On 13 February 2014 12:01, R A Brown <[log in to unmask]> wrote:

> On re=reading Comrie on "telic and atelic" (op.cit. pages 44
> to 48), it seems that this notion requires real world
> knowledge and is not a purely grammatical concept.

Actually, the Wikipedia article about telicity ( criticises this point of view,
ending with the words: "This type of exercise can serve as a reminder of
the futility of trying to link linguistic semantics to the real world
without considering the intermediary agent of human cognition."

>  For
> example, he says in his main text "... the sentence _John is
> singing a song_ describes a telic situation, since this
> situation has a well-defined terminal point, namely when
> John comes to the end of the song in question."   But then
> we find in a footnote: "Actually it is difficult to think of
> a sentence which must be interpreted as describing a telic
> situation.  Thus if John were to repeat the same song
> incessantly, then even _John is singing a song_ would be
> atelic ...."
The problem is that Comrie is trying to refer to real world events *on
their own*, while we should be looking at real world events *as they are
conceptualised by the speaker*. The distinction is not trivial. If you look
at real life events in general, naturally everything has an end. Even the
universe itself may end in a Big Crunch. Does that mean atelic verbs
phrases don't actually exist? Of course not, because even if no event
happens indefinitely, we, as human beings, can *conceptualise* and describe
events as carrying on indefinitely, if only because the end of such event
is irrelevant to what we are talking about.

Once you have that in mind, it's not an issue to describe telicity as a
purely linguistic concept, because that's what it is.

Once again, I think the "quantisation and cumulative reference" definition
hits the mark pretty well when it comes to telicity. Look at your example:
"John is singing a song". Imagine he does so for 4 minutes. So we can cut
this 4-minute event into 2 2-minute subevents. Can those two events be
accurately described as "John is singing a song"? Yes of course! In that
case, the expression "John is singing a song" has cumulative reference,
i.e. it is atelic.
On the other hand, "John sang a song" is telic, even if it refers to
exactly the same event. That's because the way the event has been
conceptualised here is different. Once again, imagine that John sang a
4-minute song. If we cut this event into two, can those two events be
accurately described as "John sang a song"? No, they can't. The first
subevent, in particular, ends before John has finished singing, and thus
cannot accurately be described by the phrase "John sang a song". The
expression "John sang a song" doesn't have cumulative reference. Rather, it
is *quantised*. And thus telic.

As you can see, the same event can be described using an atelic or a telic
expression. It just depends on how the speaker conceptualises and describes
the event.

Telicity is a property of *expressions*, not of events. So of course trying
to define telicity using real world event descriptions is bound to fail.

>  or have any examples of natlangs that do something
>>> similar?
> Examples of split subject system where the choice between
> Nom/Acc and Erg/Abs is determined by aspect can be found
> (with checking, I believe something along these lines
> happens in Georgian - must look out my notes later today),
> but I'm not aware of such systems being due to telicity.
Telicity is often confused with perfectness, and indeed those two have a
lot in common. But they are different enough that they can be combined
(indeed, in English "I have written a book" is perfect and telic, while "I
have been writing a book" is perfect but usually atelic).

> As one of the objects of conlanging, at least for some of
> us, is to experiment with ideas, why not just tryout
> Scenario A and Scenario B and see what happens.  Even if, in
> the end, an experiment proves unsuccessful, it does IMO give
> some insight into the way language works.
True, but only if one is using a useful definition of telicity. If the
definition is unclear or incorrect, whatever happens will not give us much
Christophe Grandsire-Koevoets
President of the Language Creation Society (

Personal Website: