Tag Archives: semantics

Dowty’s Aspectual Verb Classes & the Infamous Semelfactive Controversy

Enjoy this entry by guest contributor XX — MLR.

In the study of Semantics, aspectual verb classification attempts to identify the internal temporal constituency of verbs used in varying situations. This basically means that we divide verbs into what are (arguably, as we will soon find out) four main classes, depending on the semantic meaning associated with them when used in sentences that represent different situations in the world. These, for the most part, undisputed classes are (1) states, (2) activities, (3) achievements, and (4) accomplishments.

States Activities Achievements Accomplishments
No external change External change & homogenous description, & have an agent External change & homogenous description, & no agent External change, & heterogeneous description, & have an agent
know, believe, have, desire, love, understand run, eat, walk, swim, push a cart, drive a car notice, fall asleep, recognize, find, reach, die, lose draw a circle, build a house, paint a picture, recover from illness

States are verbs that do not reflect external change, such as know or believe. Activities include verbs that do have external change, have a homogenous description (there is only a process with no endpoint in the action), and have an agent, such as run or eat. If you say “Susan eats cheeseburgers,” there is no requirement in the sentence that there is some point in which Susan stops eating cheeseburgers. Achievements reflect external change and are homogeneous (but this time have an endpoint with no process) and do not have an agent. Achievements, according to Dowty, include notice, fall asleep, receive a letter, break, knock, and shatter. In “Susan notices the stop sign,” we have an instantaneous action of noticing, with no requirement that there was some kind of process involved before and after Susan noticed. Finally, accomplishments reflect external change and are heterogeneous (so they include both a process and an endpoint) and have an agent. Accomplishments are verbs like draw a circle, and build a house. In both of those examples we can imagine that there is a process and an endpoint: Susan has a period of time in which she is drawing the circle, and an endpoint in which the circle has been drawn.

While these four aspectual verb classes are for the most part undisputed among semanticists, some have proposed that there are a few remaining verbs that don’t seem to fit into these categories. The proposed fifth aspectual class called semelfactives may be necessary to cover such verbs as break, knock, rap, tap, and shatter. To better illustrate why these verbs seem to be special, I will now run them through Dowty’s diagnostic tests he has devised to determine aspectual class for verbs in English.

It’s important to note that these particular verbs are argued to be achievements by those who do not think a fifth class is necessary, including Dowty himself. Both semelfactives and achievements describe an instantaneous event with no endpoint, but it seems that they differ in that semelfactives can sometimes have an agent. The following three diagnostic tests shown below check to see if the verb in question has an agent:

[1] John persuaded Mary to break the promise.

[2] Sam forced Sally to knock on the door

[3] ? Julia persuaded the window to shatter.

The question mark in [3] signifies that the sentence is unacceptable on a semantic basis. In other words, there is nothing wrong with the structure/syntax of the sentence, the individual meanings of the words put together don’t make sense. This first test shows us that while semelfactives do not always have an agent, they differ from achievements that they can, depending on the sentence. For example, [3] would be acceptable if changed to ‘Julia persuaded Bill to shatter the window’, just as ‘John persuaded the promise to break’ no longer remains acceptable.

[4] Knock on the door!

[5] ? Notice Sally!

[6] ? Explode the house!

[7] Tap on the window!

Putting the verb into an imperative construction is another way to test if there is an agent. Again, we can see that, depending on the sentence and the situation, semelfactives can sometimes have an agent. If we changed [5] to ‘Notice how beautiful Sally looks today’, it turns into an acceptable sentence.

[8] Sam carefully broke the candy bar.

[9] Alex deliberately shattered the glass statue.

[10] ? Mary deliberately noticed the painting.

[11] ? James deliberately fell asleep.

This last test (adding the word carefully/deliberately) shows that the undisputed achievement verbs notice and fall asleep can still be proven to be just those – achievements, while verbs like break, shatter, tap, knock, etc. are shown to have agents in these sentences. Since it does not make semantic sense to be able to say someone deliberately noticed something, we can see that there is no hint of sentience (which would otherwise provide for proof of an agent) in the verb.

[12] Julia is breaking the rules.

[13] Sam is tapping his fingers on the desk.

[14] ? Mary is knowing the information.

Those arguing that semelfactives cannot fit into any of the other three aspectual classes (states, activities, and accomplishments) use this next diagnostic test to prove that they are not states. Example [12]-[14] show that these verbs work in the progressive form, and so they reflect external change.

[15] ? Sally shattered the window for an hour.

[16] ? Ben broke the lamp for an hour.

Checking to see if these verbs are accomplishments or activities is also quite simple and painless, as we just need to find if there is a process involved in the action. Placing them in the sentences above, we can see that their unacceptability shows us that these verbs represent an instantaneous action, involving no or very little process.

So, what have we learned today from this entry? Even as I was in the process of writing this out, I felt a certain amount of un-satisfaction. What is fifty percent interesting, fifty percent frustrating about the field of semantics in general is that studying meaning and acceptability is a total gray area of relativity. In my opinion, this is the major reason linguists have not been able to come to an answer about whether or not these verbs need a fifth aspectual class. Is it really unacceptable to say that Ben broke the lamp for an hour? Is it acceptable to say that Sally shattered the glass? Or can it only be that glass shatters after Sally hits it with the baseball? I seem to remember that even my undergraduate classmates had a hard time agreeing on what ‘sounded okay’ to them. So, the unfortunate circumstance to the end of this entry is that I don’t have an answer for this issue. Hopefully, however, this can help ignite some thoughtful discussion (or at the very least thoughtful thinking) about aspect, semantics, and beyond. :]



Leave a comment

Filed under Linguistics