. And what is true of the consonants is still more true of the vowels. There is much to show that the European scale of three short vowels—_ă_, _ĕ_, _ŏ_—is more primitive than the Indic single vowel _ă_, in which three distinct vowel-sounds of the parent-speech have coalesced, but we cannot infer from this that the three vowel-sounds of the parent-speech were actually _ă_, _ĕ_, and _ŏ_. Indeed, when we remember that the Greek ἕκατον (for ἕν-καντον) corresponds to the Latin _centum_, while _ferentis_ is represented by φέροντος, it is quite clear that the Latin _ĕ_ must have developed out of one or more sounds which were distinct from it. In dealing with the hypothetical Parent-Aryan it is best, with Brugman, to symbolize these three primitive vowels as _a_¹, _a_², and _a_³.[193] It is possible that some at least of the earlier sounds out of which more than one articulate sound have afterwards developed, were of a vague indeterminate character, not properly-formed vowel utterances. Professor Max Müller[194] quotes authorities to prove that in the Sandwich Islands _k_ and _t_ are undistinguished, and that “it takes months of patient labour to teach a Hawaian youth the difference between _k_ and _t_, _g_ and _d_, _l_ and _r_.”[195] The confusion between _k_ and _t_, however, has already been explained by the similar fact observed in Samoan where the sound has actually changed within the last fifteen years, a distinctly-articulated _k_ becoming an equally distinctly-articulated _t_. But even in English we find people saying _a cleast_ instead of _at least_, while at Paris and elsewhere the lower classes say _amikié_ for _amitié_, _charkier_ for _charretier_, _crapu_ for _trapu_.[196] So in the old Paris argot _j’équions_ stood for _j’étais_, and in Canada the uneducated part of the population says _mékier_ for _métier_, _moikié_ for _moitié_. Bleek, again, writes of the Setshuana dialects: “One is justified to consider _r_ in these dialects as a sort of floating letter, and rather intermediate between _l_ and _r_, than a decided _r_ sound.”[197] To these instances of confusion between two consonants which Professor Max Müller believes to be “a characteristic of the lower stages of human speech,” may be added the fluctuation between two forms of the same sound in the North German dialects, where no distinction is made between surd and sonant _mediæ_, as well as in many of the Armenian dialects.[198] But we must bear in mind that this childlike inability to distinguish between sounds may be due to two very different causes. It may be a result either of the sound being formed at the neutral point, as it were, intermediate between two distinct sounds, or of the ear being unable to discriminate between different articulations. The latter cause is analogous to colour-blindness, and has most to do with the imperfections of childish utterance or the substitution of _r_ for _l_ so often heard; the other cause is of a purely phonetic character, and takes us back to the time when man was gradually fashioning the elements of articulate speech. This infantile state of language had probably been long left behind by the cultivated speakers of the Parent-Aryan; indeed, the very existence of the three vowels marked _a_₁, _a_₂, and _a_₃, would imply that such was the fact. If there was any confusion in the pronunciation of their words it would have to be ascribed rather to sound-blindness than to imperfection of utterance.

The regular action of Grimm’s law may be interfered with by the influence of other laws, just as in physical science the regular action of the law of attraction may be interfered with from time to time. Foremost among these disturbing agencies is the accent. K. Verner has shown[199] that the position of the accent has occasioned that apparent disregard of Grimm’s law in the Teutonic languages which has produced _mutter_ and _vater_ (O. H. G. _muotar_ and _fatar_) by the side of _bruder_ (O. H. G. _brôpar_), _sieben_ (Goth. _sibun_) by the side of _fünf_ (Anglo-Saxon _fîf_), _schwieger_ (O. H. G. _swigar_ = ἑκυρὰ, _so-cru-s_) by the side of _heil_ (Greek καλός), or such a curious change in the conjugation of the same verb as the Anglo-Saxon _lîðe_,“I sail,” but _liden_, “sailed.” The same cause has brought about the varying representation of an original _ſ_ now by _s_, and now by _z_ or _r_. In the Veda, _bhrâtar_ is accented on the first syllable, like the Greek φράτηρ, _mâtár_ and _pitár_ on the last, again like the Greek μητήρ and πατήρ. _Sieben_ answers to the Vedic _saptán_, the Greek ἑπτά, whereas _fünf_ is the Vedic _pánchan_ and Greek πέντε. _Schwieger_ similarly goes back to the Vedic _´swa´srû´_, Greek ἑκυρά, just as the O. H. G. _snura_ from _snuza_ goes back to the Vedic _snushâ´_, Greek νυός, in contradistinction to _nase_, _nose_, the Vedic _nâ´sa_, the Lithuanian _nósis_. If we turn to the verb, we find that in Anglo-Saxon, whereas the present _lîðe_, “(I) sail,” corresponds with a Vedic _bhédâmi_, and the singular of the past tense _lâð_ with a Vedic _bibhéda_, the plural of the preterite _lidon_ corresponds with a Vedic _bibhidús_.[200]

There are other influences besides that of the accent which may change and mar the face of words. Although every change takes place in strict accordance with phonetic laws, and is consequently capable of explanation, the occurrence of the changes is more or less sporadic and arbitrary. That is to say, they may act upon one word and not upon its neighbour. In _should_ or _would_, for instance, _l_ has been assimilated to _d_, but in _fold_ and _cold_ it still maintains its existence. Such changes may be either independent or dependent on the action of surrounding sounds. The diversification of the Teutonic _a_ into _e_ and _o_, or the transition of the Latin _ĭ_ and _ŭ_ into Romanic _e_ and _o_ are instances of independent change. So, too, the modern English pronunciation of the vowels with passive lips, and the consequent loss of the intermediate vowels _ü_ and _ö_, is another example of the same facts. Wherever, indeed, these intermediate vowel-sounds exist, we may feel sure that the lips take an active part in articulation. In all these cases the change happens in the formation of the sound, uninfluenced by the neighbourhood of other sounds. The extension of a simple vowel into a diphthong may also be brought under this head, though the presence of the circumflex accent seems to have much to do with it. On the other hand, changes in the dentals, the passage of _z_ into _r_ and _r_ into _l_, or the transition from a guttural to a palatal and a dental, are all examples of purely independent change. When we find an Aryan _kw_ (_k_²) and _gw_ becoming _ch_ and _j_ in Sanskrit or τ in Greek, we merely see the gradual forward movement of the tongue, which is moved with less exertion towards its tip than towards its root. The change of Aryan _kw_ and _gw_ into _p_ and _b_ in Greek (as in πίσυρες and βίος[201]) is held by Sievers to be due to a sudden “leap” in the articulation, _k_ and _g_ partially assimilating the second part of each compound into _p_ and _b_, and then falling away altogether.

Most of the changes recorded in Grimm’s law may be brought under the head of independent change. No doubt the transition of _g_, _d_, _b_, into _k_, _t_, and _p_ in German is partially dependent upon the accent, but the growth of an aspirate out of a _tenuis_, as exemplified in the Irish pronunciation of English, is probably due to nothing but an increase in the energy and duration with which our breath is expired. The want of the stress accent brings about the shortening and loss of final vowels, the tonic accent, on the other hand, tending to lengthen them.

The changes caused by the action of one sound upon another may be divided into those which are due to assimilation, and those that are not. In either case the time occupied in pronouncing the changed sound remains the same as it was before; it is only in cases of independent change that it may differ. Assimilation is effected in one of two ways. The relative positions of the vocal organs needed for the pronunciation of two sounds may be made to approximate, as in the reduction of _ai_ (_a_ + _i_) to _e_, or the time that elapses between the pronunciation of two sounds may be reduced or destroyed altogether, as when _supmus_ becomes _summus_. Where the change is not due to assimilation, it will be found to depend on an alteration in the time needed for the formation of two or more sounds.

Assimilation may be regressive, progressive, or reciprocal. Regressive assimilation is where a sound is assimilated to that which follows it, as in ἕννυμι for ϝεσ-νυμι, from the root _vas_, or ποσσί, for ποδ-σι (ποδ-σϝ-ι), and γράμμα for γράφ-μα(τ). Progressive assimilation is the converse of this, as in στέλλω for στελ-yω, μᾶλλον for μαλ-ιον, _mellis_ for _melv-is_, or the Æolic ἔστελλα for ἔστελ-σα. Regressive assimilation largely preponderates in our Aryan languages, progressive assimilation in the Ural-Altaic ones; and it is very possible that Sievers is right[202] in tracing this contrast to the difference of the accentuation, which in Ural-Altaic falls upon the first syllable of the word, while in the parent-Aryan it fell for the most part on the final syllable. Böhtlingk[203] says, very appositely: “An Indo-Germanic word is a real whole of such a kind that the speaker has uttered the whole word, as it were, in spirit, as soon as he has pronounced the first syllable. Only in this way can it be explained how a syllable (or sound) is modified in order to assist the pronunciation of the syllable (or sound) that follows it. A member of the Ural-Altaic race forces out the first syllable of a word—that part of it, namely, which has the accent—little caring for the fortune of the rest; on this he next strings in more or less rude fashion a few more significant syllables, only thinking of a remedy at the moment when he first feels the want of one.” As for reciprocal assimilation, an example of it may be found in the reduction of _ai_ to _e_ quoted above, where both sounds influence one another.

Assimilation may be either complete or partial. There are sounds which can never be thoroughly assimilated to each other, _bn_, for instance, can never at once become _nn_, only _mn_. Partial regressive assimilation meets us very frequently in the classical languages; _e.g._, λεκ-τός from the root λεγε, ἤνυσμαι from ἀνυτ-, δόγμα from δοκ-; partial progressive assimilation is rarer; e.g., πάσχω for πάσκω from παθ-σκω.

The changes dependent on the presence of a second sound, which are not due to assimilation, are necessarily produced by varying the time needed for pronunciation. Of these the most striking is metathesis. Metathesis must be referred rather to a mental than to a phonetic origin. Our thought and will outstrip our pronunciation, the result being that the sound which ought to follow is made to precede, or else the vocal organs are shaped prematurely for the formation of a sound which ought to be heard later, the consequence being that the sound which should come first has to come last. Metathesis, in fact, is similar to the rapidity, or rather relaxation, of thought which leads us sometimes to write or speak a word which belongs to a subsequent part of the sentence; and it may be of two kinds: either the place of two sounds may be simply inverted, or the second sound may be made to precede the first by two or three syllables. How easily the first case can happen is shown by the phonograph, where each syllable that has been uttered can be reproduced backward by merely turning the handle of the machine the wrong way. _R_ and _l_ are the most subject to metathesis, then the nasals; the other consonants vary according to their relationship to the vowels. More regular than metathesis are the insertion and omission of consonants, as in ἀν-δ-ρὸς, ἄ-μ-β-ροτος, τέτυφθε for τέτυφσθε, _rêmus_ for _resmus_. Somewhat different are the insertion and omission of vowels, the first of which goes under the technical name of _Swarabhakti_. This name was imported from the Hindu grammarians by Johannes Schmidt,[204] to mark the growth of a short or reduced vowel from a liquid or nasal, when accompanied by another consonant. Thus _ănman_, “name,” became _ănă-man_, and then, by the loss of the first vowel and the compensatory lengthening of the second, _nômen_ and _nâmâ_. _Swarabhakti_ is, however, incompatible with the acute accent. We may find examples of it in the slow pronunciation which in English turns _umbrella_ into _umbĕrella_, and _Henry_ into _Henĕry_.[205] Prosthesis, or prothesis, the insertion of a short vowel at the beginning of a word before two consonants, is another illustration of _Swarabhakti_. There are many nations which find a difficulty in pronouncing two consonants at the beginning of a word. Thus the Bengali calls the English _school_ _yschool_, the Arab says _Iflatún_ for _Platon_, and the Ossete uses _a_ for the same purpose. In other cases, one of the consonants is dropped altogether, as so frequently by children and systematically by the natives of Polynesia. In Latin inscriptions and MSS. later than the fourth century we find forms like _istatuam_, _ispirito_, just as in the Romanic tongues we have _estar_ and _espée_ (_épée_) for _stare_ and _spada_, or in Welsh _ysgol_ from _schola_, _yspryd_ from _spiritus_. According to Wentrup,[206] _a_ is often used as a prothetic vowel in Sicilian; Lithuanian has forms like _iszkadà_, German “_schade_,” and Basque and Hungarian prefix a similar aid to the pronunciation. No trace of a prothetic vowel can be found in Latin; in Greek, however, such vowels are very plentiful. Thus we have ἄσταχυς by the side of στάχυς, ἐχθές by the side of χθές, ἰγνύη by the side of γόνυ, Ὀβριαρευς by the side of Βριαρεύς. In Greek, too, as in other languages where prothesis occurs, the complementary vowel may be inserted before a liquid, more especially _r_, as well as before a strictly double consonant, _e.g._, ἀμύνω by the side of μύνη, ἐρυθρός by the side of _ruber_, ὀρέγω by the side of _rego_. Even the digamma may perhaps take the prefix as in the Homeric ἔεδνον. But it is probable that no other single consonant does so, the apparent exceptions being really explained by the loss of a consonant which once existed along with the one that is left. Ὀκέλλω, for instance, presupposes ὀ-κϝέλλω (Latin _pellere_), Ἀπόλλων presupposes Α-κϝολιων, “the son of the revolving one” (Sanskrit _char_, Greek πέλομαι). In other cases we are dealing not with a prothetic vowel, but with a part of the primitive root: ὄνομα, for example, is shown by the Irish _aimn_ and Old Prussian _emnes_ to be more original than the Sanskrit _nâmâ_ or the Latin _nomen_, and to stand for an earlier _an-man_; and ὄνυξ, the Latin _unguis_, the Irish _inga_, is earlier in form than the Sanskrit _nakha_ and the English _nail_ (_nagel_)[207]. We may discover a tendency in Greek to adapt the prothetic vowel to that of the root, though it is hardly so regular as in Zend roots beginning with _r_, where we find _i-rith_ for _rith_, but _u-rud_ for _rud_. Sanskrit, like Latin, shows an inclination rather to drop initial vowels than to add them, but even in Sanskrit, Curtius has pointed out[208] the Vedic _i-raj-yâmi_ from _raj_ (_rego_) and _i-radh_, “to seek to obtain,” from _râdh_. As for the loss of a vowel, it is too familiar to every one to need any illustration.

More akin to metathesis is epenthesis, which closely resembles the Teutonic _umlaut_. Epenthesis is especially plentiful in Greek, where κτέν-yω becomes κτείνω, χερ-ιων χείρων, λόγοσι λόγοις, ἐλαν-ϝω ἐλαύνω, νερϝον νεῦρον. Probably λέγει for λεγειτ is to be explained as resulting from the epenthesis of ι (λεγειτ for λεγετι), just as λέγεις stands for an earlier λεγεσι. Epenthesis thus presupposes a mouillation or labialization in which the articulation of the consonant is absorbed, as it were, by that of the _i_ and _u_. The greater the participation of the lips and tongue in the formation of these vowels, the greater will be the tendency towards epenthesis.