Ramnants of Optative and Secondary Endings in Latin


coclea mod
日本語 / japāniski / יפנית
Hello forum,

I am trying to understand how the Latin subjunctive was originally optative in PIE.

From Monteil (1986) Eléments de phonétique et de morpholgie du latin [translated into Spanish p 368; not that I read Spanish particularly well but it was the only version available on the Web],
Por el contrario, la forma *-<-yeH₁ del sufijo, inicialmente característica del singular, ha podido extenderse al plural en algunas formaciones. En los verbos en -ā-re (1ᵃ conjugación), el latín presenta en efecto formas del tipo stēs, amēs, donde la vocal -ē- (como la muestra el tipo corresondiente osco deivaid "iuret", procedente de *dei-wā-ē-t) remonta a una contracción de -āē-. Algunos lingüistas han querido indentificar, en la vocal -ē- segundo elemento de este grupo, el antiguo morfema indoeuropeo de subjunctivo temático (amēs<*ama-ē-s, como el gr. τιμᾶς<*τιμᾱ-η-εις). Pero la desinencia secundaria -d<*-t, claramente documentada por el osco deiuaid, supone necesariamente una antigua formación optativa (v. p. 365).
we learn that Oscan forms subjunctives with secondary endings. The subjunctive morpheme -ē- was, therefore, originally an optative marker since in PIE subjunctive forms require primary endings, and optative secondary endings.

The PIE reconstruction of verb endings by Szemerényi Oswald (1970) Introduction to Indo-European Linguistics [translated into English p 234] as below suggests, compared with forms in Old Latin such as feced, kapiad, sied, classic forms such as fecit, capiat, sit are results of later convergence of -t and -d:

I am not sure how strict PIE was in distributing primary and secondary endings according to the tense and mood (primary for present and subjunctive; secondary for past and optative). Even if I'd take it for granted that the distribution was pretty rigorous, I still don't understand why it was necessary for Monteil to cite an Oscan example in order to prove a case for Latin. Especially intriguing is that Latin was not lacking in secondary endings in older records.

Anyone could explain the ethos behind Monteil's reasoning? His "claramente documentada por el osco deiuaid" implies that what was obvious in Oscan was less so in Latin. What would that be?
Last edited:
  • sokol

    Senior Member
    Austrian (as opposed to Australian)
    Hello Flaminius,

    and no, I too can't make sense of this, offhand.
    The chapter of the Szemerényi quote (available on Google books where however only parts are made available) is 9.2.1 Active endings in the present and aorist system, which one can find in the German edition of Szemerényi (4th edition of 1990) on page 247.
    I'm giving the full paradigm of Latin endings according to Szemerényi (page 248 in the German edition) for those who don't have full access to a copy:

    Primary endings:
    1st/2nd/3rd sg. -M ....-S ...-T
    1st/2nd/3rd pl.. -MUS -TIS -NT
    Secondary endings:
    1st/2nd/3rd sg. -M ....-S ...-D
    1st/2nd/3rd pl.. -MUS -TIS -NT

    Szemerényi also emphasises in the text that this system (and its distribution) of primary (present, subjunctive) and secondary endings (aorist, optative) wouldn't and couldn't be deducated at all from languages like Latin but is only obvious through Aryan, Hittite and Greek where they're clearly differentiated.
    This may become clearer if one takes a look at the Hittite endings, again according to my German copy of Szemerényi (p. 248):

    Primary endings:
    1st/2nd/3rd sg. -MI ....-SI ...-TSI
    1st/2nd/3rd pl.. -WENI -TENI -NTSI
    Secondary endings:
    1st/2nd/3rd sg. -M ....-S ...-T
    1st/2nd/3rd pl.. -WEN -TEN -IR

    And Aryan languages also differentiate in dual.

    I can only offer a guess - probably Monteil just wanted to emphasise with this that in Classical Latin third person singular secundar ending only was -T (due to the later phoneme neutralisation you mentioned) and that Oscan still retained -D / -T differentiation.
    Or probably he discounted those few ancient Latin inscriptions where -D (PE) and -T (SE) were still differentiated as not sufficient to prove the point.

    I don't know when Oscan died out exactly, and if there even exist more ancient Oscan documents than ancient Latin ones (that is, from the latter of course such ones which too did differentiate -T/-D here): probably even those ancient Latin documents are even more numerous; Monteil's statement however suggests that this were not so, and that Oscan were better documented.

    I guess that Oscan alone too only differentiates PE and SE in the third person singular, and that by Oscan testimony alone this system of PE and SE would not be deductible at all, but probably Oscan consequently differentiated them - while in Latin this only is the case in but a few ancient inscriptions.


    Senior Member
    As I understand it, only Oscan attestations are capable of showing the two vocalic elements (ai < *āē) and the secondary ending d at the same time. So, yes, there are a few examples of -d in Old Latin, but the hypothesized contraction *āē > ē had already taken place, and Monteil wants to establish a link between the ending and this sequence of vowels.

    Still, the Oscan example doesn't prove that the ē has to be associated with the optative. It could be a thematic subjunctive morpheme followed by an optative ending, why not? Since we know that the Italic subjunctive is a mix of PIE subjunctive and optative morphology (but maybe only at the level of hybrid paradigms, not hybrid forms — I really don't know).

    So I won't say anything about the quality of the overall argument, but I think it's clear why Monteil doesn't use Old Latin evidence: there isn't any. Now, whether the conclusions for Oscan carry over to Latin, that's for you to decide…
    < Previous | Next >