peculiarities of the Ethiopic consonants compared with the Arabic were developed in Abyssinia, and have all an African character.

That character (see I. 8, 24, 25, 35, 57, 74; III. 126, 161) exhibits generally the tendency to utter the consonants without that tension which is given to them by pressure of breath from the chest, and this is apt to detach the consonant from the vowel which follows it (Def. 26). This tendency is to be seen in all the changes which the Arabic consonants have undergone in Ethiopic.

The failure of the tension from the chest rendered it necessary either to speak with breath from the chest without tension or to utter the consonants with the breath that was in the mouth or above the larynx, pressing this on the seat of the utterance by contraction of the parts behind. The latter tends to give hardness by the compression, the former to reduce the consonant to a breathing. Both tend to cause the decay of those gutturals, which require for their due utterance tension from the chest. The tenuis q indeed can be uttered with compression of the cavity between the larynx and the root of the tongue; and the utterance of the post-palatal k in the same way tends, in the effort to contract the space behind, to move the closure of the tongue backwards so as to produce q; and thus sometimes this consonant was favoured, k being restricted to a weaker utterance.' But was reduced so as to approach to h; and x and x gradually gave up their tension, and came to be uttered like h,2 though in some cases the effort to give tension without pressure from the chest hardened these consonants to q, k, or g.3

The effort to compress the breath in the mouth, in order to make the utterance sensible, was unfavourable to the soft consonants ď, 0, and, and these were early given up; but t and were strongly uttered, the former "with a raising of the root of the tongue against the hinder part of the gums," "4 the latter with a dental sibilation; d was preserved as well as d; but t tended to prevail over it. And though there are many exceptions, the more usual correspondences are tort to d in the other languages, t to d' and 0, t' to d d z s or s, and d or z to 9.5 It is better in Ethiopic to write instead of t. For in Amharic there is a true t'in addition to the t, though

t' originally was ante-palatal.

The same tendency to compression produced, among the labials, puttered explosively with compression of the mouth, and an aspirate p' in which the aspiration is sent over the tongue to the lips producing an accompanying sibilation. In the Gã also (I. 62) there is a labiolingual f

8

The dental sibilant s tended to prevail over the ante-palatal s,' because it admitted a larger cavity between the tongue and the palate, by contraction of which a sibilation was more easily produced.

9

The detachment of the consonant from the vowel which follows it, appears in the peculiar utterance of p, in which "the breath puffs off from between the lips before the vowel is heard;' "10 and also in the 1 Dillmann, p. 39. 2 Ibid. pp. 34, 38. 3 Ibid. p. 40.

4 Ibid. p. 43. 5 Ibid. pp. 44, 48, 52. 6 Gesenius, Hebrew Lexicon, p. 778. 8 Zimmermann, p. 5. 9 Dillmann, p. 51.

7 Dillmann, p. 45.

10 Ibid. p. 45.

tendency of the gutturals and post-palatals q, X, k, and g, to incorporate w before any vowel except u or ō. This w sounds breath which would. be lost to vocal utterance in the beginning of the vowel if this were uttered through open organs after a consonant which involved little pressure of breath from the chest (Def. 26). Being close it lets little breath pass, and it produces a compression of breath, the removal of which reinforces the vowel following.1 This feature is found in many African languages, which also tend to insert y in the same way by reason of their palatal nature.

The vowels u and ō combine more closely than the others with the post-palatal and guttural consonants, so that probably the breath for their utterance presses on the organs before the closure is opened.

The tendency to incorporate w is brought into action generally where an original u has been either changed into another vowel or v, or absorbed by the consonant as w on account of the affinity of the consonant for it. And this may take place not only when the u follows the consonant immediately, but even when it follows a preceding or following consonant. But sometimes the w is taken by g when g with w takes the place of k, or q without it, and sometimes by k, when kw takes the place of q or X, X; the w making the softer consonant harder and more guttural, and therefore less different from the consonant for which it stands. Sometimes also the w is taken when such occasions for it are not present; as, on the other hand, sometimes w is not taken when such occasions might seem to invite it.2

121. Ethiopic, like Arabic, admits open syllables with a short vowel accented or unaccented; and, like Hebrew, it admits closed syllables with a long vowel without requiring, as Hebrew does, that the vowel should be accented. It also admits two consonants at the end of a word. And every syllable must begin with a consonant, and, as originally formed, only with one. The general rule is that before two consonants at the end of a word the vowel must be short. But when the first of the two is a guttural or post-palatal spirant, an a preceding it must be long; and when it is y or v it may sustain a long vowel before it.4

3

The concurrences ă + i and ă + u generally form the diphthongs ai, au, but often the long vowels ē, ō, which may also arise from ia, ua. If the first vowel be long the second becomes a semi-vowel.5

The post-palatal and guttural spirants are helped in their utterance by a vowel preceding or following them. The vowel for which they have most affinity is a; but if they have another vowel than a, then an a preceding is, by attraction of the spirant with this vowel, apt to be changed to e. They tend to lengthen a preceding vowel, giving their breath partly to it, and are themselves weakened thereby, and may be lost; but instead of giving breath to the vowel they may take breath from it and reduce it to ě. When uttered with an a following them they have an attraction for the accent.6

The semivowel v, which was probably uttered from the throat as

1 Dillmann, p. 67. 4 Ibid. p. 58.

2 Ibid. p. 41-43.

Ibid. pp. 63, 64.

3 Ibid. p. 55-57.

Ibid. p. 68-74.

well as from the lips, is in Ethiopic much weaker than y, and the vowel u than i (see 75); the muscular action of the organs in uttering y and being much the stronger. Yet as a first radical, y is very rare, and v very frequent, the language being kept guttural by the tendency to combine w with the gutturals and post-palatals.1

A final q of a verbal stem assimilates to itself an initial k of the person ending; and a final t or d of a noun assimilates to itself t of the feminine ending; t and d before s become s.2

The accent is most frequently on the penultimate syllable, more frequently on the antepenultimate than on the ultimate. A vowel long by nature or position has an attraction for the accent, as well as a syllable with a strong meaning. There are many enclitic monosyllables.3

A long vowel in a syllable tends to reduce the vowels in the adjacent syllables; a and u prefer ě, but i, which takes less breath, is content with ǎ.1

122. Pluriliteral verbal roots are formed by repeating a whole root, generally reduced to a monosyllable, or the last two radicals of a root, or by inserting n, sometimes r, after a first radical. The duplications express ideas which involve repetition, movement, duration, intensity, completion; but generally the simple roots from which they were formed are no longer found. Sometimes in a root consisting of a closed syllable repeated, the second consonant is assimilated to the third, so as to double it, and thus (and thus only) roots are formed whose first and second consonants are the same.5

Verbal roots also consist sometimes of a triliteral root with a formative prefix, being originally derived forms, which came subsequently to be thought as simple verbs; and sometimes they consist of a triliteral root or short noun with ya, va subjoined, which as final syllable of a root, whether triliteral or pluriliteral, has generally a causative or transitive significance.

Less frequently a guttural spirant is added instead of y or v. Nominal stems also are turned into verbal stems without dropping their nominal formatives.

Roots with more than three letters are so numerous in Ethiopic that they form a sixth or seventh of all the roots of the language.7

123. This large development of roots having more than three radicals is a remarkable feature of the Ethiopic language. Their mode of formation is for the most part quite according to the genius of the Syro-Arabian languages. Many of them, as has been said, are regular derived forms from triliteral roots. And the reduplication which shows itself in others is not only to be seen in the second, fifth, ninth, and eleventh derived forms of the Arabic verb and in some of the Arabic quadriliterals, but is in agreement with a tendency which may be observed in these languages to strengthen an idea by repeti tion rather than by a comparative element, owing to their weakness in comparative thought (66, 92, 117). The formation, however, of a

1 Dillmann, pp. 82, 104. 5 Ibid. p. 101.

2 Ibid. p. 84.

6 Ibid. pp. 105, 111.

3 Ibid. p. 90. • Ibid. p. 91.

7 Ibid. p. 107-113.

root with a transitive or causative significance by subjoining an element instead of prefixing one, does not agree with the true SyroArabian subjectivity. For the original root to which this addition was made, being placed first, must have been thought in its general associations among the facts of the world, showing the predominance of an external interest, instead of being limited by a subjective prefix to the thought of it as launched from a subject to an object. But the most noteworthy character of these pluriliterals is that they are thought as roots, not as derivative stems, the roots from which they were originally formed having for the most part disappeared from the language.1

Now, in the process of this displacement, the original roots must have become quite merged in the new formations; for if they had continued to be felt in these in their integrity they would have still remained in the consciousness of the race. The new formations, as they were used in speech, must have become abbreviated and reduced in meaning, and the original roots been thereby so weakened as to lose their original significance. So that in this feature of the language we have evidence of a contraction of the object thought by the mind in a single act such as might be expected from African influence (see II. 3).

The old roots in these formations might be regarded as having an analogy to Indo-European roots, which are not found separate. But it is only in these formations which have added elements either before or after the roots that such analogy is apparent. The reduplicated roots are not agreeable to the Indo-European genius, which affects its roots not so much with reduplication as with relative or comparative elements.

This tendency to contract the single acts of thought would be favoured by any weakness of the sense of the root or of the derivative element in the ideas which the formation was used to express. And only in those formations which had such weakness would it show itself by reducing them to a radical idea. But the extent to which it prevailed in Ethiopic compared with the Asiatic members of this family, and the extent to which the derived forms of the verb supplanted in the same way the simpler forms, show the reality of its operation. To this cause also is due the prevalence of the formation of causative of reflexive, which was facilitated by the reduction of the reflexive.

124. In the Ethiopic simple triliteral stem, the vocalisation of the third singular perfect is the same as in Arabic, except that in intransitive verbs the i and the u of the second radical 2 have both become ě, showing weakness of subjectivity.

The second form of the Arabic verb is in Ethiopic also, with the same significations, but generally the simple form is not retained along with it. And when the simple form is retained along with it there is scarcely any difference of meaning. The two last radicals of a root are sometimes repeated to express continuance or periodical 2 Ibid. p. 116.

1 Dillmann, pp. 107, 109, 111.

VOL. II.

F

repetition, or the play of colours. Less frequently the last radical is; doubled to express continuance or completeness, or a clinging state.1

The third form also is in Ethiopic, but it is not very frequent, and is partly replaced by its own reflexive form. And those verbs which have the third form either do not occur in the simple form or in the second, or if they do, the meaning does not differ.2

A fourth or causative form is formed in Ethiopic from each of the three preceding ones, in the same way as in Arabic from the simple stem. Often enough the simple stem is no longer in use along with its causative, but only the second form; the simple stem having been weakened by being merged in the causative.

The causative of the second form is much more uncommon than that of the first or simple form. It rarely has the same meaning as the second form. Sometimes it exists along with the causative of the first form, and generally with a different meaning, though sometimes with the same.4

The causative of the third form is very rare, as that form itself is little used.5

There are reflexives of the first, second, and third forms, all, like the Arabic fifth and sixth, formed by prefixing ta. The reflexive formations are the only expression of the passive; there not being sufficient sense of the verb in its effect in the object to maintain the passive 5 (79, 102). The reference to the reflex object being direct, the verb may often govern an indirect object." As the third form is used to express an action reaching to an object, its reflexive may either have the same meaning or may express reciprocity.7

Causatives are formed on the three reflexives by prefixing has; but as the first two reflexives differ less in meaning from the first and second forms than the third reflexive from the third form, the causatives of the first and second reflexives are much oftener replaced by the causatives of the first and second forms than the causative of the third reflexive by that of the third form. This last causative is consequently much more frequent than the others. It expresses causation of the reciprocal, even though the third reflexive be no longer in the language; or causation of gradual completeness or preparedness, though the third reflexive either does not occur or is found only in quite another signification.8 For the derived form tends to put out the simpler form corresponding to it, by reducing it to a mere part of an idea.

Thus of the twelve verbal stems almost every one may be formed independently of the others from a verbal root or from a nominal stem. But it is not to be supposed that any root has the twelve stems. The richest is gabra, which has six in ordinary use. The more prolific roots have five, namely, a first, second, or third, a causative, a reflexive, a causative of reflexive, and a reciprocal. The most have only an active, a reflexive, and perhaps a reciprocal or a causative of

1 Dillmann, p. 117-119.

4 Ibid. p. 122.

7 Ibid. p. 126.

2 Ibid. pp. 119, 120.

5 Ibid. p. 123.

8 Ibid. p. 127-130.

3 Ibid. p. 121. • Ibid. p. 124.