Understanding perfective aspect

[Travis B.]

In that case, how do these sorts of systems tend to be organised, and what sort of aspects do they tend to have instead of a perfective and/or imperfective?

I've seen a few that use perfect (retrospective), continuous/progressive, and prospective aspects. This is essentially a past, present, & future tense system, but because it's aspect it's relative to the time of the event being discussed, rather than relative to the time of speaking.

That seems similar to modern English's aspectual system, aside from that English also has a habitual and stative verbs. The ability to say things like "The farmer will have milked the cow" and "The farmer was going to milk the cow" demonstrates this.

[bradrn]

In that case, how do these sorts of systems tend to be organised, and what sort of aspects do they tend to have instead of a perfective and/or imperfective?

I've seen a few that use perfect (retrospective), continuous/progressive, and prospective aspects. This is essentially a past, present, & future tense system, but because it's aspect it's relative to the time of the event being discussed, rather than relative to the time of speaking.

That seems similar to modern English's aspectual system, aside from that English also has a habitual and stative verbs. The ability to say things like "The farmer will have milked the cow" and "The farmer was going to milk the cow" demonstrates this.

Well, English does have both the perfect and the prospective, but the ‘core’ of the English aspectual system (so to speak) is still a perfective/imperfective distinction. It is certainly tempting to analyse this as a three-way perfect/progressive/prospective system — especially since the perfective is unmarked — but I don’t think that’s right, as the perfect and prospective can both co-occur with the perfective and imperfective. I think a better way to analyse English is as having two separate aspectual systems which are orthogonal to each other, one being perfective/imperfective and the other being perfect/none/prospective:

	perfective	imperfective
perfect	have seen	have been seeing
none	see	is seeing
prospective	going to see	going to be seeing

[Vardelm]

That’s a pretty interesting system!

It is, and I'm using for my current project!

The Wikipedia page on relative tense might be of interest regard this system.

But I don’t think that’s quite what I’m looking for — to me, that seems like more of a variation on a tense system than an aspectual system per se. Sure, it’s not actually tense, but I don’t think calling it aspect is quite right either.

Yep, I agree. From what I gather, there is mixed opinion on whether it's aspect or tense. It shows that the area between the 2 categories can become quite muddied.

I don't recall seeing any other complete, aspectual systems, really. The only thing I can think of is that you have a language that doesn't mark for absolute tense, relative, tense, or perfective/imperfective at all. It would deal with time simply using adverbs & adjuncts. It might have a few of the more "miscellaneous" aspects thrown in there like an inchoative & terminative.

[bradrn]

I don't recall seeing any other complete, aspectual systems, really. The only thing I can think of is that you have a language that doesn't mark for absolute tense, relative, tense, or perfective/imperfective at all. It would deal with time simply using adverbs & adjuncts. It might have a few of the more "miscellaneous" aspects thrown in there like an inchoative & terminative.

I know that some languages (e.g. Northeast Ambae) have a system where tense, aspect and mood all get combined into one set of particles (or one set of TAM particles + several sets of optional aspectual modifiers, in the case of NE Ambae); such systems don’t necessarily operate along perfective/imperfective lines. But that’s already getting pretty far from a purely aspectual system; perhaps there exist no aspectual systems which have a minimum of tense or mood marking but don’t have a fundamental perfective/imperfective distinction.

[Vardelm]

perhaps there exist no aspectual systems which have a minimum of tense or mood marking but don’t have a fundamental perfective/imperfective distinction.

It seems like probably not, at least to the extent that you're thinking about it.

Perhaps take a look at Malay grammar for a system that has no tense/aspect marking and consider what you want to add onto that?

[Kuchigakatai]

1. We have already established that there exist semantically ‘prototypical’ perfective and imperfective events, which always get marked with the perfective and perfective aspects. But does anyone have any more information about how languages assign aspect to non-prototypical events? (e.g. the stative always gets assigned to the unmarked aspect, if there is one; I’m particularly interested in similar generalisations.)

No idea about generalizations like you want, but one thing that happens is using the form normally associated with perfective aspect as a perfective change-of-state form when the verb doesn't have the normal distinction. While it is true Mandarin 是 shì 'to be' cannot take any such aspect marking, 有 yǒu 'to have' can mean 'obtained sth' with the perfective particle (or clitic if you prefer):

她很有名气。
3SG very have famous.air
(lit., "She very much has famous air")
'She is / was / is going to be very famous.'

她有了名气，甚至还上了电视。
tā yǒu-le míngqì, shénzhì hái shàng-le diànshì
3SG have-PFV famous.air, even still be.on-PFV television
(lit., "She obtained famous air, [and] even went on TV")
'She became famous, and even appeared on TV.' (where "became" is a perfective, not an expression of gradual change)

他们有发现。
tāmen yǒu fāxiàn
3PL have discover
'They make / are making / were making / will be making discoveries.'

试验新药时，他们有了自己的发现。
shìyàn xīnyào shí, tāmen yǒu-le zìjǐ de fāxiàn
experiment new.med when, 3PL have-PFV self GEN discover
(lit., "When experimenting [with] new medicine, they obtained their own discovery")
'Testing new medicine, they made their discovery.'

[evmdbm]

[In fact, boundedness I would say is the number one characteristic of perfective. Bounded and unbounded might almost be better terms than perfective and imperfective. Perfective/bounded means the event in question is conceptualized as having a beginning, middle and end – i.e. bounded in time, a whole event. Imperfective/unbounded means you're not thinking of it that way. You're thinking of it as a) part of an event, something ongoing, most likely a part of the middle of the event, perhaps while something else was happening; b) as a state, a description of a general truth/state; or c) as something habitual, something that happens again and again. With the habitual sense, it's really a sequence of perfective events, but the focus isn't on them as bounded events but the fact that it keeps happening, almost like a segmented stative, so again, a description of a general truth/state.

How do you square this with the gnomic aorist of Ancient Greek?

I have seen it suggested that the aorist is perfective in aspect (https://en.wikipedia.org/wiki/Aorist). That makes sense to me because it refers to a single discreet act and so imperfective/perfective cuts across the imperfect/aorist/perfect distinction in Greek. Even the gnomic aorist is said to be perfective. I wonder whether that's because what we're describing is the habitual state as a simple fact, much like the simple fact of the action (I walked). And the aorist also gets explained as describing a simple action, as opposed to the imperfect - ongoing action, or perfect - completed action.

Maybe we're better seeing the habitual imperfective not as a statement of general truth, but as focussing on the ongoing nature of a particular event/action?

[hwhatting]

If I shouldn’t think of was as "perfective" and was being as "imperfective", then how else am I supposed to think about those?

Well, English does have both the perfect and the prospective, but the ‘core’ of the English aspectual system (so to speak) is still a perfective/imperfective distinction. It is certainly tempting to analyse this as a three-way perfect/progressive/prospective system — especially since the perfective is unmarked — but I don’t think that’s right, as the perfect and prospective can both co-occur with the perfective and imperfective. I think a better way to analyse English is as having two separate aspectual systems which are orthogonal to each other, one being perfective/imperfective and the other being perfect/none/prospective:

	perfective	imperfective
perfect	have seen	have been seeing
none	see	is seeing
prospective	going to see	going to be seeing

I don't know why nobody has commented on this - even, several commenters seem to accept the premise. But I'd rather say that it's not true that English has a basic perfective-imperfective distinction. It distinguishes a continuous from a simple form; the uses of the continuous form are indeed generally imperfective, but the simple form includes uses that are imperfective, like e.g. habitual actions or general truths.

Does this mean that, in a language where perfective is marked (like my conlang), stative verbs will usually get the imperfective rather than the perfective?

That's at least also true for Slavic languages like Russian or Polish. In these languages, you can form perfective verbs based on the imperfective stative verbs if you need them (e.g. Russian lyubit' "love" - polyubit' "fall in love, love for some period").

In particular, a semantically perfective event can only be in the present tense if it is punctual; this is because a present non-punctual event must extend into the past and future, and so cannot be seen as a single whole while staying entirely inside the present.

Well, at least in Russian, perfective events cannot be expressed as being in the present at all, even if they are punctual. What is formally (and historically) the present tense of perfective verbs designates either the future or is used for gnomic utterances about expected results.

[*] Languages generally contrast ‘perfective’ and ‘imperfective’ forms of verbs.

Many languages do, but many also don't. Standard German, for one, doesn't.

[bradrn]

I don't know why nobody has commented on this - even, several commenters seem to accept the premise. But I'd rather say that it's not true that English has a basic perfective-imperfective distinction. It distinguishes a continuous from a simple form; the uses of the continuous form are indeed generally imperfective, but the simple form includes uses that are imperfective, like e.g. habitual actions or general truths.

My thinking has changed somewhat since I wrote that comment. I now consider English to have a basic distinction between simple and progressive aspects; ‘simple’ is a better name than ‘perfective’ due to the highly marked nature of the English progressive. (‘Progressive’ is a better name than ‘continuous’ as the latter is usually considered to include stative usages.)

Does this mean that, in a language where perfective is marked (like my conlang), stative verbs will usually get the imperfective rather than the perfective?

That's at least also true for Slavic languages like Russian or Polish. In these languages, you can form perfective verbs based on the imperfective stative verbs if you need them (e.g. Russian lyubit' "love" - polyubit' "fall in love, love for some period").

In particular, a semantically perfective event can only be in the present tense if it is punctual; this is because a present non-punctual event must extend into the past and future, and so cannot be seen as a single whole while staying entirely inside the present.

Well, at least in Russian, perfective events cannot be expressed as being in the present at all, even if they are punctual. What is formally (and historically) the present tense of perfective verbs designates either the future or is used for gnomic utterances about expected results.

Indeed, I have learned that these are key reasons why the Slavic imperfective is considered to be unmarked relative to the perfective.

[*] Languages generally contrast ‘perfective’ and ‘imperfective’ forms of verbs.

Many languages do, but many also don't. Standard German, for one, doesn't.

‘Generally’ does not mean ‘all’. I meant that most languages with a morphological category of aspect tend to make a primary distinction between a more perfective aspect and a more imperfective aspect, especially those with a binary or tripartite system. As far as I can tell, Standard German has no morphological category of aspect at all.

[hwhatting]

‘Generally’ does not mean ‘all’. I meant that most languages with a morphological category of aspect tend to make a primary distinction between a more perfective aspect and a more imperfective aspect, especially those with a binary or tripartite system. As far as I can tell, Standard German has no morphological category of aspect at all.

That's closer to my understanding.
There is an additional point that in the Slavicist tradition, from which I come, "aspect" is only used for the perfect-imperfect distinction; everything else that gets named "aspect" by other traditions is called "Aktionsart". So in that framework, your statement "most languages with a morphological category of aspect tend to make a primary distinction between a more perfective aspect and a more imperfective aspect" doesn't even make sense, because there is no other kind of aspect.

[bradrn]

‘Generally’ does not mean ‘all’. I meant that most languages with a morphological category of aspect tend to make a primary distinction between a more perfective aspect and a more imperfective aspect, especially those with a binary or tripartite system. As far as I can tell, Standard German has no morphological category of aspect at all.

That's closer to my understanding.
There is an additional point that in the Slavicist tradition, from which I come, "aspect" is only used for the perfect-imperfect distinction; everything else that gets named "aspect" by other traditions is called "Aktionsart". So in that framework, your statement "most languages with a morphological category of aspect tend to make a primary distinction between a more perfective aspect and a more imperfective aspect" doesn't even make sense, because there is no other kind of aspect.

Well, actually, I can’t say I disagree with this statement too much! The term ‘aspect’ lumps together a whole bunch of different categories, and it’s only because languages tend to treat them in the same system that we get away with it. There is of course the usual perfective—imperfective continuum (which basically comes down to boundedness), but there’s also perfects and prospectives (relative tense), inchoatives and terminatives (phase), and various forms of pluractionality (internal structure), amongst other categories. In principle these are all orthogonal; in practice, languages tend to conflate them to various degrees.

[evmdbm]

Can I ask a related, but tangential question? Telicity is often defined in terms of presenting an action as being complete, and perfective verbs carry the meaning of completeness (or boundedness). I'm struggling to see the difference - although there obviously is one. Telicity can be marked by differential object marking, ie accusative for telic actions and partitive for atelic ones in Finnish, but you would never mark perfectivity on a noun.

Can someone give me some help here?

[bradrn]

Can I ask a related, but tangential question? Telicity is often defined in terms of presenting an action as being complete, and perfective verbs carry the meaning of completeness (or boundedness). I'm struggling to see the difference - although there obviously is one. Telicity can be marked by differential object marking, ie accusative for telic actions and partitive for atelic ones in Finnish, but you would never mark perfectivity on a noun.

Can someone give me some help here?

Telicity is aktionsart — an inherent property of an action. All actions are either atelic or telic, and changing either action to the other type is basically a derivational process: e.g. English atelic eat vs telic eat up, or German atelic kämpfen ‘fight’ vs telic erkämpfen ‘achieve by means of a fight’ (examples from Comrie).

By contrast, perfectivity is aspect — how we see an action. The sorts of events which are perfective or imperfective differ from language to language, though one can find certain prototypical events which are perfective in most languages. Generally speaking, the prototypical perfective event is an action or negligible (or neglected) duration resulting in a change of state, whereas the prototypical imperfective event is one seen as a process which does not necessarily have an end. Different languages have different relations between telicity and perfectivity: in some languages (especially Slavic), only telic events may take the perfective, whereas in others (e.g. English) even atelic events may be marked as perfective.

EDIT: Possibly an example might help. Here are all four possible combinations of telicity and perfectivity (well, progressivity) in English:

1. (nonprogressive, telic) I ate up the food
2. (progressive, telic) I am eating up the food
3. (nonprogressive, atelic) I ate the food
4. (progressive, atelic) I am eating the food

Comparing (1) and (2), the most natural sentence for me is (1) certainly. (2) focusses on the eating process itself, but implies that I finished all the food — I am eating up the food, but I don’t think I’ll finish it sounds downright odd. The English ‘perfective’ has a pretty wide range of use, which is why I called it the ‘nonprogressive’ above (emphasising its unmarked nature compared to the progressive), so (3) sounds fine, but in a language with a more typical perfective/imperfective distinction, it would sound as odd as (2). Finally, (4) is of course fine.

[Ephraim]

Can I ask a related, but tangential question? Telicity is often defined in terms of presenting an action as being complete, and perfective verbs carry the meaning of completeness (or boundedness). I'm struggling to see the difference - although there obviously is one. Telicity can be marked by differential object marking, ie accusative for telic actions and partitive for atelic ones in Finnish, but you would never mark perfectivity on a noun.

Can someone give me some help here?

Telicity is aktionsart — an inherent property of an action. All actions are either atelic or telic, and changing either action to the other type is basically a derivational process: e.g. English atelic eat vs telic eat up, or German atelic kämpfen ‘fight’ vs telic erkämpfen ‘achieve by means of a fight’ (examples from Comrie).

By contrast, perfectivity is aspect — how we see an action. The sorts of events which are perfective or imperfective differ from language to language, though one can find certain prototypical events which are perfective in most languages. Generally speaking, the prototypical perfective event is an action or negligible (or neglected) duration resulting in a change of state, whereas the prototypical imperfective event is one seen as a process which does not necessarily have an end. Different languages have different relations between telicity and perfectivity: in some languages (especially Slavic), only telic events may take the perfective, whereas in others (e.g. English) even atelic events may be marked as perfective.

I may disagree with some parts of your post.

From my understanding of the literature, there is actually little agreement about how to define telicity (and not all authors would use the term). However, there seems to a consensus that a distinction should be made between examples such as:
5. I built a house. (accomplishment according to Vendler, i.e. durative and telic)
6. I walked. (activity according to Vendler, i.e. durative and atelic)

Typically, telic predicates are thought of as being materially or inherently bounded or having some sort of inherent or natural endpoint, result or conclusion, while atelic predicates don't. Importantly, the fact that there is a natural endpoint to an action doesn't mean that the endpoint has been reached or that it will ever be reached.

Punctual predicates are much more problematic. Some distinguish between punctual atelic predicates (often called semelfactives) and punctual telic predicates (achievements), but this distinction only makes sense according to some definitions of telicity (perhaps involving some type of event-external change-of-state). According to some other definitions, punctual predicates are necessarily telic. Of course, not everyone agree on the punctual–durative distinction either.

It's an interesting question what exactly telicity (or Aktionsart in general) is an inherent property of. I've often seen it described as a property of the predicate (although not always with a good definition of the term predicate), but it may also be described as a property of the "scenario", "eventuality", "event", "situation", "action", "action phrase" or the like (I use the term predicate here for simplicity). It is sometimes said to be a property of the verb, which is a bit unfortunate (as is the term lexical aspect).

Now, the first part that I would disagree with is the statement that changing an atelic action to a telic one is basically a derivational process. Although languages may have derivational marking (or other types of marking) that correlate with telicity, atelic predicates can often be made telic simply by adding some sort of bounded object or adverbial, or vice versa. See the following examples from English:
7. I drank some wine. (no bounded quantity of wine => no specific endpoint for the action => atelic)
8. I drank a glass of wine. (bounded quantity of wine => action finished when the glass is finished => telic)
9. I drank myself to sleep. (specific change-of-state => action finished when result is reached => telic)

EDIT: Possibly an example might help. Here are all four possible combinations of telicity and perfectivity (well, progressivity) in English:

1. (nonprogressive, telic) I ate up the food
2. (progressive, telic) I am eating up the food
3. (nonprogressive, atelic) I ate the food
4. (progressive, atelic) I am eating the food

Comparing (1) and (2), the most natural sentence for me is (1) certainly. (2) focusses on the eating process itself, but implies that I finished all the food — I am eating up the food, but I don’t think I’ll finish it sounds downright odd. The English ‘perfective’ has a pretty wide range of use, which is why I called it the ‘nonprogressive’ above (emphasising its unmarked nature compared to the progressive), so (3) sounds fine, but in a language with a more typical perfective/imperfective distinction, it would sound as odd as (2). Finally, (4) is of course fine.

I may also slightly disagree with your examples in that I don't think 3 and 4 are necessarily atelic. I'm not a native English speakers so my intuitions here may be a bit off, but I think 3 and 4 can have both telic and atelic readings.

A common test for telicity in English is to use for- and in-time adverbials, i.e. "for an hour" and "in an hour":
10. I ran *in/for an hour. (atelic)
11. I ran ten kilometres in/*for an hour. (telic)

I think your example 3 might work with both types of time adverbials, but probably with a difference in meaning:
3a. "I ate the food for an hour."
3b. "I ate the food in an hour."

My intuition is that "the food" in 3a probably refers to a specific type or class of food (like "the food at this restaurant"), but not to any bounded quantity of food, while 3b probably refers to a specific meal or to a quantity of food that's bounded in some other way.

Since 4 is present progressive, it might be awkward to use duration time adverbials.

[bradrn]

From my understanding of the literature, there is actually little agreement about how to define telicity (and not all authors would use the term). However, there seems to a consensus that a distinction should be made between examples such as:
5. I built a house. (accomplishment according to Vendler, i.e. durative and telic)
6. I walked. (activity according to Vendler, i.e. durative and atelic)

Typically, telic predicates are thought of as being materially or inherently bounded or having some sort of inherent or natural endpoint, result or conclusion, while atelic predicates don't. Importantly, the fact that there is a natural endpoint to an action doesn't mean that the endpoint has been reached or that it will ever be reached.

Punctual predicates are much more problematic. Some distinguish between punctual atelic predicates (often called semelfactives) and punctual telic predicates (achievements), but this distinction only makes sense according to some definitions of telicity (perhaps involving some type of event-external change-of-state). According to some other definitions, punctual predicates are necessarily telic. Of course, not everyone agree on the punctual–durative distinction either.

It's an interesting question what exactly telicity (or Aktionsart in general) is an inherent property of. I've often seen it described as a property of the predicate (although not always with a good definition of the term predicate), but it may also be described as a property of the "scenario", "eventuality", "event", "situation", "action", "action phrase" or the like (I use the term predicate here for simplicity). It is sometimes said to be a property of the verb, which is a bit unfortunate (as is the term lexical aspect).

I agree with all this. I personally tend towards the view that punctual predicates must necessarily be telic.

Now, the first part that I would disagree with is the statement that changing an atelic action to a telic one is basically a derivational process. Although languages may have derivational marking (or other types of marking) that correlate with telicity, atelic predicates can often be made telic simply by adding some sort of bounded object or adverbial, or vice versa. See the following examples from English:
7. I drank some wine. (no bounded quantity of wine => no specific endpoint for the action => atelic)
8. I drank a glass of wine. (bounded quantity of wine => action finished when the glass is finished => telic)
9. I drank myself to sleep. (specific change-of-state => action finished when result is reached => telic)

…

I may also slightly disagree with your examples in that I don't think 3 and 4 are necessarily atelic. I'm not a native English speakers so my intuitions here may be a bit off, but I think 3 and 4 can have both telic and atelic readings.

A common test for telicity in English is to use for- and in-time adverbials, i.e. "for an hour" and "in an hour":
10. I ran *in/for an hour. (atelic)
11. I ran ten kilometres in/*for an hour. (telic)

I think your example 3 might work with both types of time adverbials, but probably with a difference in meaning:
3a. "I ate the food for an hour."
3b. "I ate the food in an hour."

My intuition is that "the food" in 3a probably refers to a specific type or class of food (like "the food at this restaurant"), but not to any bounded quantity of food, while 3b probably refers to a specific meal or to a quantity of food that's bounded in some other way.

Since 4 is present progressive, it might be awkward to use duration time adverbials.

You are of course completely correct here. (The perils of writing technical posts at 1am…) In particular, the fact that 3–4 are unspecified for telicity is a good explanation for why both of them sound natural, whereas 2 sounds far less natural than 1.

[Moose-tache]

It's telling that languages frequently have regular morphological means to differentiate perfective and imperfective versions of otherwise identical verbs, but regular morphological means to differentiate telic and atelic versions of the same verb are quite a bit less common. As already noted, the telicity of a verb may be lexically inherent, or it may be specified by other phrases in the clause. The same verb may be telic or atelic simply depending on how you parse a sentence. Consider the sentence "I will only knock once." It could mean "I will make one quick, loud noise on the door (telic)," or "I will not come knocking on your door a second time (here the knocking may be done any number of times without completion, thus atelic)." The hallmark of aktionsart is that it describes meaning, not form. In other words, we don't care how the language grammatically codes information, only how the events are literally organized in time. A language with no TAM morphology or syntax at all could still be analyzed for aktionsart. Perfectivity, on the other hand, and all aspect, is about how the language actually tells us about time and temporal shape. This has the consequence that the same term might not mean the same thing in every language. "Telic" is "telic," but "progressive" in Spanish is not "progressive" in English. Languages also get to decide how many aspectual contrasts to make, while no language gets to say whether doing something until its innate conclusion is telic or not.

To put it succinctly: If you ask whether or not the sentence "John hit the bear" is telic, you're asking me if John's action had an innate conclusion. if you ask whether or not the sentence "John hit the bear" is perfective, you're asking me if this English sentence is telling us that John's action had an innate conclusion.

[hwhatting]

Very well said.

[evmdbm]

Oh this helps a lot. Thanks. I should find the paper by Vendler that Ephrain alludes to.

[zompist]

Languages also get to decide how many aspectual contrasts to make, while no language gets to say whether doing something until its innate conclusion is telic or not.

That's probably 99% true, but it worries me-- absolute statements about language universals usually turn out to be wrong. I'm highly dubious that there is such a thing as universal semantics.

You and others have given examples where the exact same action is described as telic or atelic, which suggests at the least that telicity involves the speaker taking a viewpoint. This can easily be conditioned by the language involved.

E.g. English often marks telicity with "up"-- "I ate up the candy", "I tore up the ticket", "I read up on Lacan", "I wrote up the bug", "I thought up a use case.". I think the Mandarin object marker ba generally marks telicity-- it often indicates that something is taken care of. Both the English and Mandarin markers are optional; I don't know that that's so different from (say) choosing imperfectives, in languages where this is optional. It's also easy enough to imagine languages that don't provide a marker like this. (We can't use "up" with every verb.)

[Vardelm]

To put it succinctly: If you ask whether or not the sentence "John hit the bear" is telic, you're asking me if John's action had an innate conclusion. if you ask whether or not the sentence "John hit the bear" is perfective, you're asking me if this English sentence is telling us that John's action had an innate conclusion.

I agree as well; this is nicely put.

3. (nonprogressive, atelic) I ate the food
4. (progressive, atelic) I am eating the food

(3) sounds fine, but in a language with a more typical perfective/imperfective distinction, it would sound as odd as (2). Finally, (4) is of course fine.

My intuition is that "the food" in 3a probably refers to a specific type or class of food (like "the food at this restaurant"), but not to any bounded quantity of food, while 3b probably refers to a specific meal or to a quantity of food that's bounded in some other way.

My interpretation is that #3 would be telic. Compare:

A) I ate food.
B) I ate the food.

I think most of the time, B would be seen as telic: the speaker ate food and finished it; that is, they ate all of the food they are speaking about.

With that said, I think there is room for atelic interpretation like Ephraim mentions with a type/class of food, such as at a restaurant. In that case, there is a reasonable assumption that there is so much food that there is no way anyone could eat all of it. Here, "the" indicates a class, rather than a specific amount of food. Interestingly, a friend posted an article on Facebook this morning about the word "the" which is relevant here.
https://www.mentalfloss.com/article/535 ... -does-mean

It's an interesting question what exactly telicity (or Aktionsart in general) is an inherent property of. I've often seen it described as a property of the predicate (although not always with a good definition of the term predicate), but it may also be described as a property of the "scenario", "eventuality", "event", "situation", "action", "action phrase" or the like (I use the term predicate here for simplicity). It is sometimes said to be a property of the verb, which is a bit unfortunate (as is the term lexical aspect).

Given the differences in telicity when adding adverbials or even just "the", I very much agree with this. I don't see telicity as a lexical property of a verb, but of a phrase as a whole.

I don't know if other conlangers had the same thought, but I had understood that Finnish used the partitive or accusative cases for objects because a phrase was telic/atelic. Instead, it seems that the phrase would telic/atelic due to which one was used, much in the same way that adding the definite article or time adverbials changes telicity in English. This, plus the understanding of telicity rising from an entire phrase rather than just from a given verb and the form used made me much more comfortable with the concept (even though I still struggle with it a bit!).

You and others have given examples where the exact same action is described as telic or atelic, which suggests at the least that telicity involves the speaker taking a viewpoint. This can easily be conditioned by the language involved.

Interesting thought, and I think I would agree on the viewpoint. Back to eating food in a restaurant, is the speaker talking about all of the food at the restaurant or just the food they were given? If it's not a restaurant, are they even aware there was more food around?