IN

STUDIES IN THE MALAGASY LANGUAGE:

NO. V. THE COMPOUND VERBAL PRefixes.

N my article "On the Inflexion of the Verb in Malagasy," in No. IV. of this magazine, I tried to show that the causative and reciprocal prefixes in Malagasy (mampan, mampàha, mampi, mampiha, mifan) are compounds, formed in a very simple and systematic manner by a combination of the corresponding simple ones (man, mi, maha, miha), placing a new prefix before the verbal noun of the verb formed by the simple one (e.g. mamely, to beat; mi-family, to beat one another), only subjecting this juxtaposition to the general laws of euphony; e.g. manao, to do; fanao, manner of doing; mam-panao (for man-fanao, which would be an impossible combination in Malagasy), to cause one to do a thing. And I have up to this time seen no reason for changing my opinion.

But a writer in the last number of this ANNUAL (No. X. p. 255), briefly reviewing a little French pamphlet by Père Jean, says that if this author is right in considering the Malagasy causative prefix mampi to be derived from the Malay memper, "the theory of Mr. Dahle would of course fall to the ground." He has, however, himself some doubt as to the correctness of this derivation, chiefly because it seems to him that the causative sense in Malay is produced less by the prefix memper, than by the affix kan generally coupled with it.

The suspicion intimated is certainly reasonable enough; for, as a rule, the causative sense in Malay clings to the affix kan, and does not at all depend on the per (=ber) in memper. In Malay the causatives can be formed in two ways, viz.-(1) By the prefix men (=Malagasy man) and the affix kan. (2) By the prefix memper (i.e. men-ber) and the affix kan (applied at the same time). It is only quite exceptionally that memper alone renders a verb causative.*

If we now compare with this the compound prefixes (as I consider them) in Malagasy which I have enumerated above, we shall have to make the following remarks:

(a) All the Malagasy verbs in mampi have a causative sense; whereas the Malay verbs commencing in memper, as a rule, only get a causative sense when the affix kan is added.

(b) In Malagasy the only manner of forming a causative verb is to make it commence in mamp (mampi, mampan, etc.); whereas in Malay the causative verbs are more frequently formed without the syllable per (in memper), and only by the affix kan and the prefix men. This proves that the syllable per in Malay is generally of no importance whatever for the causative sense, while the p (pi, pan, etc.) in Malagasy is essential to it.

(c) If the Malagasy mampi is to be "derived" from the Malay memper, what then has become of the final ? This letter may in these languages

See L'Abbé P. Favre's Grammaire de la Langue Malaise, Vienne: 1876; p. 117-126.

easily pass into or d and even into z (Malagasy zato, Malay ratus), but does not often fall out altogether where it really belongs to the root.

(d) Provided that mampi could be explained as suggested by Père Jean, what then about mampan, mampaha, mampiha, and mifan, which are evidently all of them formed by analogous rules? Any one who undertakes to explain one of these from a Malayan source must try to account for the others too, as they must undoubtedly all go together.

I venture to think that these observations, brief as they are, will prove sufficient to show that Père Jean's "derivation" on this point at least is a case of "lucus a non lucendo." And if this is the strongest blow my theory is to be exposed to, I see no reason to feel nervous about it.

It was, again, not to be expected that a compound prefix could be proved to be "derived" from another language, for such forms seldom are so obtained. The simple prefixes may be (and often are) essentially the same, but the manner of combining them is often characteristic of each individual language; and it was not to be expected that it would be otherwise in Malagasy, which is certainly one of the finest and best developed languages in the family to which it belongs.

But although each language may have its own peculiar way of combining the simple prefixes, it may be anticipated that there would be some analogy in this respect between languages of the same family. And so there is. So far, a reference to the Malay memper does not only not overturn my theory, but is a very strong argument in its favour. For what is the Malay memper? Nothing but a compound prefix. Just as mam-pan-on my theory-is composed of man-man, and mam-pi of man-mi (through the mediums of verbal nouns in fan, fi, corresponding respectively to the verbs in man and mi), so is the Malay memper composed of the two prefixes men and ber, both of which may be used alone to form a verb, the first one generally in a more transitive, the second one in a more intransitive sense, corresponding nearly to man and mi in Malagasy; e.g. ber-pukul, to beat; but memukul, if an object is added. When mem and ber are combined in Malay, the affix kan is generally added at the same time; and this then makes the verb causative; e.g. ber-anak, to have children, mem-per-anak-kan, to cause to have children. The peculiarity of Malagasy as compared with Malay is, that the combination of the two prefixes is effected by means of the verbal noun of the verb formed by the first one of the two as pointed out above); and that this combination of the prefixes is in itself sufficient to render the verb causative, without any additional affix as is required in Malay.

It may perhaps be said that my "verbal noun," as an intermediate link, is, at the best, only a hypothesis. This I admit, but it is a hypothesis which explains the facts consistently with the euphonic laws of the language. Besides this, there are a good many facts which go far towards proving that this hypothesis is the true explanation of the matter in hand. I may mention the following:

1. The conception expressed by these verbs (causative and reciprocal) is certainly a compound one, and it is therefore only natural that the form of the verb should also be compound.

2. In the cognate languages compound conceptions (compound actions

and states) are generally expressed by compound prefixes (as we have already seen to be the case in Malay). It was then to be expected that such should also be the case in Malagasy.

3. If mampan is not=man+man, or mampi not=man+mi, etc., how is it that the first form invariably forms the causative of verbs whose simple prefix is man (e.g. mandèha, màmpandéha); and the second one, as invariably, the causative of those whose simple prefix is mi (e.g. mijàly, màmpijàly), and never vice versa ?

4. But if it must be admitted that these prefixes are compounds, then the only way of explaining their form in Malagasy is to consider the verbal noun (of the forms fanao, fandèha, fijàly) the intermediate connecting link. Man-mijaly, for instance, could never become mam-pijaly but through the verbal noun fijaly; but man-fijaly naturally becomes mampijaly.

5. This procedure is also a very simple and logical one. The verbal noun is simply treated as a new secondary root, and the new prefix placed before it simply adds its own force to that of the prefix of the primary verb from which the verbal noun has been formed (e.g. mijaly, to suffer; fijaly, suffering; mampijaly, for man-fijaly, to cause suffering); no other changes taking place than such euphonic modifications as always must occur when a prefix terminating in a consonant is to be joined to a root beginning with a consonant that does not agree (euphonically) with it.*

The meaning of these prefixes becomes very simple and clear on my theory, the whole being governed by the following rules:

a. Active (transitive) prefixes may be combined with active ones, or in other words, reduplicated, as in mampan for man-man. In this case the verb generally becomes doubly active, i.e. can be construed with two objects.

b. An active prefix may be placed before a neuter (intransitive) one. This will, as a rule, give us a causative verb, but one that cannot take a double object.

c. A neuter prefix before an active one (mi-fan for mi-man) gives the verb the sense of reciprocity, i.e. the subject is both acting and acted upon; e.g. mifamely, to beat one another (mamely, to beat). In this case the prefix mi seems to have almost a passive sense, a sense which also, though seldom, is found in the simple verbs in mi; e.g. misàsa is 'being washed'.

d. If the active prefix (man) in the simple verb has a neuter sense, or if the neuter prefix (mi) has an active sense, which is exceptionally the case, this will appear also in the compound forms of them; e.g. mampivély can take a double object (against the rule b), because mively is transitive, notwithstanding its neuter prefix. On the contrary, mampandeha would

A Malagasy scholar may object, that although mampi is euphonically correct, we should rather have expected mamijaly, as roots with for v or p for their initial letter generally drop it after such a prefix and change then to m. This is true; but we have still traces of an older practice in such verbs as mambòly, mambóatra (for mamoly, mamoatra), and this analogy has been followed in the compound prefixes. In ordinary compounds, n-f always becomes m-p; eg, dlom-pidina for olona fidina,