Perfective-Imperfective

[Larry Trask]

I can't possibly reply at length to Pat Ryan's vast posting, but I'll
try to deal with the major points.

First, it is a plain fact that our linguistic terminology is much less
orderly and consistent than we might wish.  Nobody knows this better
than I do: I've had to grapple with it in writing my several
dictionaries.

Second, aspect is an area in which our terminology has long been
exceptionally variable and inconsistent.  Partly that's because the data
are so messy and hard to interpret.  And partly it's because there are
two independent traditions of aspect study in Europe, the Slavic one and
the "western" one.  Being based on the facts of different languages,
these traditions differ in many respects, not least in terminology.
It is, however, the western tradition which has generally prevailed
among general linguists, at least in the English-speaking world, insofar
as anything has prevailed.

Now, I know of five scholarly book-length studies of aspect published in
the last 15 years: those by Dahl, by Comrie, by Binnick, by Smith and by
Verkuyl.  Verkuyl's book I exclude below, since the author, for his own
reasons, deliberately refrains from offering plain-language definitions
of the aspectual categories he recognizes.

Of the others, Comrie, of course, uses about the same definition of
`perfective' as that found in my dictionary.  Smith does the same.
Binnick offers no view of his own, but observes that Comrie's definition
is now the one most widely used.  He cites some examples, involving
wording like "indivisible situation" and "integral action".  Dahl does
not deny this observation, but confesses to dissatisfaction with this
definition, on the interesting ground that it is too restrictive: it
excludes certain forms traditionally called `perfective'.  But even Dahl
expressly denies the identification of `perfective' with `completive':
he regards the two as quite distinct, and he emphatically does not
equate the perfective with completion.

So, in place of the wide variation in terminology of 50 years ago, we
now have a near-consensus among those who have investigated aspect most
carefully.  The definition of `perfective' in my dictionary is the one
now most widely used, and I suggest that we should, for once, agree on
this definition, as a small step toward the goal of unifying our
terminology -- a goal which I trust is shared by Pat Ryan.

This near-consensus has developed in the last several decades.  Older
works often use different terms, but that is no excuse for choosing
*some* older term or definition and trying to continue it today.
Anyway, neither Pei's outdated dictionary nor general-purpose English
dictionaries like Merriam-Webster can reasonably be cited as authorities
on the present-day technical terminology of linguistics.  The use of
`perfective' cited in MW, in AHD, in the second edition of the OED, and
in other general dictionaries, is outdated and no longer in general use
among linguists.  The only rational response is to change the
dictionaries, not to insist heatedly that the definitions must be right
because they're in print.

By the way, I happen to be the consultant on general linguistic terms
for the next edition of the OED.  My job is to scrutinize the existing
and new entries written by the editors.  While most of the draft entries
I see are pretty good, I occasionally have to correct entries which are
inadequate or outdated.  And I can assure you that, when we get to
`perfective', I will certainly be advising an update of the entry in the
second edition, which is now woefully out of date.  And Pat Ryan will
then be able to consult the OED for the latest word on the use of
`perfective' in linguistics. ;-)

As for the etymology of `perfect', this is utterly irrelevant.
Committing the etymological fallacy -- insisting that words must mean
what their etyma meant -- is the most fundamental kind of error I can
think of.  For example, our linguistic term `sentence' derives from
Latin <sententia> `feeling, way of thinking, opinion', a derivative of
<sentire> `feel'.  Does Pat, or anyone, want to maintain that `sentence'
must mean `opinion' because its Latin source did?  No?  Then why try to
claim that our term `perfective' must mean the same as Latin
<perfectum>?  (And, by the way, the more direct English descendant of
<perfectum> is `perfect' -- which doesn't mean the same as `perfective',
but which also doesn't mean `complete'.)

As for Pei, my little aside was merely to show -- correctly -- that
Pei's work was not even held in high regard by the professional
linguists of his own day.  I myself enjoyed reading Pei's books when I
was a kid, but I wouldn't dream of treating them as reliable sources
today.  In his dictionary, Pei doubtless did his best to record the uses
current in his day, but his day was half a century ago, and things have
moved on.

An example.  All my students who have ever been taught any grammar have
been taught that English items like `my', `your' and `his' are
"possessive pronouns".  This is the traditional, and established,
analysis.  It is also dead wrong.  These things are not pronouns at all,
but determiners, as can be easily shown.  So, I carefully teach my
students that these things are determiners, and warn them that
traditional writings constantly get this wrong.  It makes no difference
that a thousand books, possibly including Pei's dictionary and
Merriam-Webster, call these things "pronouns".  The tradition is wrong
and must be corrected.

As for Forbes's grammar of Russian, well.  Forbes's account is not
entirely consistent with what is said by the specialists who have
examined Russian.  All of Comrie, Dahl and Binnick (at least) provide
accounts of Russian which are at variance with Forbes's account: all
agree that the Russian contrast is not merely one of completion versus
non-completion, but something rather more subtle.

Now, Pat accuses me of being less than candid in my dictionary.  Not
guilty.  Read the preface, and consult, for example, the entries for
`continuous' and `progressive'.

Quote from Comrie:

> "As already indicated, in discussions of aspect, as opposed to many
> other areas of linguistics, there is no generally accepted
> terminology".

Correct, of course, but the work of Comrie and of others has taken us
some way toward an accepted terminology.  Anyway, Pat, weren't you
telling us just the other day that there *was* an accepted sense of
`perfective'?

> I can interpret Comrie's Gothic definition myself: I interpret *his*
> "perfective" to mean: 'a verbal action characterized as a point in
> time'; and *his* "imperfective" to mean: 'a verbal action
> characterized as points in time'.

No; certainly not.  Comrie emphatically does not define `perfective' to
mean `punctual'.  He is entirely clear about this, and he defines
`punctual' in very different terms from `perfective' in chapter 2.

By the way, I use `punctual' in my dictionary because that is *by far*
the most usual term.  The alternative `momentary' (or `momentaneous') is
rare, and I did not judge it important enough to enter in my dictionary.
Look at the five books cited above: of these, only Dahl mentions
`momentaneous' at all, and then only in a quote from a particular writer
who uses it.

Pat also seems to be uncertain as to what I mean by `superordinate'.  By
this I mean that `perfective' and `imperfective' are broad labels which
can be subdivided into several more specific aspectual types.  For
example, all of the `habitual', the `iterative' and the `progressive'
constitute varieties of the imperfective.

> However illuminating these definitions might be thought to be
> ("completive = completion"; who could have a problem with that?),
> Larry's short definition for 'perfective', i.e. 'no internal
> structure', is simply incredible. There is no possible verbal action
> which does not have 'internal structure' *in this universe*!

A misunderstanding.  The issue is not whether an action has an internal
structure, but whether it is *linguistically presented* as having one.
We are talking about linguistic structure, not about the nature of the
non-linguistic universe.

In most European languages, `lightning' is a noun, while in many
American languages it's a verb.  The real world is the same, but the
linguistic facts are different.  And you can't claim that `lightning' is
not a noun because lightning isn't a thing, or because the equivalent
word in another language is a verb.

> The sentence 'Wallace Stevens lived in Hartford' is *not* perfective
> since, as Larry correctly notes, only telic activities can be
> completed or considered to be in the process of completion. What
> anyone who speaks English without wishing to impose a pre-defined
> and artificial interpretation on it would understand was that
> Wallace lived in Hartford for some indefinite or understood or
> immaterial period of time, which makes it 'durative'; and I would be
> willing to wager that his life there had some "internal structure".

It did, but this is not relevant, since it is not linguistically
expressed in my example, which *presents* Stevens's time in Hartford
without structure.

> Larry is, of course, free to assume that only his ideas have
> validity however unattractive that position may be. What he will
> never admit is that his choices (Comrie, etc.) are *choices*, and
> not glimpses of the Eternal Truth that has been denied all those of
> us who disagree with him and his *choices*.

Really?  I claim no eternal truth: I am merely reporting on the
contemporary usage of professional linguists.  It is you, Pat, who want
to claim that Pei and Forbes revealed the "Eternal Truth" decades ago,
and that we are wrong to deny it.

[responding to Lloyd Anderson]

> I think it is unpardonable for the possibility to exist that "what
> is CALLED 'Perfective' in the grammatical tradition of a particular
> language . . . may not bear any relation to what is apparently the
> same term intended in its universal sense."

Well, whatever you may think about it, that's the way the world is.
Lloyd is quite right.

> Of what conceivable value is it to take a term, 'perfective', which
> had, and apparently still continues to have a definition at odds
> which the "universal" one, and apply it to a new category?

This is not what has happened.  Rather, our use of `perfective' has
emerged from what was formerly a highly inconsistent set of usages
involving two independent traditions.

> 'Perfective' and 'imperfective' should be re-defined and restored to
> their traditional meanings.

But they have no established traditional meanings.  I advise you to read
Binnick's book, which provides an excellent summary of the historical
confusion and disagreement over aspect and its associated terminology.

> While Larry may scoff at Pei, dictionaries (after all, not written
> by linguists), I dare him to scoff at W. P. Lehmann, who, in his
> book on Proto-Indo-European synatx, *clearly* employs 'perfective'
> in the traditional sense, a category of IE which is indicated by the
> "perfect" endings.

This is not what Lehmann says.  Lehmann argues that PIE had a
`perfective' set of forms which, in the daughter languages, evolved into
other functions, including (in some daughters only) a perfect.  Lehmann
is using his terms in a way that seems good to him, but this is not
necessarily the way they would be used by anyone else.  As I have
emphasized, our use of aspectual terminology has been characterized by
colossal inconsistency.

But we now have a widely agreed sense of `perfective'.  So let's use it.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk