Latin future in -b-

Thu Nov 21 13:15:50 UTC 2002

Note: the previous posting from Guy Deutscher was truncated, so I'm
reposting his communication.

Dorothy Disterheft

************************************************************************

Dear all,

Various people sent me very helpful answers, but for some reason, none of
these seems to have been transmitted to the whole list. So in case you
have been losing sleep over the Latin -bo- future, here is a short digest.

The problem, it seems, is made complex by alleged parallel formations in
other I-E languages (Celtic), and more importantly, by the parallel
formation of the imperfect in -bam- in Latin itself. So in fact, most of
the people who responded actually favour a third alternative, which was
not mentioned in my query, namely that the -bo- future is a later
analogical formation on the imperfect in -bam, which itself is taken (by
most, not all) to derive from *bheu. (Arguments for this view in print can
be found in Baldi 1976, 1999, and Hewson and Bubenik 1997). According to
this view, then, the -bo future is related to *bhu only indirectly (via
analogy), and not as a direct grammaticalisation from *ama-bh(w)u. Carol
Justus, on the other hand, thinks that it's *possible* that *bheu
grammaticalised independently in two roles (to give imperfect and future,
as later 'habeo' in Romance: 'habeo cantatum' and 'cantare habeo'), but
she would certainly not claim that this was a secure reconstruction.

In short, in view of the responses (reproduced below), while it would be
unfair to call the *canta-bh(w)o etymology 'discredited', it's certainly
very far from 'credited'. I guess that the reason why the canta-bo future
is so often quoted by linguists must be that it is so neat to have two
parallel cycles of grammaticalisation of periphrastic future in the
history of Romance: pre-Latin *canta-bh(w)o, and Romance cantare habeo.
But since we have many other cases where we can be on less shifting
ground, it's probably better to find other show-case examples.

Many thanks to all who responded (John Hewson, Gonzalo Rubio, Carol
Justus, Martin Huld, Andreas Ammann, Paul Hopper), and a digest of their
views follows.

Guy Deutscher.

-----------------------------
>From John Hewson:

Phil Baldi's 1976 article in Language (Lg 52.839-850) is an excellent
overview of the question. When Vit Bubenik and I looked at the question
(Hewson & Bubenik, Tense & Aspect in IE Langs, Benjamins 1997) we
concluded that future tenses in IE langs are rare (only Italic, Baltic,
and some Celtic), and all late developments (most IE languages express the
future by aspect, not tense). They are definitely not inherited, but new
formations.

If you look at the whole Latin system, you will find three forms (past,
present, and future tenses) for the perfect, and similarly three forms for
the non-perfect, the perfectum/ infectum constrast being aspectual.

To form the past and the future in the perfect, -erat and -erit (etc) are
added to the perfect stem, the future having the endings of the ancient
subjunctive. Here we can see that what was originally the stem of the verb
"to be" (-er-) has become the marker of the non-present, with -at and -it
distinguishing past and future. In other words Latin formed a future
perfect with the inflections of an ancient subjunctive of the *es- stem of
the verb "to be".

In the three tenses of the infectum the -ba- is also a formation from the
*bhu- stem of the verb "to be" (see Baldi 1976); this form is not
problematic. The problem is the future. Looking at the total picture, and
avoiding an atomistic approach, it is clear that just as -er- came to be
used in the perfectum as a non-present marker, -b- likewise came to be
used
in the infectum as a non-present marker, adding the subjunctive endings to
form a future paralleling the future perfect. In other words, the Latin
futures in -bo, -bis, -bit, are a late analogical formation. You will find
this late out more coherently in Hewson & Bubenik 1997:192-194.

--------------------------------------------------------------
>From Gonzalo Rubio:

I think you may want to take a look at Philip Baldi, _The foundations of
Latin_ (Berlin: De Gruyter, 1999), pp. 397-98. As Baldi points out, the
future must be understood in relation with the preterite in -ba:-. Both
formations are parallel, as in the case of the perfect and pluperfect
(amavero, amaveram). The
imperfect in -ba:- most likely exhibits an (aorist?) optative form of
the verb "to be" (*bhu-a:-m) preceded by a participle or participle-like
form. Heinz-Dieter Pohl argued (in 1992, reference in Baldi) that the
/b/ could be an "epenthetic" consonant of sorts preceding an optative
marker -a:- (this was proposed before Pohl, I'm sure). But this seems
rather far-fetched (no epenthetic b's occur in that context normally in
Latin).

---------------------------------
>From Carol Justus:

I don't know what other Indo-Europeanists think, but in general IE
etymology is a hypothesis-forming process. One proposes a plausible
etymology, then checks related data to see to what extent it is confirmed
(or not) by independent findings.

On the one hand *bheu does fit the sound correspondences. It is also to be
compared with the imperfect, however. Do both both -ba- imperfects and
-bo- futures have a similar origin (Jay Jasanoff has written on the
imperfect, I forget where, but someone out there probably recalls)? It is
true also that in the evolution of the Romance languages we find one verb
giving rise to more than one tense form. I am thinking here of 'have'. The
standard view that I learned some decades ago was that the French future
(e.g., trouver-a 'will find') is built on the infinitive to which a form
of 'have' was suffixed and phonetically reduced. The perfect by contrast
was formed with 'have' as an aux (e.g., a trouve have found'). Did *bheu
'be' similarly serve as suffixed aux that had two separate evolutions? It
seems that the value of this etymology is in raising this kind of
question. More interesting yet would be the use of 'be' rather than 'have'
at the Latin stage of IE evolution.

On the other hand, if you maintain, which many do not, that the Sanskrit
and Greek verbal systems represent the oldest layer of IE, then you will
want to find something that looks more like a Sanskrit and Greek future
form to relate to Latin. The suffix -su- with -s- has that advantage, but
I can't quite see how the sound correspondences would work. Personally, I
would favor *bheu and the questions it raises over what seems like an ad
hoc *-su- to prop up a wobbly hypothesis about language grouping and
chronology.

The etymology *bheu in and of itself is also more important within the
framework of grammaticalization as a hypothesis in terms of the questions
that it raises, and it should also be treated that way. For example, if in
IE it happened that a form of 'be' was suffixed to form a new tense, then
might similar things have happened elsewhere? If they did, then 'be' looks
like a pretty good origin for the Latin future and imperfect. This would
be hypothesis confirming information. But to elevate this etymology to a
proof is a bit tricky. Heine & Kuteva's World Lexicon of
Grammaticalization would seem to help lay a foundation for this kind of
inquiry.

--------------------------------------------------------------------------------
 -----
>From Andreas Ammann:

I've come across yet a third proposal in: Kurzova Helena (1993): From
Indo-European to Latin: The Evolution of a Morphosyntactic Type.
Amsterdam: Benjamins. On p. 188 she claims that we are not dealing with
*bh(w)u-, but rather with *dhe:- (colon instead of the usual macron for
vowel length here), just like, according to her, in the case of the
imperfect in -bam, bas, bat... That would be the same marker as in the
Germanic dental preterit, i.e. a root for 'put' (which shifted to 'do' in
Germanic). Latin -b- is one of the word-medial reflexes of IE *-dh-, the
other being -d-. It seems clear that the -b- in the Latin future is the
same as the one in the Latin imperfect, but what is not clear is which of
the two is older. Some authors say its the future, some say its the
imperfect, and the respective other form arose analogically. (I find it
interesting that during my little sampling I havent found a statement that
they may both have been grammaticalized at about the same time.) Anyway,
Kurzova argues that the imperfect form came first. If you want to derive
the -b- from the desiderative, you have to think of the imperfect as later
(or totally unrelated). There is a danger of circularity here, of course.

---------------------------------------------------------------
>From Martin Huld:

The problem arises when the account of the Latin future is expanded to
include data beyond Latin.  The Faliscan future carefo can be regarded as
the same as Latin -bo with Faliscan retaining the intervocalic voiced
spirant that occluded to -b- in Latin.  Beside the future in bunt (3pl) is
the imperfect in -bant whose Oscan counterpart fans again preserves the
original spirant value.
 The traditional explanation is that -bo et all. is analogical to -bam et
all. on the model of eram vs. ero.  That means that -bam represents a
subjunctive of *bhueA-, but fui, OIr. boi point to PIE *bhouH-, which
means unmotivated
schwebeablaut.  This is probably the reasoning behind Sihler's
pronouncement on the future in -b- "This is at bottom the same formation
as the imperfect in *ba- and, like that formation, is partly transparent
and partly opaque. (New Comparative Greek and Latin Grammar p. 558).

Nevertheless, there are other sources for a Proto-Italic voiced bilabial
spirant besides PIE *bh(w) and that is the problem.

Matters become more complex when Celtic (the Irish f-future)is brought
into the equation.  Like the Latino-Faliscan bo/fo future; the Goidelic
f-future is not found in the other branch of Celtic, Brythonic.  On the
face of it, Irish
-f- (-b in word final position but that is a later development) should
reflect *-sw-, which would mean (as Thurneysen in Old Irish Grammar p. 398
[and also
Pedersen] maintained) that the comparison is simply invalid, but he notes
that some have argued that both forms arise from *-bhw-, but there is no
independent evidence that Proto-Irish *-f- can arise from *-bhw-.  Without
giving details, Meillet held that Le futur latin ... amabo, monebo, audibo
fal. carefo, pipafo a son corresponant exact dans le futur irlandais en
-b, -f-(Les dialectes indoeuropeens 37).

Even if Thurneysen's account is correct and the Irish is merely an
accidental
similarity and the future arose analogically from the imperfect by
association
with the regular paradigm ero (from an IE thematic subjunctive) and eram
(of obscure origin), the periphrastic account still leaves the formation
of the stem unaccounted for, and that fact probably accounts for
Szemerenyi's hesitation.  For what its worth, Buck (1933:278-81) expresses
the same view,
that periphrasis with *bheuH- is the most probable explanation, but the
details of the analogical development are unclear and there is no credible
explanation for the stem formation.

As is often the case, the handbook accounts record the surface details but
have overlooked the footnotes which all along have acknowledged the
difficulties in the periphrastic explanation.