PJ 302.7 Ha I •^ t " "L* * 8 \ ^« A^ >v* v V\^ /./*** •-. ^ ^ "V- <-. ■ ' ANALOGY SEMITIC LANGUAGES DISSERTATION PRESENTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY V - ABEL H. HUIZINGA it BALTIMORE: press of isaac friedenwald co. i8qi. « ™ % 5»M *• ANALOGY IN THE SEMITIC LANGUAGES. PART I. The received opinion on the nature of analogy as a linguistic phenomenon, and on the relation of analogy and phonetic law, may be stated as follows i 1 i. The phonetic laws are absolutely without exception. There are not two classes of sound-changes, regular and irregular, syste- matic and sporadic. 2 In speaking of phonetic law, however, it must be remembered that the idea of law is conditioned by the sphere in which it works and the material to which it is applied. We cannot speak of phonetic law in the same sense in which we speak of a natural law in physics or in chemistry. The student of linguistic phe- nomena should always take into account the individuality of the language-user. 2. Whatever cannot be explained by regular processes of phonetic law must, in the main, be due to the influence of analogy. Most, if not all, apparently irregular and exceptional forms which cannot be brought under any known phonetic law, or which seem to violate such laws, have been formed directly after the model of other forms without etymological consciousness, simply by the power of association. These two forces, viz. phonetic variation and formation by analogy, are regarded as the most potent in bringing about indi- vidual instances of linguistic changes. Thus Sievers, in his article on Philology in the ninth edition of the Encyclopaedia Britannica, speaking of these two forces, says : " They generally work in turns and often in opposition to each other, the former frequently tending to the differentiation of earlier unities, and the latter to the abolition of earlier differences, especially to the restoration of conformity disturbed by phonetic change. Phonetic change affects exclusively the pronunciation of a language by substitut- 1 See Misteli, Lautgesetz und Analogic in Lazarus' und Steinthal's Zeit- schrift fur Volkerpsychologie und Sprachwissenschaft, XI 365-475. 2 Cf., however, BAL 98 2, A. J. P., V 171. ing one sound or sound-group for another. Analogical change is confined to the formation and inflection of single words or groups of words, and often has the appearance of being arbitrary and irregular. It is beyond our power to ascertain whence analogical changes may start, and to what extent they may be carried through when once begun. All we can do is carefully to classify the single cases that come under our observation, and in this way to investi- gate where such changes are especially apt to take place, and what is their general direction." Starting with these general premises, it is the purpose of this article to study the operation of analogy in the Semitic languages, and to present, in a systematic way, the results of this study. In a study of this kind we are confronted by three questions : i. What is the relation of analogy to the characteristics and structure of the languages in which it occurs ? Are its nature, its manifestations, and the scope of its application at all modified or conditioned by the well-known peculiarities of these languages ? 2. What individual instances of analogical formations are found in these languages? 3. How are they to be classified ? In conducting our investigations we may start from the well- known fact that the whole structure of language, in its grammati- cal forms and categories, is, in a sense, analogical. It exhibits the operation of what we may call constructive analogy. The working of analogy as a constructive force in building up groups of similar words and forms may be stated as follows : It is obvious that different stems, forming different words and present- ing different sounds and combinations of sounds, are used to express different ideas ; and further, that different modifications of the same stem, whether by internal vowel changes, or by the addition of pre- fixes, infixes and suffixes, express the same idea under different mod i- fications and relations. But that different words should undergo the same changes and receive the same additions in the form of prefixes or suffixes in order to express the same relation or modi- fication of the original idea, is clearly the result of analogy whereby words are grouped together in classes, and within these groups the change which is applied to one is applied to all. This forma- tion of groups or classes of words and inflectional forms, and the application of the same inflectional change (using this term in its widest sense) to every word belonging to the same group, are the result of analogy. Each group is governed by a prevailing 5 analogy, and each individual of the group is treated in its develop- ments and its modifications to express different relations, in accord- ance with this prevailing analogy. The Semitic languages are peculiar in exhibiting with great clearness and fullness the effects of this constructive analogy by the regularity and uniformity of their structure. So, for example, in the inflection of the verb we find that the general analogy which is normally exhibited in the stems with strong and firm consonants holds good for all verbs, and the deviations from this model of the strong or regular verb are only modifications owing to the peculiar nature and feebleness of certain consonants. From the simple form of the primitives, called the Qal or first form, are formed according to an unvary- ing analogy in all verbs the verbal derivatives, sometimes called forms, or stems, or conjugations, each distinguished by a specific change or added element, with a corresponding definite change in its signification, such as intensive, causative, etc. In other lan- guages where such formations exist they are usually regarded as new derivative verbs. But in the Semitic languages they are incomparably more regular than in the Indo-European lan- guages. 1 In these cases we have no reason to suppose that the present uniformity had to contend with original diversity. It may have been so, but the presumption is that it was not so.. But the case is different when we consider another marked uniformity in the structure of these languages, viz. the fact that all inflectional stems have, or are assumed to have, three stem-consonants. As the languages have comedown to us, we find a striking uniformity of appearance, but we have reason to suspect that it is at the expense of original divergency. In this case we have an instance of analogy partly as a disturbing and partly as a constructive influence. There are indications that the number of tri-conso- nantal stems was originally much smaller than at present, but in the course of linguistic development bi-consonantal stems were made tri-consonantal by the addition of another consonant until finally the latter formed the majority. 2 And although we have reason to suppose that the inflection of bi-consonantal stems was originally to some extent peculiar and different from the inflection 'See Kautzsch-Gesenius, Hebrew Grammar, 25th ed., Leipzig, 1889, §$39, 2 and 41. 2 Cf. Lagarde, Symmicta, I 122 (Gottingen, 1877); Deutsche Schriften (1886), 285 ; Bildung der Nomina (1889), 215. 6 of tri-consonantal stems, yet the preponderance of tri-consonantal stems was so strong that even those bi-consonantal stems which remained for the most part gave up their individuality, in various ways assuming the appearance of tri-consonantal stems and con- forming to their method of inflection. 1 It may be assumed, then, as an established fact that the present uniformity in the appearance of the languages, namely, the pre- dominance of tri-consonantal stems, is at the expense of original diversity. Still, even here, after the uniformity had once been established, analogy works as a constructive force in the further inflection of these stems. But leaving for the present the consideration of analogy and the analogical structure of the Semitic languages in this sense of the term, let us examine the subject of analogy in its narrower, more specific sense, in the sense in which the word is generally used by recent writers, such as Misteli and Sievers ; let us see what instances of such analogical formations we have in these languages, how they may be most conveniently classified, and how they are related to the fundamental structure and characteristics of these languages. Whatever theory we may adopt as to the original form and constitution of the (so-called) weak verbs, this much at the least is certain, that in their present form they present the appearance of verbs regularly inflected after the model of the strong or perfect verb, modified, however, by the peculiarities of the weak conso- nants found in the stem. Add to this the fact that in some of the Semitic languages certain consonants (e. g. in Hebrew the gut- turals) have certain peculiarities which give rise to corresponding peculiarities of inflection of the stems containing such consonants, and all the apparent irregularities of Semitic verb-inflection are accounted for. These different peculiarities give rise to different classes of verb-inflection, according to the ordinary denomination, verbs '"a, K"3, n"S, etc. But knowing something of the nature of these weak conso- nants, something of the nature of the differences which distin- guish these different classes of stems in their various formations and inflections on the one hand, and something of the nature of analogy as it is commonly understood, and as it is exhibited in 1 Compare Whitney, Language and the Study of Language, p. 302 sq. ; Stade, Lehrbuch der Hebraischen Grammatik, §i2a, I and §142-144 ; Kautzsch, Grammatik des Biblisch-Aramaischen, §§45 and 46 ; Noldeke, Syrische Gram- matik, §§41 and 57. other families of languages on the other hand, we are led to expect the occurrence of analogy just here. We are led to expect that the peculiarities which distinguish one class of weak verbs from another, the barriers, as we may call them, which separate the different classes from each other, should be entirely or par- tially disregarded and the different forms confused. And such we find to be the case. We have not only the general fact that all the inflection of those weak verbs which were originally bi-con- sonantal in their stems is analogical, i. e. based on the analogy of the stems with three stem-consonants, but we have also a great many specific cases of analogy. We find many individual instances of verbs of one class treated as if belonging to another class, and hence inflected after the analogy of that class, or even disregard- ing the weakness or peculiarity entirely and inflected directly after the analogy of the strong verb. We find also a few instances where the strong verb is inflected after the analogy of the weak. All such cases of analogical formation, therefore, which affect the real or assumed stem of any word comprise one class with three subdivisions. Class I. Analogical formations in the inflection of the verb or in the forma- tion of verbal derivatives with reference to the different classes of stems. Under this head we have three subdivisions : A. Confusion of the different classes of weak stems. B. Weak stems after the analogy of the strong. C. Strong stems after the analogy of the weak. This law of tri-consonantality , if we may so call it, in the stems of words, is one of the most prominent characteristics of the Semitic languages, and this class of analogical formations which has just been considered is closely connected with this same char- acteristic, in that stems which in their original form did not have three stem-consonants are conformed to the analogy of regular, original, tri-consonantal stems. Another peculiarity of the Semitic languages is the function of the vowel and the use made of differences and changes in vocali- zation to differentiate various classes of words and to construct different inflectional forms. Thus, in Arabic, qatala is the type of the active transitive verb, qatila of the intransitive, and qutila of the passive. 1 'Cf. Lagarde, Bildung der Nomina, p. 7 (ZDMG XLIV 536). 8 We have also many phonetic processes whereby vowels are changed, lengthened, shortened, etc., in the various processes of inflection. These changes and variations are so light and delicate that we may expect some confusion at times, and such we find to be the case. Sometimes these phonetic processes are firmly maintained, enabling us to ascertain the law which governs them. But in a great many cases forms are transferred from one class to another, and changes take place which are not warranted by any phonetic law. All these analogical formations connected with the vocalization of the different words and forms can be comprised in one class. This gives us Class II. Analogical formations involving changes and confusion in vocali- zation. Still another peculiarity of the Semitic languages is their method of inflection and of the formation of derivative words by pre- formatives, informatives, and afformatives, and the resulting dis- tinction between servile and stem-consonant. These formations and distinctions are peculiarly subject to confusion, and" hence give rise to many analogical formations which may be comprised in one class with four divisions. Class III. Analogical changes in the formative elements of words. A. Mistaking servile or formative element for part of the stem. B. Mistaking a stem-consonant for a servile. C. Analogical changes in the formative elements themselves ; influence of one formative element upon another. D. Addition of serviles and formative elements where they do not belong. I have thus shown, in a preliminary and provisional way, the possibility of the existence of three different classes of analogical formations affecting respectively the stem-consonants, the vowels, and the formative elements of the different words and forms. But when we consider that every Semitic word can be analyzed into these three elements, viz. its consonantal stem, its vowels, and its formatives, and that these three elements are in the main so strongly marked and so clearly separated, we see at once that this analysis has furnished us a basis for the classification of ana- logical formations which, though perhaps not so profoundly philo- sophical as some other modes of classification which might be adopted, still has the merit that it is simple, practically convenient, and exhaustive, and most of all, that it presents these analogical phenomena in closest connection with the structure and the char- acteristics of these languages. The results of our study up to this point may be summed up in the following propositions : i. The whole structure of the Semitic languages and all the formations and inflections of words are analogical, using the term analogy in its widest sense. 2. Using the term analogy in its restricted and more usual sense when speaking of it as a linguistic phenomenon, those cases of analogical formation which do occur are connected most closely with the structure and various characteristics of these languages, and thus they are easily accounted for, in fact they occur just where this structure and these characteristics lead us to expect them. We find, then, these two factors in the structure of the Semitic languages : on the one hand a notable degree of regularity in the recurrence of certain fixed types of forms and in the application of inflectional modifications ; on the other hand, certain deviations from these types and normal processes under the influence of a disturbing analogy. But when we look more closely into the nature and relations of these two factors it becomes evident at once that they sustain a most intimate relation to each other. They are not the result of separate and distinct linguistic forces, or to go back of the language to the mind of the language-user, of separate and distinct faculties of the mind. They are the result of the same law working under different conditions. As a lin- guistic law we call it the law of the group. As the product of the human mind we ascribe it to the power of association. By the constitution of the human mind each word is felt to be a member of a distinct class or group, and not simply an isolated individual ; and the whole philosophy of analogy as a linguistic phenomenon may be thus briefly expressed : Whenever the law or type of any group has been able to main- tain itself, the individual conforms to the law of the group to which it belongs and the formations are normal. But when the law breaks down and the distinction between the groups is disregarded, the individual, instead of conforming to the group to which it belongs, is transferred to another group, with which somehow it has become associated, and the result is a disturbing analogy. IO It is true that the unwarranted application or extension of analogy beyond its legitimate bounds is a marked feature of lan- guage. It is this that has given rise to the expression mistaken or false analogy. It might better be called disturbing analogy. One of its most frequent effects, as is pointed out in the words of Sievers, quoted above, is the obliteration of existing differences. But it is wrong to regard analogy simply as a disturbing influence. In fact, we might almost say that this is only an incidental effect. It is better to regard analogy as that which binds together the different individual members of each group of words or linguistic elements. Or to speak more accurately, it is natural for the mind to associate words which although different still are felt to be members of one and the same group, and to apply to all the inflectional changes which it has been taught to apply to one. This extension of the same inflectional changes to all the different members of the same group, this formation of different words from different stems after the same type or model, is not the work of memory, as is also pointed out by Brugmann in his book " Zum heutigen Stand der Sprachwissenschaft," p. 79. It is the work of the power of association or combination, or, as it might also be called, the analogical faculty. We are not concerned primarily with the name of this faculty, however. But that with which we are concerned is the fact that so-called analogical formations are not to be regarded as something isolated and entirely distinct from those formations which are called normal, but rather they are the results of the same mental process, and show the operation of the same law working under different conditions. Starting from the principle that analogy is the modification of an existing form, or the origination of a new form after the model of another form with which it is associated, different schemes for the classification of such analogical phenomena have been pro- posed. These different schemes are presented and reviewed by Delbriick, " Einleitung in das Sprachstudium," p. 108 fg. Accord- ing to him these classifications are made from three different points of view : 1. According to the nature of the psychological activities which are concerned in the different formations. 2. According to the nature of the words affected by analogy. 3. According to the results of the operation of this force of analogy. The first of these principles is the one adopted by Misteli in II the article already referred to (Zeitschrift fiir Volkerpsychologie und Sprachwissenschaft, XI 365-475, XII 1-26), and after him by Wheeler, in his treatise on "Analogy and the scope of its appli- cation in language." ' As this is by far the most satisfactory of existing methods of classification, I shall present it in some detail, using mainly the phraseology of Wheeler in the statement of the different classes, and illustrating them by examples taken from the domain of Semitic. According to Sievers, the influence of analogy tends largely to the abolition of earlier differences. Similarly Wheeler (op. cit. P- 35) : " The operation of analogy in language is in every case ultimately conditioned and determined by the natural quest of the mind for unity to replace multiplicity, system to replace anoma- lous diversity, and groups to replace monads. The office of the psychical factor in the development of language is therefore the maintenance and re-establishment of the groups which the phonetic laws tend to disrupt, and the creation of new groups. It aims to eliminate purposeless variety." To this must be added the important statement on page 37 of the same work : " The formation of new groups very commonly appears as a readjustment of old groups. Changes in the char- acter and use of the phonetic material of a language often cause a form or number of forms to be severed from one group and attached to another." Wheeler thus classifies the phenomena of analogy : I. Likeness of Signification and Diversity of Form. Two words entirely diverse in form, but which are capable of application to one and the same object or idea, may, through the influence of this limited likeness of signification, be confused into one word by the process known as " contamination of form." No instance of this kind in Semitic has occurred to me. II. Affinity of Signification and Diversity of Form. Words totally dissimilar in form, but expressing ideas of like category, are made to approximate slightly in form through the extended appli- cation of some sign of category or through the extended use of some element or combination of elements of sound which has come to be recognized as characteristic of a group. As an illustration of this kind of analogy the following has occurred to me : The common people often say masoner for mason. The ending er came to be regarded as expressing trade or calling, 1 Ithaca, N. Y., 1887. Cf. also A. J. P. V 165-85, X 202. 12 from such words as painter, joiner, carpe?iter, farmer, etc. Hence they sought to bring mason into the same category of form by adding er and making masoner. Instances in Semitic are the following among others : The prefix m, to form the passive participle of the Arabic first form, of which the ground-form is gat-til. See Barth, Nominalbildung, p. 178, In Arabic this principle often works in the extension of existing groups. See Barth, Nominalbildung, p. 135, §85^. The form qatil, formed nominally from the i- imperfect, is used to denote masses, collectives. Then in that sense it is formed from stems which have no *- imperfect, especially to denote masses or collec- tions of animals, kalib ' dogs,' da'in ' sheep,' etc. Cf. also the Mandean pronouns ac/mun, achtun, achtochun and achnochun (Nold., Mand. Gramm., p. 86). III. Likeness of Function and Diversity of Fo7'm. Words differing in form are reduced to groups upon the basis of likeness of function, i. e. of likeness of use in the economy of the sentence and for the expression of like modifications of thought. An illustration of this kind of analogy is found in the forma- tion of the Hebrew infinitive absolute of the Niphal. See Barth, op. cit. p. 72, §49$. The regular infinitive absolute of the Niphal is seen in ^DDJ. But as in the Piel and Hiphil a certain assonance was perceived between the imperfect and the infinitive absolute used to strengthen it, e. g. ^2pT\ ~»3^ Ex. 23, 24, a similar form was originated in order to produce a similar assonance in the case of the Niphal rrteft rosn, D^X rf?Si> I Sam. 27, 1, etc. This ana- logical formation almost entirely displaced the organic. Other illustrations of this are seen in the formation of prepo- sitions with suffixes in Hebrew and in Ethiopic. Compare ^$ The 1 is organic in the first two cases, because they are to be referred to the stems ^K and HU. Starting from such stems it is extended analogically to stems of other prepositions where it has no place at all. IV. Contrast of Signification and Partial Likeness of Form. Words of contrasted signification and of partly similar form are grouped in couplets, and a further approximation in the outward form is the result. 1 Cf., however, Beitr. z. Assyriologie I 160 (ZA, IV 375). 2 Cf. ZDMG, XLII part 3 (A. J. P. X 234) and Lagarde, Mittheil. II 231. 13 An instance of this kind of analogy is given by Praetorius, Ethiopic Grammar, p. 86, §99. Eth. wad 1 a, he has gone out, has in the subjunctive ida, and imperative da'. These forms ought to be inflected thus: tede'i, ide'd or del, de'-ti. They follow, however, the analogy of their opposite ibd\ from bo a, he has gone in, and hence we have the forms ted&z, iddii, 7'da a, etc. 1 Another illustration of the analogy of the opposites is the Hebrew flXfy last (stem prp) after the analogy of \W#r\, first. V. Likeness of Signification and Partial Likeness of Form, Words whose stems have a like signification and are similar but not like in form are grouped together upon the basis both of meaning and form, and a levelling of the form of the stems is the result. Under this head Wheeler brings the following classes of ana- logical formations: A. Levelling between different cases of like stems. As an instance, somewhat similar at least to the instances men- tioned by Wheeler under this head, maybe mentioned such cases as I have given under CI. Ill, Div. B., the Assyrian plural iidti for iddti, from ittu, which stands for idtu, 2 the feminine of idu, hand. B. Levelling between the different forms for person and number in the same tense of the verb. Analogical influence of this kind is very conspicuous in the inflection of the perfect of the Semitic verb. The original paradigm probably was as follows, in the singular : s qatal a. ' he has killed.' qatal at. ' she has killed.' qatal ta. ' thou hast killed.' qatal ti. ' thou (fern.) hast killed.' qatal ku. ' I have killed.' In Ethiopic, the first person qatal-M has influenced the second person masculine and feminine, so that they now have qatalkd, qatalki. In the other languages the reverse of this process took place ; kti of the first person was changed to tfi under the influ- ence of the analogy of the second person, while in Hebrew, by a further analogical change, id was changed to ti under the influence of the possessive suffix i. A 1 Cf. Hebraica II 6, i. 2 See Hebraica I 178, 5 ; cf., however, Delitzsch, Prolegomena 46 and 115. 3 See Noldeke, Die Endungen des Perfects, ZDMG, XXXVIII 407 sq. 4 Cf. SFG 53 below ; ZDMG, XXXVIII 419; XL1V 539, 1 ; Wright, Comp. Gramm. 175. 14 C. Levelling between the different parts (tenses), etc., of the same verb. An instance of this kind of analogical formation is seen in the vocalization of the Hebrew perfect and imperfect Piel as com- pared with the corresponding forms in Arabic. In Arabic, which probably comes nearest to the primitive, we have qattala, yuqat- tilu, while in Hebrew we have h$p, }tSp*. The ^-vowel of the second syllable of the perfect is probably due to the influence of the corresponding vowel of the imperfect. The z'-vowel of the first syllable is then perhaps due to vowel-harmony. D. Levelling between derivative and primitive. No instances of this kind in Semitic have occurred to me, per- haps because there generally exists such a close connection between derivative and primitive. For this classification Wheeler claims practical exhaustiveness. He says : " Under the five main categories which have been thus far established and discussed may be classified nearly if not quite all the phenomena usually associated with the action of analogy." I have preferred, however, in my treatment of analogy to look at the phenomena from a different point of view, and adopt a different method of classification. The predominant characteristic of all Semitic analogical formations, as I have already shown, is the formation of groups, and the disturbance of these groups by transferring individual words, forms, portions, or characteristic elements of forms from one group to another, taking them from a group where they exist organically and applying them to or placing them in a group where they do not belong. These dis- turbances or transferments will naturally affect either the conso- nantal ground-stem of a word, or the vocalization of the stem, or those formative elements, prefixes, infixes, suffixes, by which different inflectional forms or derivative words are formed. Hence the classification which I have given will logically follow, and I trust that a study of the material as I have arranged it, in the second part, under these different heads, will bear out its suitability. This material will appear in a subsequent number of this Journal. *5 PART II. The following are the principal instances of analogical forma- tion in the Semitic languages. Class I. Analogical formations in the inflection of the verb, or in the formation of verbal derivatives with reference to the different classes of verb-stems. In this class are comprised only such cases as have reference to the different classes of verb-stems. All other cases are placed under their proper heads (Classes II and III). Division A. Confusion of the different classes of stems, either in the inflec- tion of the verb or in the formation of verbal derivatives. In Hebrew the two classes of verbs r\'h and wh are frequently confounded. Sometimes, however, this confusion is due simply to the Massoretic pointing. r\"h after the analogy of K'6. K3^> 2 Kings 25. 29 'he changed' for T\W. So also from the same stem &y&) Lam. 4. 1 and K3£" Ecc. 8. 1. Kna* 'he is fruitful' Hos. 13. 15, st. ma. N?? 'to consume' Dan. 9. 24 for n?3 (inf. constr. Piel). NSnip Jer. 38. 4 'relaxing' for nSDD (prtcp. Piel). n«"|b for '"nb Zeph. 3. 1 'rebellious' (prtcp. Qal), st. mD. D^pp Lam. 4. 2 ' weighed ' as if from fe6o. D^Fi Deut. 28. 66 'hanging' pass, prtcpl. Qal as if from &6n. So also D-1K7Jj1 2 Sam. 21. 12 ^p. fr\?D*.- 2 Chron. 16. 12 'and he was sick' for ilWl. ■IKBnjI 2 Kings 17. 9 'and they covered' as if from astern NDn, for nsn. **?.*_! 2 Kings 17. 21 2^riD 'and he removed' as if from S13, for 1*J from mx i6 K.in-}K Prov. i. 10 'be not willing' as if from a stem NIK for ra«. KTV? 2 Chron. 26. 15 'to shoot' as if from a stem KV for m\ similarly 2 Sam. 11. 24 and Prov. 11. 25. K'6 verbs after the analogy of n'6 verbs. Several forms from K73 'to withhold,' 'to restrain,' are formed as if from a stem n?3, so *flK?3 Ps. 119. 101 (this, however, may be simply due to the Massorites), ^n?| 1 Sam. 25. 33, v3 1 Sam. 6. 10, nj»? Gen. 23. 6. So also many forms from NfH 'to heal' are formed as if from nfH, as na*| Ps. 60. 4. Imperative Qal with vowels as if from nys-|n Job 5. 18 'they heal.' nns"}3 Jer. 51. 9. -1ST 2 Kgs. 2. 22. Perfect and imperfect Niphal. »TJer. 8. 11. Piel. So also npia Prov. 12. 18 'babbler' from stem KD1 (cf. Ktp?P). npv Ruth 2. 9 from HP? 'to thirst.' •te Ezek. 28. 16 and rfe Job 8. 21 from «!?» ? 'to be full,' also ni*6p Jer. 25. 12, n^D Ex. 31. 5. nVttfe> Prov. 8. 13 from «3? 'to hate.' n«oq Ez. 33. 12 from «tpn 'to sin,' n'K3p_ 2 Sam. 21. 2 from fctii? 'to be jealous.'. These forms as compared with ^^P, etc., clearly show the influence of the analogy of ni?|. The example of nxttq, moreover, leads us to suspect that in KDin Is. 65. 20, KDn Ecc. 8. 12, and in other similar cases we have something more than a "confusion due to the Massorites." 1 '•HvStf Ps. 139. 14 and ^v^ Ex. 33. 16 are from &6d to sepa- rate, distinguish,' as if the stem were nba. Wpto) Job 18. 3 from Kptp < to De unclean' as if the stem were HDtD. So W¥P? 2 Sam. 3. 8 from K*p 'to find ' as if the stem were nm 1 It is well known that the verbs N"? are in the Mishna regularly inflected as verbs !Tv; see Geiger, Lehrb. zur Sprache der Mishna, p. 46; Strack- Siegfried, Lehrb. der neuhebr. Sprache (1884), §§o,8c and 105; cf. Wright, Koheleth (1883), p. 488. i7 n2m Jer. 49. 10 from K^n 'to hide' as if the stem were nnn, also H2nn T Kgs. 22. 25. So also O'GJ Jer. 26. 9 and tn«^3n Zech. 13. 4 from K33 'vaticinatus est' as if the stem were H2J (for 123). Similarly ffl23nn 1 Sam. 10. 13 and n*33"ljn 1 Sam. 10. 6. Again, a similar confusion is observed between V"v stems and Y'V stems. The following forms from V"V stems are formed as if from a stem Y'V : |-VV Prov. 29. 6 'he will sing' (st. pi). p»Y Is. 42. 4 ' he shall be bruised' (stem f n). [Cf. Eccl. 12. 6.] W* Ps. 91. 6 'he shall waste ' (st. YiB>). 1-13 Ecc. 9. 1 ' to explore ' (st. "112). WM Nah. 1. 12 'they shall be cut down' (st. ?n). TT2fi Is. 24. 3 'she shall be plundered' (st. TT2). p12Fi Is. 24. 3 'she shall be emptied' (st. Pp2). ptw; Ps. 115. 7 'they handle' (st. WB). VPQ Jud. 3. 24 'covering' (st. TpD). iWn Hab. 3. 9 'she was made bare' (st. ir\v). ftV. Hab. 2. 17 'it terrifies them' (st. nnn). VTXi Prov. 11. 15 'he shall be broken' (st. VV\). So also in Syriac these two classes of verb-stems, viz. V"V and Y'V are frequently confused. See Noldeke's Syriac Grammar, §§ 58, 105, 126B, 174E, 175, 176, 177A for these and similar instances in Syriac. The following forms from Y'V stems are formed as if their stem were V"V. 2'^ Josh. 2. 16, inf. of 2-1^ 'to return'; i»3 Jer. 48. 11, Pause, 'he is changed,' st. "ViD; and "IP} Ezek. 48. 14, Hiph. from the same stem. In Syriac all the verbs V"V form their forms with prefixes after the analogy of verbs |"S. See Noldeke, Syriac Grammar, §i78C. This is also the explanation of all those forms with prefixes from verbs V"V in Hebrew, in which the first stem-consonant is doubled. They are based on the analogy of verbs |"B. Instances are Dfy from WOI 'to be silent.' 1p\ stem Yip 'to bow the knee.' D^ stem D?on 'to finish.' ^ stem Wn ' to be slack, feeble.' F$\ stem nn^ ' to be low.' W. stem "|2D 'to tumble, fall.' 3D> stem 33D 'to turn' and 3D»_, Hiphil from the same stem. The same formation is observed in the case of some forms from Y'V stems. *®\ stem >1D 'to draw back, depart' s®\ stem ^jd 'to cut off.' ?*?Pl Hiphil stem 7-1T 'to shake.' JVp? Hiphil stem n-lD 'to move.' ^.J Hoph. stem b-1D 'to be long.' n^n and 0*3* 'to lay down,' 'to cause to rest' Hiph. stem rvo. According to Stade, §535b, the form fiJ.^Dfy impft. 3d person fern. plur. stem 33D 'to turn,' with the insertion of the diphthong ai (*v) between the stem and the ending, rests on the analogy of the same form from stems PI"? (for l »"7 or v6) n ^?^. In Mandean (Noldeke, Mandean Grammar, page 83) and in Amharic (Praetorius, Amharische Sprache, p. 141) confusion between stems V"V and *"? and )"? is especially frequent. The examples are all given in the places cited, so that it is needless to give them here. According to Praetorius (loc. cit.), this is also the explanation of the form H13D, i. e. the stem 33D is treated as if it were H3D for 13 D (sabawa). On rilUp see also Wright, Arab. Gramm. §120 rem. c. and Spitta, Dialect of Cairo, §101, 3. Sometimes stems V'B and *"S are confused. As is well known, many of the stems *"S in Hebrew are originally V'B, the original initial waw reappearing in the Niphal, Hiphil, and in verbal deri- vatives with D-preformative. Not all such formations with waw are organic, however. Thus, as is pointed out by Haupt (S. F. G. p. 22, note 1) the verb in* 'to know' is a verb cum yddh origi- nario, as appears from the Assyrian idi, tidi, and Ethiopic 'aydea notum fecit. Hence the Hiphil ^H" 1 ", Syriac 'awda* and skawda' are analogical formations. Some forms from stems V"V show the influence of *"2 analogy to which they bear an external resemblance. So VW. Job 18. 7, impft. stem "m 'to be narrow.' The verb ^n n (Nold. Neusyr. Sprache, p. 231, §109). 2ND 'to be worth' has some forms from fcOD 0'6). 20 p&O 'to spit' has in the inf. KplKn after the analogy of K"2 (Nold. §110, p. 233). V"h and w? are confused (p. 239, §111). See also p. 248, §114. The verb 3.T (p. 254, §117) forms its conjugation from several different though correlated stems. There are many instances of such confusion of the different classes of weak verbs in Mandean. The instances are too nume- rous to be mentioned here in detail. See Nold. Mandean Gram- mar, p. 24, §22. Verbs tertiae V after the analogy of verbs tertiae \ P. 82, §74. Verbs Y'V after the analogy of V"V and conversely V"V after the analogy of >"!?. Verbs mediae X and v after the analogy of 117. Verbs *>"h after the analogy of V"V, and so conversely verbs V"V after the analogy of verbs »"?. Nominal formations from different stems showing the influence of i"? stems are given on p. 104, §90. P. in, §94. (Mand.) forms from *'6, Y'h treated after the analogy of forms from stems V"V. P. 236, §i77a. Verbs tertiae gutturalis after the analogy of stems Y'7. P. 243, §180. Verbs K"3 are treated after the analogy of verbs 1"S, >"D Verbs Y'V and V"V are habitually confounded (Mand. Gram, p. 247, §183). So p. 255, §189. Verbs mediae V or K are treated after the analogy of verbs Y'V. Cases of the influence of one class of weak stems upon another in Amharic are given in Praetorius, Amharische Sprache (Halle, l8 79)> P- I 4 I > § io 3- As they are all fully cited there and dis- cussed at some length, it is needless to mention them here in detail. Other cases of the confusion of different weak stems in Arabic are given by Barth, Nominalbildung, p. 45, §3ob. The adjective-form qdtil from stems Y'V, preserves the w in a few cases such as tawil ' long/ hawzd 'zealous.' But in most cases they follow the analogy of stems v, y, with the usual change of djS to ajji {aiii'). Examples are maiiit 'dead,' 'aiiid, jaiiid 'good,' haiiin 'light, easy.' Other instances are given p. 188, §i27c, qaiiim, haiiit, saiiid. 21 Class I. Division B. Stems with weak stem-consonants after the analogy of stems with strong stem-consonants. Also original bi-consonantal stems after the analogy of tri-consonantal stems. As was pointed out in the first part of this* article, many stems in the Semitic languages which seem to have three stem-conso- nants were probably bi-consonantal in their original condition, and assumed an additional stem-consonant in order to make them tri-consonantal by the force of the prevailing analogy. Other stems, instead of becoming completely tri-consonantal remained more or less imperfect and preserved some traces of the original bi-consonantal condition. In different languages also they attained to different stages of completeness. Thus, from the original bi- consonantal stem Dp we have in Hebrew Dip and Qp T (with a long vowel), and in Arabic qdma. The Hebrew intensive is B£1p, and the Arabic qauuama, and Aramaic D'[2 ( n £*i?? Dan. 6. 8). Here it will be seen that the original bi-consonantal stem is more con- spicuous in Hebrew, and the form of the tri-consonantal stem is more completely attained in Arabic and in Aramaic, qauuama standing on exactly the same plane as qattala. Finally, taking the inflection as it now stands, where the pecu- liarity or feebleness of one or more of the stem-consonants, or the originally bi-consonantal nature of the stem gives rise to different classes of verbs, each with its own special paradigm (according to the current denomination V 'S, |"3, ff'V, T\"h» etc.), we find many individual cases which, disregarding these peculiarities, leave the paradigm to which they ought normally to conform and follow the analogy of the strong or perfect verb throughout. We find, then, that we can distinguish three closely related cases : (i). Stems in which all traces of the original bi-consonantal nature have disappeared, and which have three stem-consonants in all of their forms. (2). Stems in which the analogy of the tri-consonantal stems is established as part of the regular paradigm in some of the forms. Here we find such cases as the Arabic qauuama from qdma, Aramaic HOJidJ from E£. Compare also Noldeke, Modern Syriac Gramm. §42, where it 22 is observed that those nouns which in the earlier language were bi-consonantal, such as dim ' blood ' and shim ' name,' have become tri-consonantal in modern Syriac, dimmd and shimmd. Under this head may also be placed the forms treated by Barth, Vergleichende Studien, Z. D. M. G., Vol. 41, p. 603 fg. Because of the reluctance of the Semitic languages to have nouns with only two stem-consonants, masculine nouns of this class of forms of which he is treating (viz. shortened derivatives of stems 1"? and *"?) are but seldom found in the northern Semitic languages, and in Arabic hardly ever. In consequence of the prevalence of the law requiring three stem-consonants, the greater number of such shortened nouns assumed the feminine ending, not to denote sex, but simply to compensate the loss sustained by the disappearance of the third stem-consonant. Cf. Hebrew rigG? 'drink' from the stem np& y njj 'body.' In Arabic hize from the stem haziia * to obtain ' and many others. So also the "feminine ending" is added in those cases where the first stem-consonant has disappeared. T\j% and Arabic lidatu n from the stem T?1 (wa/add), It 'to bear.' This "femi- nine ending" has become so entirely a part of the stem as to remain in the plural. Compare ninjH. < doors,' Hfn^P ' bows,' HIJW 'watering-troughs,' ninp? Ez. 13. 18 'bands, pillows.' (3). Individual cases of single forms where the normal para- digm or type of the class to which the stem belongs is disre- garded and the form is based directly on the analogy of the strong or regular verb. In ^?_^ Ex. 23. 22 'I was an enemy' and in &1.W 'weary,' we evidently have a later development of a consonantal jjW/£ between the two original consonants of the stem. 1 With regard to *\?P Ti however, we must note that Barth, Nomi- nalbildung, §iob, regards it as transposed for *|ffj. The assimilation of a vowelless | is a familiar fact in Hebrew morphology. In cases where it is retained, the analogy of other consonants not subject to such assimilation doubtless has its influence. So «Wjn Is. 58. 3 'ye exact,' TKJ? Jer. 3. 5 'He will keep,' Vl$3? Deut. 33. 9 ' They will guard,' 3p3? Job 40. 24 from 2pJ ' to bore through,' •1Spf Is. 29. 1 from ?)p3 ' to come round,' said of feasts. 1 Cf. Haupt, Zeitschrift fur Assyriologie II 276, and Beitrage zur Assyrio- logie I 296 below. 23 T™Jl) Ezek. 22. 20 'to melt,' Hiph. inf. of ^ru. piipn Jud. 20. 31 Hophal from P^n 'to cut off.' In Assyrian n is not assimilated to the following dentals : enzu 'goat,' bintu 'daughter,' erisu 'weak.' See Haupt in Hebraica I 227. In ™ Ps. 119. 61 'to surround' Piel of TO; "HR 'to make blind,' *$£?& Ps. 119. 28 'preserve me alive !' Piel of D-ip; B#??n] Dan. 1. 10 'so ye should endanger,' Piel of the stem 2-in; to"!*B?n Josh. 9. 12 'we provided ourselves with food,' Hithpael denominative from TV 'food,' from the stem 1-1S 'to hunt,' we have instances of forms of stems with weak stem-consonants treated after the analogy of stems with strong consonants. In a few cases we find, instead of the usual form of the con- struct infinitive from n"b stems (J"l1?}), a form which is based on the analogy of the same form from the strong stem ?ttp ; such forms are nfettj 'to do,' HS*] 'to see,' n'^p 'to get,' ^ 'to drink.' Such forms as Dnj?| 'boves,' D*niq 'thickets,' D^jq 'forces,' D^jji 'bucks,' a> 1$ c cities,' when compared with the more organic form of such stems as seen in D*t?1£> 'whips,' CV'P 'thorns,' D?S 'terebinth-trees,' show the influence of the analogy of the form In the different forms of the verba mediae geminatae (or WV verbs) there are many forms which seem to point to an original bi-consonantal stem, e. g. no, while others are formed as if there were three original firm and equally legitimate stem-consonants, 22D. The former are the more organic, more in accordance with Hebrew phonetic law ; the latter are analogical. Thus by the side of such forms as PI 'he has beaten small,' OH 'he is warm,' and ">P 'it is bitter,' we have other forms with the second stem-consonant repeated. TT3 'he has plundered,' Bpn 'he is warm,' T!£ 'he has measured,' 23D n. In stems beginning with a sibilant the n of this reflexive prefix is transposed so as to come after the sibilant instead of before it. Thus, from TOSP 'to keep,' 'to watch,' the imperfect reflexive is 1©5p? 'he will observe (for himself).' This mode of forming the reflexive from stems beginning with a sibilant by means of an infixed instead of prefixed / is found in nearly all the Semitic languages. In Arabic and Assyrian, however, this mode of forming the reflexive by means of infixed t {ta or tau) is not confined to stems beginning with a sibilant, but the reflexive particle is universally infixed. 1 In this case the strong proba- bility is that it was not so in the primitive form of the language, but that it started, as in Hebrew, with the stems beginning with a sibilant, and then the influence of these formations gradually spread until it included all stems. Only two other explanations are possible ; either the primitive Semitic form was indifferently prefixed or infixed t, or else it was universally infixed, and became prefixed in Hebrew, Aramaic, etc., except in the case of stems beginning with a sibilant. Praetorius, in his article "Athiopische Grammatik und Etymologie," in the first number of the Beitrage zur Assyriologie, edited by Delitzsch and Haupt (p. 25 and note), says that it is not unlikely that the reflexive t was originally infixed in Ethiopic. The proof of this which he cites, however, in the case of kaddna 'texit ' seems very weak. 1 In Assyrian the t is infixed not only in the reflexive derived from the Qal (iptdlix), but also in the reflexive derived from the intensive stem (uptallix). In the latter formation the t is prefixed in Arabic (V and VI), not infixed as in the eighth form. 26 Delitzsch, on the other hand (Assyrian Grammar, §83), remarks that the reflexive prefix was originally prefixed and not infixed, and that traces of this prefixing are still found in Assyrian. This, too, favors the position that the change from prefix to infix first took place in the case of stems beginning with a sibilant, to avoid cacophony, and was then, by the force of analogy, extended to all stems. A similar case of the extension of a phonetic change beyond its organic limits by the process of analogy is found in Amharic, in the linguistic phenomenon known as 'palatalization' 1 (German, Mouillirung), in which the pronunciation of a consonant is affected by a certain liquid modification occasioned sometimes by a following z, /, or ^, and sometimes also taking place without the influence of these sounds. In the cases where the palataliza- tion is occasioned by the sounds mentioned, certain modifications in vocalization are phonetically legitimate. In some cases, how- ever, these changes in vocalization take place where they are not organic, but analogical. There is no /, /, or ^-sound to occasion them, yet the language acts as though there were, simply because the phenomenon of palatalization is usually accompanied by these sounds. In the pronominal suffix of the 3d person plural, masculine or feminine, dtjdw, for instance, the liquid modification of the / is not occasioned by a following i,j> or £ sound. Yet the language acts as though there must be a latent ^-sound, and in some cases places it there ; so in the form bilaljiw 'when he told them.' So also it often happens that a stem which originally ended in w or j lost its final stem-consonant in the modern Amharic by a process of phonetic decay. The theme thus shortened is simply inflected as though it never had another stem-consonant, or else a trace of the vanished final j is left in the palatalization of the preceding stem-consonant. In some cases, however, where this palatalization has taken place without the influence of a final/ the language proceeds on the presumption that there must have existed such a final stem-consonant. Thus the verb manatja never existed in the form manataja, yet it forms its imperfect, not imantj, as it should do, but imanatj, as though there were a final stem-consonant j. See this whole matter fully treated and illus- trated in Praetorius, Amharische Sprache, p. 45, §23b, and pp. 237-240, §200. 1 Cf. Beitr. z. Assyr. I 257, n. 9. 2/ Class II. Analogical formations involving changes and confusion in vocalization. In Semitic we distinguish in the nominal formations between sub- stantives of internal formation (Stade : Nomina innerer Bildung ; Barth : Schlichte Nomina) and substantives of external formation (N. a'usserer Bildung). Substantives of external formation are formed either by prefixes or by suffixes. Substantives formed directly from the stem without consonantal additions are divided into different classes according to differ- ences in their vocalization. According to Barth, Nominalbildung, these different nominal formations in Semitic exhibit many instances of the analogical influence of one form or class of forms upon another in the matter of vocalization. The following is a brief and summary statement of the funda- mental positions of his book : The nominal formations of Semitic are formed either from the perfect or from the imperfect verb-stems. There are three classes of perfect verb-stems, distinguished by three characteristic vowels, d, z, it. Forms with a are transitive. Forms with / and u are intransitive. Each of these classes of perfect-stems gives rise to a corresponding class of nominal forma- tions likewise distinguished by the three characteristic vowels, a for the transitive, and i and u for the intransitive. There are also three classes of imperfect stems, distinguished likewise by the three characteristic vowels u and i for the transi- tive, a for the intransitive. Each of these three classes, again, gives rise to a corresponding class of nominal formations. In many of these nominal formations we observe that the char- acteristic vowel of the class to which it belongs has been correctly maintained. These call for no further attention. In other cases, however, we find that a change has taken place. That is to say, the verbal stem, either perfect or imperfect, as it now exists, has a different vowel from that of the corresponding nominal forma- tion. In cases such as these there are two possible explanations. Either, in the original form of the language the verbal stem existed in two forms with two different characteristic vowels, one of which subsequently ceased to be used, surviving, however, in the corresponding nominal formation ; or else the anomalous characteristic vowel of nominal formation is the result of analogy. 28 I have noted the following cases as cited by Barth : § 17/3. Intransitive nominal forms of the form qidiill, legitimate when formed from ^-perfects, but sometimes also found derived from /-perfect stems. Thus 'umudd (perf. z) 'full of youthful strength.' Sometimes intransitive adjectives of this form are formed from transitive stems. Thus sumull 'hard, firm,' kubunn 'avaricious.' P. 36, §24. Intransitive nominal forms of the form qidl derived from /-perfects. In Arabic : zuhd from zahida 'to be white, glossy.' husr 'loss' from hasira 'to lose.' 'ujb from 'ajiba 'to be astonished.' ruhb from rahiba 'to fear.' rugb 'avidity, voracity' from ragiba 'to wish, desire.' In Hebrew: I»N 'faithfulness,' from |P« (?) 'to be faithful.' "ipn 'want' from "ipn 'to want.' In Syriac: tfjri 'anger' from \Y\ 'to be angry.' KJBttJ from \$U 'to be strong, heavy.' K37?^ from IB& K3C»i 'laughter' from TO 'to laugh.' tfpfe from D^p 'to be whole.' K|?ttJ 'length' from pm Barth, p. 43, §28c. Instances of the form ^/z/ intransitive from ^-perfect stems. sarV ' rapid ' from saru'a 'to hasten.' 7#/zVhard' from jaluda 'to be hard, firm.' karim ' noble ' from karuma. kabir ' old ' from kabura. kabira, fut. #, 'to be advanced in age,' is, however, also given in the dictionaries. katir ' many ' from katura. l azim 'great, large, immense, huge' from 'azuma. In the Targums: TV 'dry,' -p£> 'quiet,' yen 'sleeping,' 765>n 'dark,' from stems having u in the perfect. The same phenomenon is observed in nominal derivatives from imperfect stems (see page 96, §64d,e). Thus the Hebrew infini- tive ?bp : is formed not only from ^-imperfect stems, where it is 2 9 organic, but also from intransitive verbs with ^-imperfect, e. g. P\'to sleep,' 3HK 'to love,' ^H 'to cease,' *W? 'to hate,' «T 'to fear.' These are clearly the result of analogy. In other cases, however, we must consider the possibility that in a previous period of the language the two forms with two different charac- teristic vowels existed side by side, the one having survived in the present form of the verb-stem, the other in the corresponding nominal formation. Thus in Hebrew imperfect a is the regular imperfect of stems having a guttural as second or third stem-consonant, while the corresponding nominal forms often point to an original u or i form. Such are ?J? 'to strike,' *\V] 'to be angry,' vfrf 'to send.' So also transitive qitl forms in Arabic from verbs mediae guttu- ralis which now have only imperfect a.fil'^oX'sihn 'charm,' rihle 'passage, journey.' Here it is more probable that the uniform ^-imperfects are themselves the result of a later modifica- tion under the influence of analogy. So also in the case of transitive nominal forms of the form qdttl or qiftl derived from the imperfect stems. The imperfect i itself has been mostly crowded out by the prevailing analogy of the ^-imperfects, while the original vowel has often been preserved in the corresponding nominal formation, especially the infinitive. (See Barth in ZDMG 43, 177-91.) The following instances are given. Infinitives with suffixes : onrjp (-129 'to sell'), D^pa, n;u£ Infinitives without suffixes: "lfe$2 Deut. 26. 12 'to tithe'; also the noun "VVp 'tenth'; YTtfb 'to help' 2 Sam. 18. 3 (Kethibh), Y?d2 'to receive a portion.' Sometimes even when the z-imperfect was preserved, and not crowded out by the prevailing analogy of the ^-imperfect, its nature was misunderstood and this misunderstanding led to further analogical formations. In many cases the language mistakenly assumed them to be Hiphil-imperfects, with which they completely coincided in form, and then by analogy produced other Hiphil forms to correspond with them. So especially in the case of stems y 'V. From DT>P1> ^> etc., were formed the Hiphil-forms V2D, wbD, vnnp, pyip, Trn, e tc. The same thing took place in case of the other stems. According to Barth (op. cit. p. 119, §78a) : Whenever a transi- 30 tive form qitl of any stem points to an original /-imperfect, and when this stem occurs in some languages in the Qal-form, but in others only in the Hiphil, the presumption is that this Hiphil is of secondary formation, based on a misconception of the original /-imperfect. Thus the common Semitic form T3T, Arabic dikr, Ass. zikru points to an original /-imperfect stem Jazkir; compare Aramaic "13*1? with a on account of 1. As, however, this /-imperfect has disappeared from the Qal in Hebrew, it is probable that it was retained as a Hiphil, and then the other Hiphil forms were formed as a later analogical develop- ment. Arab, liadana 'to take in the arms' has at present only the ^-imperfect. From this we have Hebrew |> % n 'arm, bosom,' and Arab, hudne 'embrace.' But the Hebrew ^¥0 and Arab. hid?i point to an original /-imperfect. From ^21 'reverse, contrary,' and npaq 'overthrow,' we infer the former existence of a form IIPO- as j^^/-imperfect. This is not found now in north-Semitic. But Syriac has Aphel in the same sense as Hebrew and Syriac Qal. Compare also Hebrew Tjsrin Job 30. 15, which presupposes the Hiphil. In the same manner the noun siqi (watering) and the imperfect iasqi explain the formation of the later Hiphil, flgpn and Aram. W£, resulting in crowding out the Qal which must have existed at a former period ; cf. ^\?P, Syriac KJVj?E> etc. So also the Hebrew V91 'libation,' form qatil, Deut. 32. 28, points to an original /-imperfect. Hence VP** Ps. 16. 4, *J©!1 Gen. 35. 14 are regarded by Barth as originally /-imperfects Qal. Being erroneously regarded as Hiphil-imperfects they furnished the basis for further analogical Hiphil-formations. Another case is **Di?*i- Being regarded as a Hiphil-imperfect it furnished the basis for further analogical Hiphil-formations, but H?nf?# DvnjpD, and ntPDjpp point to an original /-imperfect of the Qal. P. 116, §77a, 3. In Arabic the transitive form qitle is formed organically from /-imperfect verb-stems to express the mode or manner of the action expressed by the verb. After this meaning had become established for this form it was extended by the force of analogy to verbs of every class. P. 126, §82. Transitive nominal formations qiitfd, q at '&l formed from the imperfect. Where such formations exist from stems which at present do not have u as the characteristic vowel of the imperfect we must 3i again, as elsewhere, distinguish two cases ; either at an earlier period the w-imperfect existed side by side with an i- or «-imper- fect, and was gradually displaced, or else the ^-nominal forma- tions are the result of analogy. Instances of the former case are seen where the Arabic has only /-imperfect while Hebrew and Syriac have ^-imperfect ; so Arabic huluk ; ^?D! and jp&/#r. But many such cases are probably the effect of analogy. The following are instances of infinitives of this class where there existed no corresponding ^-imperfect : uutilb from uataba 'to leap.' iiurfid from uarada 'to arrive.' uujUd from uajada 'to find.' iiuj-tib from uajaba 'to be necessary.' uu' ill from ucCala 'to seek refuge in.' uuj&z from uajaza 'to be brief.' P- !35. §85d. The infinitive qatil is used to form collectives. Organically it belongs, of course, to z'-im perfects, but it is often found where there is no corresponding /-imperfect, especially to denote collectives of the names of animals. So kalib 'dogs,' da" in 'sheep,' mctiz 'goats,' naqid ' cattle.' e. In Ethiopic the infinitive qatil has become the regular verbal infinitive for every class of verbs, and as such is formed from all the different conjugations. This analogical extension of the infinitive form qatil has completely crowded out the old parallel form qatil. f. So also in later Hebrew (Mishna) the feminine n^ttp, used to denote abstract notions is formed without regard to the char- acter of the stem. The instances given are, nD^Dn, rip*TB# n?*31 r\w>, htdKp nnw. vofox\, n«»i. P. 144, §92. Transitive qattlat. In the case of the words nil$j and n ?!P we are unable to determine whether we have the evidence of a former /-imperfect of these stems, or analogical formations induced by the signification of the same form from other stems. P. 174, §I22C. Speaking of the active participial form qdtdl (or qattiV) he says that a genetical connection with the infinitive qiittil is beyond question, and both go back in their origin to the ^-imperfect. But just as it was shown (in paragraph 82 B of this same work) that the infinitive qiitHl was often formed analogically when there appears no w-imperfect, so also tire participial form qdtiil is used to form an active participle from a number of stems which show only the /-imperfect; so darUb (by the side of dariU) 32 'striking,' hattif 'buzzing, sounding, clanging,' gastim 'violent' (but gasama is given with w-imperfectj, kasllb 'gaining, ac- quiring.' P. I75,§i22a, note I. Instances of active participles of the form qatM formed from stems having /-perfects, through the influence of analogy. In Arabic : daliHk 'laughing,' 'amfil 'doing,' icCiis ' despairing.' In Hebrew: £?s: ^fe> 2 Sam. 5. 8 and 21 nn-in^ Hosea 3. 1, are cited as instances of this form. From the Mishna, pplDtfn 'busying themselves,' Para 4. 4, naiD3n rPT 'dripping olive,' Pea 7. 1 Tip^n 'be watchful,' Ab. 2. 14 »nB> 'dwelling,' etc. P. 178, §i23a. The form qdtdl as passive participle is formed in Hebrew from all stems indiscriminately. P. 186, §i26a. The participial form qdtil organically connected with /-imperfects is formed in Arabic from verbal stems of every kind. d. In Aramaic this form has become the regular passive parti- ciple, and as such is formed from every transitive verb. P. 201, §i36b. Participle qdtil. This form arises from the transitive /-imperfect. The form from the /-imperfect, rather than that formed from the z<:-imperfect, became the prevailing one through the influence of the forms of the participle in the derived conjugations, which show an /-vowel after the second stem -con- sonant. The only difference between the participle of the Qal and that of the derived conjugations lies in the fact that the latter prefix an m, while the former, not taking the prefix, lengthen the vowel of the first syllable. After this form was established as the participle of the transi- tive /-imperfect, it was extended to all transitive imperfect stems, and also to those stems which are intransitive in structure but transitive in meaning. In a few cases it is extended to verb-stems intransitive in meaning. Arabic, dmin 'sure,' sdlim 'safe,' dd'td 'far.' Eth. rdte 'true.' Aram. ;Vr[ 'fearing,' N^n 'rejoicing.' Heb. n?.?lT 'base, abject.' I have stated these principles and given these illustrations just as they are given by Barth. I must say, however, that to my mind his whole position and the value of all his discussions are 33 extremely doubtful. For, first, as he himself admits in many places, we do not know how far these forms, which apparently show a different vowel from that of the form from which he derives them under the influence of analogy, may go back to other forms which no longer exist, but which, when existing, had the same vowel as the nominal formation derived from them. The whole of §100, for instance, on the imperfect infinitives of the derived conjugation is made extremely uncertain by the admission which he rightly makes, that formerly these imperfects may have shown an u as well as an /-vowel. This but shows how uncertain the whole matter is, far too uncertain, in fact, to form the basis of scientific investigations and conclusions. But again, I do not see that he establishes with any degree of certainty the fundamental position of his whole book, viz. that all the nominal formations must be referred either to the perfect or imperfect verb-stems, whose characteristic vowel will normally be the characteristic vowel of the corresponding nominal formation, and that all deviations therefrom are the result of analogy. Why should all nominal formations necessarily come through the chan- nels of the perfect and imperfect verb-stems ? Why can they not be referred directly to the simple ground-stem, without the intervention of the perfect or imperfect verb-stems ? These are questions not satisfactorily answered by Barth, while they are suggested by the many deviations from the principle which he seeks to establish. 34 PART III. Besides such analogical formations as have just been presented and discussed, there are also many such formations in the inflec- tion of the substantives. In the various processes of inflection, such as the formation of the construct, emphatic, plural, and in the addition of pronominal suffixes, each class is governed by its own peculiar laws. In a great many cases the law of the group maintains itself and no confusion occurs. But from the nature of the case, considering the subtle nature of these formations and distinctions, we might expect occasional confusion and transitions from One class or type to another, in other words, metaplasms and analogical formations. And such, on examination, we find to be the case. Thus, according to Stade (§191), in Hebrew the nomi- nal forms gat I, git I, gut I form their plurals after the analogy of the forms gatal, gital, gutal. 1 On these plural formations last mentioned, however, we have an interesting dissertation by Mr. W. H. Salter Brooks, " Vestiges of the Broken Plural in Hebrew " (Dublin, 1883). The object of this paper is to show that the stems qatl, gitl, gutl formed their plurals originally without any plural termination whatever, but simply by internal vowel changes, thus : gatal, which became gatdl, qital or git&l, gutal or gutdl, exactly like the corresponding Arabic internal or broken plurals. According to this view, then, the plural of these words is sufficiently indicated by these internal vowel changes. But subsequently, owing to the tendency to vowel corruption and obscuration, these distinctions were not sharp enough and clear enough to distinguish between the singu- lar and plural. Thus at first the Hebrew would have presented the scheme : s. ba'l. pi. bcCal ' lord.' s. rumh. pi. rumah 'spear.' But owing to the presence of the guttural consonant, the singulars were bound to become bdal, rumah, and actually did so, that is, become identical with the plurals. Hence, in such cases, the 1 Cf. Haupt, Beitrage zur assyr. Lautlehre (Gottingen, 1883), P* %9> n - 3 5 Praetorius in Beitrage zur Assyriologie, Vol. I (Leipzig, 1890), pp. 374-7. 35 addition of the external sign of the plural was necessary in order to distinguish the singular from the plural. But after it was added to such forms, the language lost sight of the distinction between singular and plural sufficiently maintained by vowel changes, and gave to the external or sound plural formation universal exten- sion. The application of the external plural sign in these cases, there- fore, is not to be regarded as organic, but as the result of analogy. On this view of the matter we have here an instance of analogical formations Class III, D, the addition of formative elements where they do not belong. In Aramaic and Syriac substantives of the type qatl sometimes form their absolute and construct state after the analogy of the type qatil. The organic form reappears in the emphatic state. (See Haupt in Hebraica, Vol. I, p. 228, note i.) 1 Thus in Syriac we have the regular normal forms in Fffl (for t&rcT) Wf» ■' door.' ">3S (for pdgdr) K*tja < corpse.' But B*S3 'soul' stands for napis (form qatil). The emphatic K£^ is syncopated from ndpisd. So also *|H?' K?J?3 'shoulder,' and ^P T /J 'king,' etc. On the other hand, the ground-form of ?V2 'lord,' is bcCl (type qatl), construct bcCal y and from this we ought to have, by the laws of Syriac phonology, the form ^4, and the z-vowel of the second syllable is due to the analogy Of forms like &>&?, ^^\ in other words, a metaplasm from the type qatl to the type qatil. So also D-i> Kp-13 'stomach,' rh**whk 'image,' KVQ> xvyp 'reason,' etc. Again, all vowel changes in the stems of words are governed by regular phonetic laws. (See, e. g. the laws of Hebrew vowel changes in Stade, §73 fg.). But in individual cases analogy often plays an important part. In such cases an individual is trans- ferred from one group to another, or a phonetic process is taken out of its legitimate sphere and applied to cases where it is not legitimate. An example is given in Stade, §78b. A long ^-vowel of primitive Semitic generally appears in Hebrew as 6 (Stade, §77a). But an original w-vowel may also appear in Hebrew as 6. This 1D 'to be long,' D|?-in from the stem W 'to stand,' the short $-vowel of the final syllable is not organic, but due to the analogy of the strong verb. Compare also •irip-in ' they were killed,' stem HID, and ■IfcO-in 'they were brought in,' stem K12. Again, in the vowel changes which a word undergoes in receiv- ing pronominal suffix. 3 we find cases of the influence of analogy. Take, for example, the substantive of the form qatal (which in Hebrew appears as ?9R' " l ?7) with pronominal suffixes. The following is the law of vowel change as stated by Bickell (p. 77> §105; : " The vowels before the suffixes are changed entirely according to universal vocal laws. The pretonic syllable should be height- ened. The syllables at a greater distance from the tone should be volatilized, and open syllables which stand before a half-vowel should retain their short vowel. Only the suffix ka forms an exception to the latter rule> before which a is heightened in the antepenult." Thus from ~^% (type qatal) 'word,' we have ^l. 'my word,' Tiyi 'his word,' D?W 'your word,' but T£\ 'thy word.' This heightening of the vowel a before the suffix ^ as com- pared with the short vowel before B? is due to the analogy of the contiguous forms "H?^' ^191' ^9\ The distinction between 1 as light suffix and &?» 1? as grave suffixes is a mere fiction of the grammarians. For instances of the influence of analogy in the affixing of pro- nominal suffixes in Syriac see Noldeke, Syriac Grammar, §145 E. 37 Another case of unwarranted application of phonetic processes under the influence of analogy is given by Stade, §3550. The organic form of substantives derived from n"? stems with the suffix of the 2d pers. masc. is seen in T,? 'thy teacher' (stem nY), spjrjO 'thy army' (stem run), spjj?p 'thy flock' (stem n:p), T$? 'thy maker' (stem POT). In other cases an analogical shortening has taken place, based on the analogy of TJR in Pause and out of Pause X#\ So we have in Pause X& 'thy field,' but out of Pause T]?. So also itffto 'thy rod' (stem ntM) sjjjjp, sjg 'thy dwelling' (st. ™) T)?P- Gen. 48. 4 Hiph. Prtcpl. stem ma, ^-^P. Piel Prtcpl. stem nw 'to command.' So also in the forms of the verb with suffixes we find instances of the influence of analogy in the confusion of vocalization. When the suffix of the 2d person sing, ka is affixed to a verbal form ending in a consonant it ought to be hard. So quite organically in $£9? Jer. 22. 24 ' I will pluck thee ' (stem priO). But when this same suffix is attached to the 3d person, fern, sing, perfect (which never had a final vowel) it is spirated after the analogy of the same suffix attached to the masculine, where the spiration is organic because this form ended originally in a vowel. In other words, gdtdldt-kd was understood as standing for qdtd- Idtd-kd, because qdtdl-kd stood for original qdtdld-kd. On the other hand, forms with hard 1 have influenced analogically forms which should have been spirated. So 33.13 Deut. 24. 13 'he has blessed thee.' Of a similar nature is the application of the pausal form of the suffix with a hard 1 to substantives. So 3*3 Prov. 25. 16 'thy sufficiency.' 7\W?, Deut. 8. 5 Piel Prt. 'thy instructor.' $m gj» (stem ruiT) 'thy answerer.' 3i?$ 'to justify thee' Job 33. 32 (Piel Inf.). 3.??P? Deut. 23. 5 Piel Inf. 'to curse thee.' Examples of forms with spirated "I where it ought to have been hard : "W?^. 'she consumed thee.' VT^J 'she bore thee.' irtan Cant. 8. 5 'she brought thee forth.' A similar case is pointed out by Bickell, §113. The imperative had no final vowel even in primitive Semitic. ;V>p.=qtul=qutul. 38 "But," says Bickell, "the Hebrew forms with suffixes presup- pose a final i=u> as in the future." In other words, the form of the imperative with suffixes is based on the analogy of the imper- fect with suffixes. Examples are : •ini!?y 'serve him/ *aajg 'give him,' stem H"0. VH& ' bear him,' stem Kfa. •inr^T ' know him,' stem 1H\ •in|?na 'let him go.' wyp« 'strengthen him.' •in|5;tn 'strengthen him.' On the other hand, in Aramaic the organic form appears with- out any vowel, thus ^?V.f Dan. 2. 24 ' bring me in/ stem ??!>, aphel; cf. Noldeke in ZDMG 38, 408. In his Vergleichende Studien III (ZDMG 43, 181) Barth has given us a treatise on analogical changes observed in the vocali- zation of the imperative and imperfect of some verbs. The following is a brief summary of his view : The imperative is formed from the imperfect. Corresponding to the three classes of imperfects, therefore, distinguished by the three characteristic vowels u, a, and /, there must have been three imperatives, u, a, and /. The w-imperative becomes ?bp s ; the a- imperative is seen in 2H^ • the /-imperative is seen in |fi» 3G£ and 11 All the forms of the 2d pers. fern. sing, which have the charac- teristic /-vowel point to an original /-imperfect. In the imperfect the vowel has been mostly changed to u through the influence of other ^-imperfects. In the imperative, on the other hand, it has been maintained. This explains the forms HPtf 'speak,' H?P 'sell,' *"??f 'pass,' n?H 'gird,' ^W 'pour,' »BP« 'collect,' ^n 'uncover.' The /-vowel in these imperative forms, therefore, is an indica- tion of the original /-vowel in the corresponding imperfect forms, just as forms like *??p 'rule,' *t?p 'exult,' point to an original u- vowel. So also in the 2 pers. masc. sing. k}1$1 Prov. 4. 13 'preserve her,' and in the forms with n?, such as ni^ 'preserve,' iTJpp 'sell,' n ?l? 'arrange,' »iBpK 'collect,' we see an original /-vowel. Not so, however, in the 2 pers. masc. pi. Here nearly all the forms which originally had the characteristic vowel u now have the vowel /. The change has been made under the influence of the analogy of those forms where the /-vowel is original. 39 Only three cases remain in which the original zZ-vowel has been preserved, viz. •I3"in from ^in ' to oppress, destroy.' For instances of forms of the Assyrian verb where the force of analogy has resulted in changes in vocalization, see Haupt, " Der Keilinschriftliche Sintfluthbericht," in KAT. 2 68, note 2. 1 By the side of the older organic forms of the present, such as ibdlut 'he lives,' isdgum 'he cries,' izdnun 'it rains,' we have later analogical formations, such as iza?ian i isakan 'he places,' where the vowel of. the last syllable has been influenced by the vowel of the preced- ing syllable. Delitzsch, however (Assyrian Grammar, §9oa, note, also §94, and §115), maintains the contrary view, that izanan is the organic and izanitn is the analogical. 2 Finally, on comparing the vocalization of the Hebrew intensive perfect and imperfect with the corresponding forms in Arabic, it seems probable that the Hebrew vocalization has been changed under the influence of analogy. In Arabic qattala, yuqattilu is probably nearer the original in form, and more organic, while in Hebrew '®P. the vowel of the second syllable has been influenced by ?W\, the imperfect. Com- pare Barth, " Nominalbildung," p. xxii fg., also p. 70. On this point Dr. Martin Schultze (Zur Formenlehre des semitischen Verbums, Wien, 1886, p. 48, §49), remarks that the derived stems are undoubtedly younger than the simple stem or Qal. It is to be assumed that they first formed the younger tense (the so-called imperfect), whose vocalization is the same in all the different dialects. But at a subsequent time, when the need was felt to form the afformative tense (or perfect), the tribes had separated. The older northern Semites retained, without reflection, the vowel of the future; the younger Arabs, however, chose, more consistently, the vowel of the Qal-Praeterite. Class III. Analogical formations with reference to the formative elements of words. 1 Cf. Beitrage zur Assyriologie, Vol. I, p. 124. 2 Cf. Barth in ZA II 377; Zimmern, ZA V 10. 40 Division A. Mistaking a servile or formative element for part of the stem. An interesting case of this kind is described by Stade, Hebrew Grammar, §400-2. It is in the formation of the 3d person femi- nine singular perfect of the n"? verbs. The ground-form was qataiat (compare Arabic radiiaf). This could become either q&tat or qdtdj'd. The former has been preserved in r\WV Lev. 25. 21, stem nbw 'to make.' J"pn 2 Kings 9. 37 Kethib, stem n}n 'to be,' and with suffix • : t t The latter is preserved in TDPi Ps. 57. 2 '(my soul) has fled (for refuge).' But the form qdtdt was generally regarded as a regular perfect tri-consonantal stem ; the servile n was erroneously regarded as a stem-consonant. Then, taking the forms qdfla and the pause- form qdtdld as models, the feminine ending d was appended, giving the form q&tata. This form has been retained in the pause, niTO, nnjn, etc. From this, after the analogy of qdt e ld, we have the usual form r, fl?\). Another case of this kind is seen in the treatment of the Niphal of the tf"^-stems. The ground-form is n&katt. The organic form is seen in irjj, stem Tin 'to be dry.' 3M, stem 2!1D, Eccl. 12. 6, stem pn, 'to break.' Then, as from TJ3 W e have in the feminine n *^3, so we have the feminines ^iJK, stem Pp3, 'to empty.' n ?pJ and the plural ^TJ, stem ^T, 'to shake.' Sometimes a n, which really is the sign of the feminine, is treated as if it were a stem-consonant. The examples of this class of analogical formations found in Syriac are given by Noldeke, Syriac Grammar, §78 and 80. The same thing has taken place in Ethiopic. See Dillman's Grammar, §133, and §137, 5b. For the Assyrian examples of this class see Haupt, SFGM 73 ; J also Delitzsch, Assyrian Grammar, §69. 1 Cf., however, Haupt, Beitrage zur Assyriologie, Vol. I, p. 16. 4i mutt&ti, pi. of muttu. Midi, pi. oilHu 'night' (st. rt*6). isitdte, pi. oiisitu 'pillar.' 1 On the Hebrew nirta, nWj?, etc., consult Am. Journ. Phil. XII 37 ; also Barth, in ZDMG 41, 605-607, and Nominalbildung, p. 4b. In Syriac the formations IJ-' HU Krrt- from nouns in ^ril-, are organic only in derivatives from stems l"?, so ^O^, pl« H?}.?^ ' image.' In forms like ■13?D# Kfll^D 'kingdom,' the ending n-1, which is merely a formative, is treated after the analogy of those words where it forms part of the stem. It might be expected that languages, starting out, as do the ancient Semitic, with a sharp distinction between stem-consonant and servile or formative element, would in the course of time and a development extending over more than a thousand years, lose the sharpness of this distinction and incorporate elements which originally were formative as part of the stem. This expectation is borne out in the study of the Amharic, the Mandean, and the modern Syriac. See Dr. Martin Schultze, " Zur Formenlehre des semitischen Verbs," p. 43, §45. Thus in Praetorius, "Amharische Sprache," pp. 130-140, under the title " Wurzelerweiterung ," we have a study of those stems of the Amharic language which have been extended beyond the original three stem-consonants, mainly by incorporating and reckoning as part of the stem those consonants which originally were formative. In many cases, however, as is pointed out by Praetorius, the incorporation of the formative element is preceded by the loss of one or more of the original stem-consonants, so that the formative element is taken as a compensation for this loss, and the appear- ance of having three stem-consonants is restored. The other methods of extending the original stem, viz. by doubling the original bi-consonantal stem, and by other additions, or by a repetition of one of the original stem-consonants of the tri-consonantal stem, do not, of course, come up for consideration in this paper. We are concerned only with those cases men- tioned by Praetorius as constituting the third method of extending the stem, namely, by incorporating a nominal, pronominal or 1 Cf. Lotz, Tiglathpileser, pp. no, 56 and 155, 27. 42 verbal formative element, or even a preposition, as part of the stem. I add a list of the cases mentioned by Praetorius, referring to his work, however, for their fuller discussion. The causative prefix a is treated as a stem-consonant in the stems amara as if it were a simple stem primae gutturalis ; but in reality it has arisen from 'amara, the causative of a stem med. gtctt. So also 'addga 'to grow.' 'asaldma ' to be a Mohammedan.' A more doubtful case is 'ardsa ' to moisten.' The causative prefix sa or its modification sa is treated as a stem consonant in the following words : sanakala ' to give offence, to be a stumbling-block ' (Anstoss erregen). shanagala 'to deceive.' shamagala 'to become old.' shanakara 'to leak.' shamatata 'to despise, scorn' (perhaps). shamana 'to weave.' shaqaba 'to ascend.' This seems to have been developed from 'ashaqaba 'to lift up,' a second causative from aqaba. So in the " Quinqueliterum " shakarakara 'to rise,' and saba- raqa 'to shine, sparkle.' The following stems have been formed by raising the t of the reflexive prefix ia to the dignity of a stem-consonant : tafid 'to sleep,' from tanahaya 'to rest oneself, to recover.' tanafasa 'to breathe, to take breath.' tanakuala 'to be deceitful.' tdlala 'to be deceived.' talalafa 'to cross.' tawdba 'to be beautiful.' tarasa forms the new double reflexive tanlarasa 'to support one's head by anything.' tanabaya 'to prophesy.' Perhaps also tdkata 'to be slow, lazy.' Sometimes the reflexive t has been weakened to d, and is then treated as a stem-consonant. So in ddsasa, 'to feel of.' Perhaps also daraqa 'to be dry' belongs here, and danaquara ' to be deaf, senseless.' danagara 'to dissuade.' 43 Here also Praetorius mentions the Ethiopic danagada 'to frighten,' and danagala. Nominal formative elements are likewise incorporated as part of the stem ; so m in the words mdshara 'to be a bridegroom/ from Ethiopic moshar 'a saw.' mashaga 'to barricade oneself.' marakuaza ' to support.' masaqala 'to hang.' manazara 'to commit adultery.' manatsha ' to spring forth.' managaga 'to hold,' or 'to open another's mouth.' mdkuata 'to vow.' mdraka 'to take captive' (also in Ethiopic). mtisana 'to recommend' (also in Ethiopic). So the prefix 'a of internal plural formations is reckoned as part of the stem in 'amalaka 'to adore God.' 'asamd 'to prac- tise witchcraft,' formed from the plural 'asm&t 'witchcraft ' with the loss of the t So also the nominal suffix t in the words gazata 'to excommu- nicate,' from the Ethiopic stem wagaza. So perhaps from a noun with the suffix at or ot we have tan&- wata ' to offer opposition, to withstand.' galamola 'to play the whore,' from gdlamot 'whore.' So the consonant n of the abstract ending an in shalatana ' to rule.' m and / are treated as stem-consonants in malakata 'to denote,' from lak y a. mogata 'to dispute,' from wag'' a 'to push, to strike.' m&lada 'to get up early.' So the consonant m of the adjective ending dm is reckoned as part of the stem in such words as quaratama ' to have the rheu- matism' and taraqama 'to fill up, to stuff' (vollstopferi) and gara- mama ' to exercise a horse.' Other noteworthy cases are the following : zawatara l to continue, go on,' in which the relative za has become part of the stem. baratd 'to be firm, strong,' in which the preposition ba has become part of the stem. Praetorius further mentions similar formations in Ethiopic, such as la'dla 'to be high,' from Ja-hsV. Also bahata and bahtawa 'to be alone,' from ba + 'ahati. 44 For a very complicated case of this kind see Praetorius, Amharische Sprache, §2o8a, b, c, d. 1 In an early stage of the language the verbal infinitive of heda 'to go' united with the auxiliary verb 'a/a 'to be' and the union became fixed both in form and in meaning. At a later period, by a " false analogy" this form, the result of this union, was considered a normal homogeneous verbal stem, and received further modifi- cations, so that its origin was much obscured. The verbal inf. of heda, which at present, in accordance with the regular process of contraction in Amharic, is hido, was for- merly haydo. This older form hay do united with a following 'a/a, and formed, not haydwd/, which the later phonetic law would require, but haydo/a, in accordance with an older law still preva- lent in Tigrifia (see Praet. Tigrina Grammar, §48, p. 69), and which accordingly we must postulate for the older period of the Am- haric. hayddla then meant ' he has gone, he has left, he is not here.' But in course of time the original meaning of the verbal stem 'to go' was lost sight of, and the negative element, from being merely secondary, came to be regarded as the primary sense ; and so the initial hay was regarded as a sort of negative particle. Then, in accordance with the analogy of the initial syllable of the negative form of the imperfect, hay became 'ay, and finally the negative m was attached, 'aydo/am 'he is not.' Since, then, the first part of this form, viz. 'aydo, had lost its original character as a verbal infinitive, it had to lose at the same time the ability to make its suffix correspond to its subject, and so it became fixed in the form of the 3d pers. masculine singular. The form 'aydo/am had the appearance of a negative imper- fect of a stem do/a, only the final a gave it the character of a per- fect form. And since the endings of the perfect were assumed for all the different persons and numbers, the form of the 3d pers. pi. 'aydd/ilm, although having exactly the appearance of a nega- tive imperfect, was not able to transfer the whole inflection to the imperfect, even though such a transfer was favored by the change of the initial hay to 'ay. Also the circumstance that the form with the relative is ydydo/a 'who is not,' and not yamdy do/a, shows clearly that the form was regarded as a perfect. In this way it 1 See also Praetorius' Beitrage zur athiopischen Grammatik und Etymologie, in Vol. I of Beitrage zur Assyriologie, especially §§1, 2, 6, 52. 45 came to pass that this old compound verbal infinitive was inflected in a mongrel way, in its initial part having the appearance of a negative 3d pers. imperii., and in its final part having the appear- ance of a perfect. These mongrel forms again have resulted in further analogical formations. Being regarded as pure imperfects 3d pers. masc. sing, they received the usual pronominal object suffixes, meaning : 1 it does not belong to us,' or ' you or thee it is not appointed to,' etc. Then from this the reflexive taddla ' to be appointed, assigned,' was formed, and even the simple ddla imposuit seems to have been used, though Praetorius says he never found it in any text. Similar cases are found in Tigrina. See Praetorius, Tigrina Sprache, §173. The reflexive prefix ta is regarded as part of the stem, and from the stem thus augmented a new causative is formed by pre- fixing 'a. See the examples given in Praetorius, Tigrina Sprache, §173. The Mandean instances of this kind of analogical formations are given in Noldeke's Grammar, p. 84, §74, p. 98, §87. The Arabic form asnata ) denominative from sane, sdnat, 'evil year,' in which the feminine ending / is regarded as part of the stem. In modern Syriac many verbs with \ as a fourth stem-consonant are denominatives from nouns which have been formed by adding the suffix an to a tri-consonantal stem. The instances are given in Nold., Neusyrische Sprache, p. 192, §95. In other quadri- literals the first stem-consonant was originally a formative element which afterwards came to be treated as part of the stem. Such are many beginning with B> and D, such as *6nD 'to change,' and "OHD 'to visit,' etc.; others beginning with n, such as nmn 'to be boiled in pieces'; others beginning with ft, such as D730 'to take out.' See the cases cited in Nold. op. cit. p. 194, §96. Some of these, formed from verbs with weak stem-consonants, have the appearance of original tri-consonantal stems. So the causatives from stems N"E> and ^"2 and 1"V ; also from stems mediae 1\ See examples given in Nold. op. cit. p. 228, §108; p. 235, §110. Also p. 243, §113; p. 245, §113. Finally, we have under this class the somewhat analogous case where, not the formative element, but a consonant which is the result of a phonetic modification is treated as part of the original stem. See an illustration given by Haupt in Hebraica I 226. 4 6 The Assyrian ittu 'side' is the feminine of idu 'hand,' and stands for idtu. The plural of idtu is idati, but sometimes it is formed from ittu, and gives us itdti. So also the construct aran 'sin,' from arnu, which is a phonetic modification of annu (resolution of the doubling by insertion of" 1 ) is such an analogical formation. See Hebraica I, p. 219, note i. 1 Class III. Division B. Mistaking a stem-consonant for a servile or formative element. We find examples of this kind in Syriac in words in which the stem-consonant n is regarded as the sign of the feminine. See the examples cited in Noldeke, Syriac Grammar, §86. Also in §71, 1, K^-1 Texvlns and NJJSDt? Trraxfi, the J of the ending is a stem-consonant, but is treated as though it were a formative element. The Ethiopic behtew, solus solitarius, becomes in Tigrifia behtti. The final u, however, was misunderstood and regarded as the suffix of the 3d pers. sing. masc. and the word was thought to mean 'he alone.' Hence the other corresponding suffixes were added to denote the other persons, genders and numbers. behtd 'she alone,' etc. (Praetorius, Tigrifia Sprache, §121, 37). Class III. Division C. Analogical changes in the formative elements themselves. Influence of one formative element upon another. The analogical changes which take place in the afformatives of the perfect in the inflection of the Semitic verb have been so thoroughly set forth by Noldeke in an article entitled "Die Endungen des Perfects," in ZDMG 38, 407 fg., that the changes and formations need only be briefly noticed here, referring to that article for a discussion of the whole question. In Syriac we have the 3d person plural ending -tin for -^ of the other languages, after the analogy of the other plural endings. So also in Assyrian we have in the permansive for kasdu some- times kasdtini. In Arabic the 3d pers. plural feminine qatalna, in place of the common Semitic form qatald, shows the influence of the corres- 1 Cf., however, Zimmern, Babyl. Busspsalmen (Leipzig, 1885), p. 12, 6. 47 ponding form of the imperfect, iaqtulna (see Dr. Martin Schultze, "Zur Formenlehre des sem. Verbs," p. 18, §15). In the 2d person singular masculine the Ethiopic has -ka for the common Semitic form -t6i. Noldeke explains this as due to the influence of the analogy of the possessive and object suffix of the same person. It is better, however, to regard it as the influence of the original ending of the 1st person sing. -kti, which has been preserved in Ethiopic, and which in that language has changed the endings of the 2d person masculine and feminine from -td and -ti to -ka and -ki, while in the other languages the reverse has taken place, viz. -k{i of the first person was changed to -tti under the influence of the analogy of the 2d person, while in Hebrew a further analogical change took place, t{L being changed to ti under the influence of the possessive suffix/. (See Haupt, SFG 52, note 10; Stade, §i79a ; Hommel, Semiten, 443, note.) In the 3d person plural masculine the Syriac forms in -d before suffixes are formed after the analogy of the imperfect forms in -{in and -Sn, which retain their original -a before suffixes. In the 2d person plural feminine the vowel -u of the Arabic ending -tunna is probably after the analogy of the same vowel in the corresponding masculine ending. So in Ethiopic the pronominal suffix 3d pi. fern, hdn is probably after the analogy of the masculine homu. In the 1st pers. plural the Hebrew -nti for the common Semitic ending -nd is probably due to the influence of -nfi in W}1% n^n as pronominal suffix is a late analogical formation and is found only with prepositions n®h\2> n^nro (Stade, §348). So ^np^p Gen. 1. 21 is based on the analogy of -ln^D. As to the analogical changes in the preformative of the imper- fect Qal I shall not attempt to discuss them here. See some remarks on this subject by Haupt in Beitrage zur Assyriologie, Vol. I, p. 17, note 20; also pp. 260,11. 27, and 328. Other instances where one formative element has influenced another are the following : The vowel of the preformative of the Hiphil participle in DW (stem E-1P) and 2pft (stem 220') is influenced by the vowel of the preformative of the perfect Q S [?D and 2pn. According to Stade (§i6ib) the H of the preformative of the Hebrew reflexive in ^IWC as compared with the more primitive form Ht, is due to the influence of the H preformative of the causa- tive. 4 8 But the question then arises : to what is the n preformative of the causative due, when in Arabic we have hqtala and in Ethiopic 'agbdra ? The form of the ending of the imperfect plural feminine of verbs V"V and V'U has probably been influenced by the correspond- ing forms of the i" 1 "? verbs. Thus n^3Dfl (stem n2D) after the analogy of n:*»W) (stem rOtf 'to go up.') The organic form appears in t P&^? (stem 3-1^ f to turn.') The form of the preformative of the perfect Hophal of verbs tf"tf and V'U has been influenced by the analogy of the same pre- formatives of verbs VS. Dpjn, stem Wp, and ?n-in, stem ??n 'to pierce, to begin' after the analogy of Tijin, stem TV, for Til 'to go down.' A case of this class is cited by Dr. Rosenberg in "Das arama- ische Verbum im babylonischen Talmud," p. 40. In the Ethpeel of verbs *"B, the * loses its consonantal power in most of the forms, and consequently the 1") of the preformative remains. But in cases where the * retains its consonantal power the n ought not to remain. If it does remain, it is due to the influence of the analogy of those cases where the H loses its con- sonantal power. In the Ethpaal the organic formation is usually found. Another case where the form of one preformative has analogi- cally changed the form of another preformative is the following, taken from Barth, Nominalbildung. The infinitive absolute (Barth, perfect infinitive) of the Niphal appears in Hebrew in three forms : fej?3, hb%n, and 7$j?n. Of these the first is the normal form. The other two are explained by Barth (Nominalbildung, p. 72, §49b) as analogical formations based on the analogy of the form of the infinitive absolute in the other derived conjugations, especially in the Piel and Hiphil. In these conjugations there was a certain similarity in sound between the infinitive abso- lute and the imperfect very noticeable when the infinitive was used to strengthen the imperfect ; cf. Ex. 23, 24 "Wfl ^W. Hence this infinitive was supposed to have been formed from the imper- fect, when in reality it was formed from the perfect. In the Niphal no such resemblance existed between the imperfect and the infini- tive as formed from the perfect, and hence a new form was formed from the imperfect which should have the same assonance to the imperfect as was observed in the case of the infinitive absolute of the 49 Piel and Hiphil. The ground of this explanation is strengthened by the fact that this analogical formation is found only when the infinitive absolute is used to strengthen the imperfect. Where it is used to strengthen the perfect the normal form is used ; cf. on the one hand *1P£?. ^DKn, Q n the other hand nri^ppJ f\b^ Gen. 3I-30- In forms like ^pxn the vowel of the last syllable is due to the influence of the old perfect infinitive of the Qal and Niphal. ?1tD}% Pjbp^ while in ni3fl the assonance to the imperfect has been made complete; cf. B7EK t^Eri, 1 Sam. 27. 1, 'escaping, I shall escape.' Class III. Division D. Inorganic application of formative elements under the influence of analogy. The true origin, nature or significance of a formative element was often misconceived, and so it came to be attached to forms where it was not organically legitimate. An instance of this class is given by Noldeke, Syriac Grammar, §5oB : "The * which was often found apparently without any special reason, came finally to be attached to words ending in a consonant "; cf. Hebraica II 104, n. 1. We have another instance in the use of the old accusative sin- gular ending & H- (accusative of direction) with "?» 3 and IP, and in cases where it is attached to dual and plural forms, thus : ftafltf? 'upwards,' HBD^ < downward.' n»«ftb 'toChaldea.' ' nWQ&n 'to Heaven.' nwy?V 'to Egypt.' nbap'from Babylon.' Sometimes there is a combination of two processes B and D of this class. The language mistakes a stem-consonant in a certain form for a formative element and then adds it where it does not belong. So in the forms of Syriac numerals with determining suffixes (Noldeke, §149). The^ which there appears is organic only in IJ"]? 'we two.' After this analogy we have the other forms P i T j ™ ' they three,' etc. In Ethiopic the pronominal form 3d pers. masc. sing, we'etu, 50 forms the 3d person plural masc. and fern, wfetdmu and weetdn by the inorganic attachment of the suffixes -drnfi and -6n (Prae- torius, Eth. Gramm. §21). Another case of the same kind in Ethiopic is the attachment of suffixes to keVi when used as a numeral. These forms keleitjc, keleHa, keleUi and keleeta, masculine, fern., nominative and accusative respectively, are based on the analogy of other nume- rals with suffixes (Praetorius, §135). Still another case is found in the forms of prepositions with suffixes. The prepositions ?£ 'to,' *W 'unto,' and /D 'over,' are to be referred to the stems vK, Htf, and vtf (see Stade, §375). Hence with suffixes we have the organic forms T^» ^<$» Tl^' etc. But the language mistook this s for a formative element, forming the plural, and then added it to the prepositions where it had no place at all. T"P? 'after thee,' T??n 'under thee," W3 'between us,' W^? 'among us.' Cf. Barth, Vergleichende Studien, ZDMG 42, 348-358, where also the views of Lagarde, Mittheilungen I 232 are successfully combatted. So too in Ethiopic all the prepositions receive their suffixes with the insertion of the long vowel £ (see Praetorius, Ethiopic Grammar, §152). This vowel is organic only in Id' Id 'upon,' with suffixes (cf. Syr. ?tf?=?tf, stem vtf). But the language mistook it for a formative element and then applied it analogically in the case of the other prepositions. Another case of this class is found in the form of the feminine plural before suffixes in Hebrew, e. g. -WTTID-ID. Losing sight of the fact that the "> was organic only in the masculine ^5-10, and that the fern. pi. construct was already expressed by the ending ill", the language adding the ending '•" through the influence of the masculine. On this subject compare the remarks of Praetorius, Tigrina Sprache, §99, the substance of which I quote as follows: In several ancient and modern Semitic dialects the pronominal suffixes are found partially or wholly in combination or even union with a nominal plural ending. The languages presenting this phenomenon are Hebrew, Aramean, Modern Syriac, Man- dean, Amharic, and Tigrina. The similar occurrence, however, of this peculiarity in these six languages is purely external and accidental. There are three different grounds and occasions of this confusion of plural ending and pronominal suffix which are still to be distinguished in its occurrence in the different lan- guages. In modern Syriac and Mandean the plural nature of the nominal plural ending was entirely forgotten and abandoned. See Noldeke, Mand. Grammar, §76 and §141. The possessive suffix pronouns have incorporated the ending ai, which originally served as the ending of the construct masc. pi. The result is that in the case of masculine nouns with pronominal suffixes the singular and plural can no longer be distinguished. The same thing is true in modern Syriac. See Noldeke, Modern Syriac Grammar, §37. In Hebrew and Aramean, however, a knowledge of the plural nature of * was retained, in spite of its erroneous application. This is sufficiently shown in the fact that besides its legitimate application in the masculine, it is found only in the plural of the feminine. So also in Tigrifia and Amharic the meaning of the plural ending in combination with a suffix was not forgotten. When, however, we find, in Tigrifia, cases where the singular has the suffix -Citkmn or -dtom instead of and alongside of the simple -kHm and -6m, this must be because the plural ending -at was thought to denote not only the plural nature of the noun to which it was suffixed, and which was its original function, but also the plural nature of the following pronominal suffix. The ending -dt, therefore, is capable of a two-fold reference — either to the pre- ceding noun, which is legitimate, or to the following pronominal suffix, where it is not legitimate. As a result of this confusion k&ken&fkiim, for instance, may mean either 'your priest,' or 'your priests.' Sometimes there is a misapprehension of the nature and origin of the formative elements of a word. See Dr. Martin Schultze, Zur Formenlehre des semit. Verbs, p. 28, §29. He there points cut that the Syriac interjectional n and the Biblical- Aramean pre- positional ? of 3d person sing, imperf. came to be regarded as the pronominal prefix of the 3d person, and as such was prefixed to the plural as well as to the singular. Before Schultze, however (1886), Mr. W. H. Salter Brooks (1883), in the appendix to his treatise "Vestiges of the Broken Plural in Hebrew," pointed out that the Syriac form with n belonged originally only to the singular, and was afterward applied analogically to the plural forms. He says nothing, how- ever, as to the original nature of this n. Another note on the prefix ? and 1 of the 3d person impft. in Aramean and Syriac respectively. As to the Aramean, see the remarks of Kautsch, Grammar, p. 79. The ? was originally the sign of the optative or precative (compare Assyrian precative particle lu). Afterwards its original signification was lost and it was applied indiscriminately. 1 Compare the examples cited by Dr. Rosenberg, " Das aramaische Verbum im babylonischen Talmud," Marburg, 1888. Dr. Rosen- berg says: "7 is prefixed after the example of the Assyrian, where in the same relations the particle lu is used, to strengthen the assertion. Gradually its force weakened until it became simple assertion and took the place of the prefix \ Because of its relationship with J, however, it changed into the same, and this new form became the exclusive one in Syriac, in Mandean with few exceptions, and in Talmudic was used pretty often." According to Noldeke, in Lazarus and Steinthal, Zeitschrift fur Volkerpsychoiogie und Sprachwissenschaft, Vol. VII, pp. 403-411 the dual originally denoted a pair {die paarweise Verbindung). This signification is still retained in Hebrew. In Arabic, how- ever, it came to mean simply twoness, and extended itself to all the inflected parts of speech. If this could be maintained it would be an interesting case of the analogical extension of an inflectional element beyond its original bounds. In a review of this article, however, by Dr. Friedrich Miiller, "Der Dual in den semitischen Sprachen," Wien, 1875, this posi- tion is questioned, and the counter-position maintained, that the Arabic dual form of the verb, qatalh, is the original. Friedrich Miiller endeavors to support the position by a comparison with the Assyrian (p. 8). This must be abandoned, however, as the Assyrian form cited is now known to be, not a dual, but the femi- nine plural. 2 We need not now examine the details of his argument. On the whole, his reasoning is unsatisfactory, and I do not accept his conclusion (p. 12 and p. 14) that the Arabic dual, in its form and in the extent of its application, represents most faithfully the condition of the parent Semitic in this respect. The position of Noldeke seems to be the correct one. A case of misconception of the nature and significance of an inflectional ending and consequent misapplication is given in Nbld. Neusyrisch. Gramm. p. 218, §102, the use of the ending -nt, which should be used only in the plural, to form the 3d person sing. 1 Cf. Haupt, Beitr. z. Assyriologie, I 17. 2 See, however, Haupt, SFG 71, and ZDMG 34. 53 A similar case is the indiscriminate use of the old Ethiopic Abtina or Abtm in modern Amharic as the title of a bishop or a saint (see Praetorius, Amharische Sprache, §760). For another case of this class see Praetorius, Tigrina Sprache, §131. The plural ending -at, which originally was the ending of the feminine plural, has in many cases taken the place of the original masculine ending -an, especially in the case of adjectives and participles. For the same phenomenon in Tigre compare Schreiber, Manuel de la langue Tigrai, §63. In this connection we must bear in mind, however, that the presumption is that in primitive times the plural endings -im or -in or -an, on the one hand, and -tit on the other, were applied more indiscriminately without regard to gender; compare e. g. ni'nx and D*#J. See Barth, Vergleichende Studien, ZDMG41, 613, 614. In studying the plurals formed by the insertion of h (see Noldeke, Proc. of the Berlin Acad. 1882, 1 178-9; Barth, Ver- gleichende Studien, ZDMG 41, 621 fg.~), we find some instances in which it is inserted probably after the analogy of other older and more original cases. Thus Arabic ummahtit 'mothers' pro- bably after the analogy of an older form abahdt ' fathers '; cf. the Sabean WOK and the Syriac «nri3« : (cf. Hebrew nino** 1 'hand- maids') is not therefore a late form. On the contrary it is primi- tive Semitic. See Barth, ZDMG 41, 624, note 2. In Aramaic (Syriac) the h was originally inserted only before the ending -dtha* The few cases in which it has been inserted before the ending -in (J) such as KH2S (cf. Sab. ^mx, const, pi.), ^£ n 'father-in-law,' KD9? 'name,' are probably of a secondary nature and formed after the analogy of forms in -dtka, Knnm KnnDn. 2 The h arose from an original w because of dissimilation. In almost all cases where we find the original stem- consonant w changed into h in the plural, we find that it immediately follows a labial. Hence the labial spirant w was changed to the guttural spirant h. 1 See also E. Nestle's paper on El, Elohim, Eloah in Theol. Studien aus Wtirtemberg, III. Jahrg. (1882), pp. 249 and 255. According to Nestle DTDX is plural to "?X and the singular ni?K a secondary analogical derivation from DTQK. This view is not accepted by Delitzsch, however. See Del., Neuer Comm. uber die Gen., Leipzig, 1887, p. 47. 2 The supposition of Stade, §182, that the insertion of the il is to give bi- consonantal stems the appearance of tri-consonantalism cannot be maintained. See Barth, ZDMG 41, 625. 54 Cases where a labial does not precede are hence to be consid- ered analogical formations. They are comparatively rare, and the different forms appear respectively in only one language ; so Arab, sanahdt, Phoenician mr6l. An interesting and valuable discussion of some analogical forma- tions of this class is given by Barth, ZDMG 42, 341-358. I shall give only the summary of his views, referring to the article itself for the details. The words for heaven, water, life, entrails, face, value, in the several North-Semitic languages have the plural form, while the corresponding forms in South-Semitic are generally singular. Cf. Heb. D*»B>, Phoen. DD0 , Syr. VQf> K*W , Psalm. IW, with the Arabic samd', Sabean 1BD, Ethiopic and Amharic samdi. Heb. i^Jp, Syriac TP' K* T », with Arab. ma\ Eth. and Amh. m&i. Heb. d*n f Syr. P?n. «Pl f Arab. ham. Hebrew (late, Mishna) D"MD or DWB, Arab, #M*a* or /»zM\ Hebrew D^B. Late Hebrew D*cn, Aram. NW. Barth thus explains these plural forms which are found in Hebrew and other North -Semitic languages. They are all derived from stems I"? and '"?, In the case of the words for heaven and water the original forms were probably samdi and mdi. These were shortened to w samai and mai, thus exactly resembling the usual construct, pi. masc. Also when these forms received suffixes they presented the same appearance as the plural with suffix. Hence the language mistook the form for a plural, and formed the corresponding plural absolute D?E1P and In the same manner, from the forms with suffixes, *p}n» T?P, ifo?, 1W, which, although really singulars, looked like plurals, were formed the absolute plurals D?*D, D"2JD, D'OB, wm. So perhaps also Bibl. Aram. P1D. 'breast,' and Hebrew En?*] 'blood-guiltiness.' According to Barth (Nominalbildung, §123), the prefix ma of the Arabic passive participle of the first form is due to the influ- ence of the analogy. of the passive participles of the other forms. This is a case of the analogical extension and application of a formative element beyond its original bounds. Sometimes analogical influence of this kind produces great mixtures of forms. Thus the Greek suffix -apios is attached to genuine Syriac words (Noldeke, §140, p. 77). 55 Here also may be classed such formations as are mentioned in Noldeke's Mandean Grammar, p. 86 fg. tichuun, dchni "we" forms achtmi "you," by inserting the eft, a contamination of form. Then, as from achtun there was formed achtochun by the possessive suffix of the 2d pi., this again pro- duced the formation achnochun for the first person. Historical and Bibliographical Note. No systematic comprehensive treatment of analogy in the Semitic languages has yet appeared. In nearly all the recent treatises on the grammar, phonetics, and morphology of these languages frequent use is made of the principle of analogy to explain the different phonetic and morphological phenomena of these languages. So, for example, Bickell (Grundriss der hebr'a- ischen Grammatik, Leipzig, 1869), English translation, Outlines of Hebrew Grammar (Leipzig, 1877), on page 42, §47, notes the fact that "the remarkable transition of/, which has arisen from #, to i in the hiphil has perhaps originated according to an errone- ous analogy from the conjugation of the verbs mediae v,y, where this i is phonetically legitimate." And on page 49 of the same work he suggests that many Semitic stems which at present have three stem-consonants origi- nally had only two such consonants, and were made tri-consonantal by the insertion of an auxiliary sound through a pervasive analogy. Also in the following among many other passages of the same work he points out cases of the occurrence of analogy. On page 66, §86, he notes that after a false analogy the ending a, to indi- cate direction, is added to a plural noun in the case of nD^G^a ' unto Chaldea,' Ezek. 16. 29. On page 70, §93, it is observed that before suffixes the plural endings of the masculine are super- added after the feminine plural ending according to a false analogy peculiar to Hebrew; cf. TUpD 'thy horses,' and TJTVfD-ID 'thy mares.' On page 100, §136, it is observed that the feminine singular perfect of verbs tertiae infirmae {galayat, galtit) is usually expanded to n n?K after the analogy of strong verbs. In "Die sumerischen Familiengesetze von Dr. Paul Haupt," Leipzig, 1879, we have many discriminating and valuable obser- vations on the occurrence of analogy in Assyrian and also in the other Semitic languages. 56 See page 52, note 10, where it is pointed out that in Assyrian izzazti (3d plur. pres. kal stem naz&zu 'to settle, to stand') instead of the organic form indzazu is based on the analogy of illakfi 'they go' (stem alaku) y the analogy being induced by the form of the imperfects izzizti and illikti,, (On these forms cf., however, Del. Ass. Grammar, §100.) In the same place it is also pointed out that Hebrew *?3K (grd- form anakii) and ^?Pi? are based on the analogy of *3S in the vocalization of .the final syllable for a?iaku and qataltu, while qataltu is for qatalku under the influence of the 2d person qatalta and qatalti. Unfortunately the treatise which is there promised us on "Associative Neubildungen im Bereich des Semitischen " has not yet made its appearance. On page 66, note 3, of this work it is pointed out that many forms of verbs 1"B (Assyrian) such as usebila ' I brought,' usisib ' I placed,' usSsd 'they led out,' are based on the analogy of forms from stems K"B. On page 73 middle analogy is pointed out in the forms muttdti plural of muttu for mun&ti, similar to the analogy observed in the Hebrew forms TY\rf?\ plural of Tffi 'door,' and Wn^j?, plural of ntfj'bow.' Two articles by the same author in the first volume of Hebraica (Assyrian Phonology and Wateh-ben-Hazael), as also several arti- cles in the first number of Beitr'age zur Assyriologie, Delitzsch und Haupt, Leipzig, 1889, contain much valuable material of the same kind which has been utilized for this present treatise. In Stade's Hebrew Grammar, Leipzig, 1879, many individual cases of analogy are pointed out in the discussion of the various forms. The greater number of these cases are cited and arranged under their respective classes and divisions in the preceding part of this treatise. On pages 110-114 of his grammar Stade treats of the various ways in which original bi-consonantal stems have become tri- consonantal . under the influence of the analogy of those stems which originally had three stem-consonants. For the rest the principal individual cases of analogy which he points out are those cases where a weak or irregular verb-stem is treated after the analogy of a strong stem, or where the different classes of weak verb-stems are confused, for instance, confusion of stems n 'v and **"?, §i43e, note 1. 57 It is hardly necessary to say that in the excellent grammars of Theodor Noldeke (Syrische Grammatik, Leipzig, 1880; Neusyr- ische Grammatik, Leipzig, 1868; Mandaische Grammatik, Halle, J 875) we have a comparatively full presentation of nearly all the analogical phenomena of the respective languages of which he treats. These different formations, as given by Noldeke, will be found cited or referred to in the preceding pages of this thesis. The same remarks to some extent apply also to the grammars of Kautzsch (Grammatik des Biblisch-aramaischen, Leipzig, 1884), Praetorius (Grammatik der Tigrina-Sprache, Halle, 1871 ; Die Amharische Sprache, 1879 ; ^Ethiopische Grammatik, Leipzig, 1886), and Delitzsch (Assyrische Grammatik, Berlin, 1889). Considerable valuable material is also contained in various essays on different topics of comparative Semitic grammar recently published. See especially Noldeke in ZDMG 37, 525, Die Verba 'W im Hebraischen; Vol. 38, 407, Die Endungen des Perfects ; Barth, Die Nominalbildung in den semitischen Sprachen, Leipzig, 1889, also his Vergleichende Studien in ZDMG 41, 603; 42, 413 ; 43, 177. The material of these articles has been cited and discussed in this paper. There is not much that is new in Wright's Comparative Gram- mar of the Semitic Languages, recently published (New York, 1890). INDEX OF FORMS CITED IN THIS THESIS, The numbers refer to the pages. Hebrew Forms. ™«, 53- 2n $, 29. ni-ina 32. t:jdk, 12, 52. w$, 22. " n?ta, 18. Wfe^, 37. T& 12. nines, 53 . ^Vpk", 24. !»K 28. in^K, 38. HP*, 38. V???, 30- 'BSK, 38. nap**, 38. ntaia, 16. no, 17. T - T ?, 23. WTO, 50. W*3»3, 50. "1J£, 24. ;im 37- nj|, 22 r4, 23. TO% 36. JCM7- run?" 7 !, 22. *vi 38. P% 2 3- ^0, 49 . P30, 29. *PP, 30- w^n, 36. rifn, 18. T^th, 18. ■inp-in, 36. PpJin, 3 6, 48. ^ID, 29. ^•?n, 18. nnnn, I7 . rir>;n f 40. nrpn, 40. npgn, 12, 49. o!?sn, 12, 49. sjrwpi?, 16. triK'3|n, 17. rpjn, 18. pJ^n, 23. w! 30. W 30. •in s _pvn, 23. tofcn," 48. ^n, 48. nojotfri, 49. n Bffy 30. niaann, 17. rpaann, 17. fei??n, 47- Dnn*ni :) 23. «iy, 15. ^ ■$B& 15. •i^m, 15. JIM, 30. ?api, 30. n$ir, 32. W, 29. ^$an 37. p?n, 38. ^n, 29. *>m 23. ^*n, 16. •inp.-m 38. ^'n, 16. n^n, 16. D*n, 54 . ^?S 23. oon, 23. n iBP, 40- ipn, 28. i? n , 30. nw, 30. •nin, 39. D^tpnn, 36. Wn )3 8. ™V, 36. TO 31. w, 23. TO, 17. »*., 17. *n:, 18. W, 36. 6o \f\V\, i 7 . ?^, 18. np", 18. nfc», 16. W"fe, 37. TC, 18. n^pi, 17. w, 17. w, 18. rfc, 16. ^i 17. ll'^ 1 !, 22. n^!, 18. "&«, 22. *i*, 22. **?*., 22. ID!, 18. 3K, 18. as*, 18. rvp!, 18. *»!, 15. "», 18. yi, 17. ?»&♦, 14. XT, 29. R, 17. ^T-, 17. m:, 17. »¥, 16. **y., 16, ™*. 17. n®» 17. ?*!, 29. **& 15. *«*., 15. ipfl^?, 25. PK*i7- *b, 15. ™??, 16. *?, 16. •ante, 16. ninpa, 22. no^3, 49. n ^, 22. •n^?^, 23. ptefe, 29. Ktl^, 16. nap 1 ?, 49 . npyp 1 ?, 49. n'apV 23. >$?, 29. 1'^, 29. 3$& 37. i^?, 36. n^nsp, 49 . ^23- n^nio, 15. W 37- ™> 54. 3"J?», 37. rras 38. n ?P, 38. D*pO, 29. nifc&p, 16. n*&&, ^ n*6p, 16. Ap, 16. '???, 38. TP9, 17. DWDp, 15. Dbp/23. ""ife®, 29. W, 37- 3W, 37- 1WP, 37. r?£, 29. npnvp, 49. nibnpp, 30. B^Oi^P, 30. Q'i?P, 47. ^i?P, 37- nnp, 29. «£>1P, 15. D*'W, 29. tt#0, 39. r^,2 4 . TW, 24. m 17. ng3j, 40. wiaj, 17. m 37. WTJ, 40. naijji 17. T3. 40. ^ 36. B?nfOJ, 36. WODJ, 16. *P?j, 12,49. DP3, 40. ™, 17. np;, 40. 2p^, 40. njw, 40. TPJ, 30. «^?3, 16. w&j, 16. Q^a? f 29. n m 38. 9W, 38. n?E? f 31. ^H3J, 40. nns-ja, 16. P T ?, 40. ^3, 53. ^P, 23. nnp, 18. •i^niD-iD, 50. >nm 38. *m 38. Tli?, 12. TO, 23. to, 23. 6i m, 23. *i#, 22. D^W, 23. D*P»tf, 23. 3^37- n ?1?> 38. nrs, 23. rfe, 4 o. ri ^'^, 40. *3*JP?, 40. WB, 29. ** 54. WE, 38. 3E1*, 37- n*9¥, 16. i^V, 23. n^np, 30. ?8P, 14, 39. WR 23. w, 13. n'Kap, 16. ri3|5, 23. nimp, 16. mnBTj, 22. nm sl 23. °W, 23. nan, 16. mv, 38. ««$?>, 24. afc^i 17. nn^, 29. HTf , 39- nns^ 23. T& 37- Dp#, 23. nV, 29. ^23. nnvhf, 36. ^»VVs4- fctap, 29. m\ 15. *jp, 32. nfcfcib, 16. p?^, 38. ngw, 22, 30. ninj#>, 22. in^, 23. *Qh, 16. H3PI, 17. pran, i 7 . Tfl?n, 12, 50. Q^n, 23. b-1&6fi, 15. DW^, 15. ^an, 22. ■injjp, 38. nj»aro, 18, 48. iW5," 17. "JVE, 16. t P'^, 48. HTDK, 31. n^n, 31. m^n, 31. nsiD:n, 32. Late Hebrew (Mishna). ppiorm, 3 2 - n»*on, 31. na^, 31. nk«w, 31. npnsi, 31. nn*np, 31. "nptf, 32. *w, 32. Aramean, including Modern Syriac and Mandean. Ksata, 25. °??, 35- F|fcO, 19. KB»3, 19. D 1?> 35- W..35. "9837, 21. fa. 35. «W, 41. Nrn^ft 4 i. k^d, 25. **% 53. K3313, 25. *W?*i 53- SDn:, 28. tniK, l8 . tol 32. achnOchun, 12, 55- KDI, 22. achnun, 12, yen, 28. achtOchun, 12. •JWTO, 38. achtun, 12, 55- *39*, 46. PDX, 19- «in, 32. "1DN, 19. K*P n , 53- WS, 30. "|OTI, 28. ^?. 35. 3SO, 19- 62 fcWSPP, 46. t&BD, 45- W)D, 19- n^a, 28. *>?"!?, 30. ■*}. 35. "jpnD, 45- ^no, 45- amin, 32. asnata, 45. ummahat, 53. baid, 32. jalid, 28. jaijd, 20. hadana, 30. hidn, 30. hudne, 30. hize. 22. haiiit, 20. husr, 28. dikr, 30. rihle, 29. rugb, 28. ruhb, 28. zuhd, 28. sumull, 28. da'in, 12, 31. KJB?, 28. «M, 19. D ?V, 35- ron, 28. Kgtn, 28. yiW, 28. mra>, 18. Ki?fe\ 28. Arabic Forms. dahuk, 32. darub, 31. salim, 32. sihn, 29. sarf, 28. siqi, 3°- sanahat, 54. saiiid, 20. c ujb, 28. azim, 28. umudd, 28. amul, 32. aiijd, 20. gasfim, 32. fi'l, 29. qattil, 12. qattala, 14. qaium, 20. kabir, 28. «m, 53- Wf£ 22. «JPB>, 28. KOW, 30. Dxn, 19. s^n, 19. t^n??, 49. nmn, 45- kabunn, 28. katir, 28. karim, 28. kastib, 32. kalib, 12, 31. ma'iz, 31. maiiit, 20. naqid, 31. uu'til, 31. uutCib, 31. uujfib, 31. uujfid, 31. uujtiz, 31. uurud, 31. hattib, 32. haiiin, 20. ia'us, 32. iasqi, 30. iuqattilu, 14. Ethiopic Forms, including Amharic and Tigrina. haydola, 44. laala, 43. malakata, 43. malada, 43. maraka, 43. marakuaza, 43. masaqala, 43. moshara, 43. mashaga, 43. manazara, 43. managaga, 43. manatja, 26. manatsha, 43. makuata, 43. mogata, 43. masana, 43. shalatana, 43. rate c , 32. sanakala, 42. shamana, 42. shamagala, 42. shamatata, 42. shaqaba, 42. shanakara, 42. shanagala, 42. shakarakara, 42. quaratama, 43. 63 bahata, 43. behta, 46. bahtawa, 43. barata, 43. talala, 42. talalafa, 42. tarasa, 42. tanabaya, 42. tantarasa, 42. tanakuala, 42. tanafasa, 42. tafia, 42. takata, 42. tawaba, 42. teda'i, 13. amalaka, 43. amara, 42. arasa, 42. asalama, 42. asama, 43. asmat, 43. abuna, 53. 'ayd61am, 43. adaga, 42. kele'Sta, 50. kele'£tu, 50. kele'&a, 50. kele'Sti, 50. we'etdmti, 50. wad'a, 13. zawatara, 43. ida', 13. ida'fi, 13. ida'a, 13. daraqa, 42. dasasa, 42. danaquara, 42. danagala, 43. danagara, 42. danagada, 43. galam6ta, 43. garamama, 43. gazata, 43. taraqama, 43. tanawata, 43. sabaraqa, 42. da, 13. aran, 46. attasab, 18. £nzu, 23. ensu, 23. §qir, 19. izzazfi, 19. izanan, 39. illika, 19. Assyrian Forms. isitate, 41. isakan, 39. itati, 13, 46. ittasuni, 19. us£bila, 18. usesu, 18. usesib, 18. bintu, 23. zikru, 30. kasduni, 46. li-ri-qu, 19. l£tat, 41. muttabil, 19. muttati, 41. Biographical Sketch. Abel Henry Huizinga was born in Kalamazoo, Michigan, August 1 8, 1859. He graduated from Hope College, Holland, Mich., in 1880, as Bachelor of Arts. The next three years he spent as a student of theology in the Theological Seminary of the Reformed Church in America at New Brunswick, New Jersey, and graduated from that institution in May, 1883. From 1883 to 1886 he pursued the study of Philology in the Johns Hopkins University, Baltimore, Md. The course of his study included Sanskrit and Comparative Philology under Prof. Bloomfield, Greek under Prof. Gildersleeve and Semitic Languages under Prof. Haupt. ANALOGY SEMITIC LANGUAGES DISSERTATION PRESENTED TO THE BOARD OF UNIVERSITY STUDIES OF THE JOHNS HOPKINS UNIVERSITY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY ABEL H, HUIZINGA BALTIMORE: PRESS OF ISAAC FRIEDENWALD CO. i8qi. V % 1&* +t- V* s y > vV r -" -'.'vjt^JSBBSfcjL*! LIBRARY OF CONGRESS 027 250 804 5 H ra man row Httfl limit ■1 im ■ mum K BUR m\ nicfliinni! HMOHHa ro Wll« Mm m HWH