These lessons introduce you to verbs and their core arguments. They build on the grammar of words and sentences.

  1. Verbs & nouns revisited
  2. Valency & transitivity
  3. Thematic roles & case
  4. Argument alignment
  5. Answers to the exercises

Verbs & nouns revisited

Verbs as functions

I've called verbs "action words", as if they were simply ways of expressing activities that take place. This misses one of the key features of verbs in many languages, namely, that they are an indispensable part of a sentence. You might even say that they hold the sentence together.

To get a better sense of how verbs work, let's agree to refer back to the example I gave it to him. As we progress, this touchstone will keep my explanations focused and give us something to check against.

I just claimed that a verb holds a sentence together, but I've given you no evidence to favor that notion. Let me illustrate my point with a sort of word game. Notice what happens when we swap out the noun phrases in I gave it to him.

English I gave it to him.
John gave it to him.
John gave the ball to him.
John gave the ball to the dog.

We could try this little exercise over and over. We might come up with strange examples like the cat gave Santa to the milk. Still, our sentences will make grammatical sense. The meanings of the words might be silly in context, but their functions remain crystal clear.

Let's try an alternative swap. Instead, let's start switching out the verb.

English I gave it to him.
*I stood it to him.
*I liked it to him.
*I went it to him.

Curious, isn't it? This comparison suggests a fundamental asymmetry between the verb and the other material in a sentence. The verb is doing something that's in a way more central than those nouns. How do we describe this? Instead of defining the verb as a word that talks about an action, we can move to describing it as the essential function word in a sentence. In fact, it almost seems to lay out the central structure of the sentence that we expect to find used, including the number and type of noun phrases that will get used. Rather than just an action word, it looks like the main verb may be the key function word in a clause.

Nouns as arguments

Given the results of our word game above, how do we look at nouns? It looks like noun use is in some way dicated by the verbs.

For now, let's say that verbs are not only function words, but that they make decisions about the nouns that will accompany them. Specifically, it looks as if verbs decide how many nouns they expect. Other evidence suggests that verbs decide what type of nouns they allow:

English I gave it to him.
*I gave she to him.
*Me gave it to him.

It's clear that verbs expect a specific form of a noun/pronoun in a specific "position". But it's just as clear that verbs don't expect specific nouns (remember that we can interchange nouns in the above sentence without disturbing its grammar). We'll call those specific positions that the verb expects to fill slots, and the nouns that fill those slots will be the verb's arguments.

In the examples with went and gave, the verbs clearly have certain arguments they expect. Too few or too many, or the wrong kinds of arguments will result in unusual sentences. The arguments that must be present in order for the verb's sentence to make grammatical sense are the verb's core arguments.

give (argument 1, argument 2, argument 3)

What's more, we noticed that the first argument takes a subject form (like I or she) and the second and third arguments take an object form (like me or her). English pronouns show these forms morphologically, but English nouns do not. We'll still differentiate subject and object nouns, since other characteristics (like placement of the subject before the verb and objects after the verb) remain the same for both nouns and pronouns. We can now specify not only how many core arguments verbs like give and go take, but what kind of core arguments:

give (subject, object, to + object)
go (subject, to + object)

We'll spend much of the rest of these lessons considering two key issues we've addressed above: 1) how many arguments does a verb allow, and in what arrangement? 2) what kind of arguments does verb allow?

Practice Exercise

1) Identify the verb and its arguments in the following sentences:

He sees everything.
Joe said that to me.
Do you speak multiple languages every day?
I walked down the stairs with my broken leg.

2) Which of the arguments above are core arguments? Which ones are optional?

3) Give the entry for each verb and its arguments in question 1 following the structure introduced above: verb (arguments).

Valency & transitivity

Valency

I mentioned that we'd discuss the number of arguments verbs allow as well as the type of arguments. At this point, we turn to the number of arguments. Valency (sometimes called valence) refers to the number of core arguments that are part of a verb's morphosyntactic structure.

Valency is specific to a single verb. A verb like run expects one argument, the subject, so it's called monovalent. Verbs that expect two, like the verb see, are divalent. Verbs expecting three are trivalent, including give. The labels "monovalent", "divalent" and "trivalent" depend on the core arguments, the arguments required for normal use of the verb in question. We could always add extras in the forms of adjuncts: she saw it [on Thursday], or she saw it [on Thursday] [with her brother].

give (1, 2, 3) trivalent
lift (1, 2) divalent
fall (1) monovalent
rain (?)

Like other verbs, give has a number of slots that take a certain number of nouns and pronouns. Give has three of those slots. It expects three arguments. Put simply, we use the verb give with a giver, a given and a "givee". To use lift, we need a lifter and a lifted. To use fall, we just need a faller. So far, the logic behind this is easy.

Then we run into a verb like rain. The verb rain doesn't fit neatly into this scheme. On the one hand, it takes a mandatory subject it. So, it looks like it has one core argument. On the other hand, it can't be switched out for another noun or pronoun. What's more, it doesn't seem to refer to anything specific. It's more of a placeholder (called a dummy pronoun or expletive).

We may argue that rain has no real arguments, but that English adds a dummy pronoun for grammatical reasons. (A finite English verb has a subject noun or pronoun.) So, rain and similar "weather verbs" have a place in this scheme after all:

rain (0) avalent
snow (0) avalent
snow (0) avalent

Transitivity

We can look at this another way, by focusing on transitivity. Transitivity looks at the verb from a different perspective: what kind of object arguments does the verb take? Subjects and objects play a direct part in the verb's action. Oblique elements are secondary. Some of those oblique elements are optional, but indirect objects are often core arguments. In English, direct objects most often immediately follow the verb (assuming the verb takes a direct object), while indirect objects may be expressed with a prepositional phrase.

give (subject, direct object, indirect object)
lift (subject, direct object, (no indirect object))
defer (subject, (no direct object), indirect object)
fall (subject, (no direct object), (no indirect object))
rain ("dummy" subject)

Transitive verbs take a direct object. Intransitive verbs do not take a direct object. Two of our verbs above are transitive: give and lift. Three are intransitive: defer (in the sense of defer to (someone)), fall and rain.

In order to avoid potential confusion, narrower terms have been added. Ditransitive verbs expect both a direct object and an indirect object. So, we can classify give as ditransitive. Notice that I have not marked a difference between verbs that take only an indirect object and verbs that want neither a direct nor indirect object - from the perspective of transitivity, both are "intransitive".

Defective verbs typically fail to represent all possible subjects. For example, recall that rain can only take the dummy pronoun it as its single core argument. From the perspective of transitivity, rain is still an intransitive verb. We might even say that rain-type verbs are even less transitive than sleep-type verbs.

For starters, transitivity and valency are two ways to think about the number of arguments that a verb has. Next time we'll move on to consider the type of arguments that a verb may have.

Practice Exercise

1) Give the transitivity values of the verbs you worked with in the last practice exercise.

2) Give the valency values of the verbs you worked with in the last practice exercise.

Thematic roles & case

Thematic roles

The last lesson focused on the number of core arguments a verb has. This time around, we're going to distinguish noun roles from noun cases. At first, it looks as if case will do the job of marking the types of arguments that a verb allows. After all, if the verb give has slots for a subject, direct object and indirect object, it follows that we can only use nominative case pronouns in subject slots and accusative case pronouns in direct object slots.

English I gave it to him.
We gave it to him.
I gave them to him.
*Us gave it to him.
*I gave they to him.

We're about ready to close the case and consider this proof positive that verbs dictate not only the number of core arguments but the case and position of each argument.

Middle English Me thinks he hath courage.
  
Spanish Yo se lo di.
'I gave it to him.'
Me gusta mucho.
'To-me likes a lot.'
  
Icelandic Ég gaf hann.
'I gave it to him.'
Mig vantar hús.
'Me needs (a) house.'

Like Modern English, the languages above use different pronoun cases. However, the case of a single argument changes depending on the verb. The argument in position 1, which we've been calling the "subject", isn't always in the nominative case (the "subject" case).

Now let's completely break our assumption that argument positions are equivalent to noun cases. Some languages are even extremely regular in their use of different cases for the same argument position, depending on the verb.

Basque Ni etorri naiz.
'Me have come.'
Nik ura dut.
'I have water.'
Nik luburua ikusi dut.
'I have seen the book.'
Ni erori naiz.
'Me have fallen.'

We'd still like to say that these verbs have subjects and objects, but we can't say that the subject is equivalent to one specific case (like the nominative) and the object to another (like the accusative). Sometimes the same subject in the same position (argument 1) will look like nik, other times we find ni.

Let's take a different approach. We'll pursue the path we started down earlier. We noticed that, when we use the verb give, we also need to include a "giver", a "given" and a "givee". Let's call those roles. In fact, since they're something like thematic performances or parts that nouns play, we call them thematic roles. We could say that "giver", "givee" and "given" are names for the thematic roles of the arguments of give. Other verbs have roles, too. See has a "seer" and a "seen", and make has a "maker" and a "made".

But there's something about roles that strikes us as more universal. The giver of give, the maker of make and the singer of sing all share something - they are all agents. So, let's call this role the "agent". Similarly, most nouns we've been calling direct objects seem to be passively involved (actions are done to them - they are given, made or seen). Let's call that role the "patient".

You'll come up with a further observation when you look at intransitive, monovalent verbs like sleep or fall. The argument we've been calling the subject of those verbs doesn't instigate the action the way an agent would. There's another role shared by sleepers, fallers and subjects of intransitive verbs more generally. Let's call that role the "experiencer".

Look again at the Basque examples above. When the subject is an agent, the noun takes one form. When the subject is an experiencer, it takes another form.

Cases vs. roles

So, what is the "role" of case, if it isn't to define the role a noun plays in a sentence?

As you saw above, different cases might be used for nouns playing different roles. On the other hand, different cases might be used for nouns playing the same role.

Icelandic Mig vantar hús.
'Me needs (a) house.'
?Mér vantar hús.
'To-me needs (a) house.'

In the first Icelandic sentence, speakers fill the "experiencer" role with a pronoun in the accusative case. The second sentence (marked as questionable here) uses a pronoun in the dative case in that same argument position. Since Icelandic speakers don't just throw a nominative noun in that position and say *ég vantar hús, it's safe to assume that there's still some connection between case and role, right?

Think of it this way. Nouns and pronouns need some sort of case to be used. Otherwise, the uncased noun is left to the side, unusable. We've already seen that verbs dictate the number and roles of their arguments. Each of those roles has an eligibility requirement - say, that a noun must be in a nominative case. Nouns trained with these qualifications are eligible to fill that role.

So, in English, we might find that all "patients" are accusative pronouns & nouns (Amir saw him), but not all accusative pronouns & nouns are patients of verbs (for him). Accusative pronouns are just qualified to fill that patient role. Roles are filled with a noun that has case. Once a noun has case, it is eligible for certain roles but ineligible for others. This explanation may justify why cases remain distinct from roles, but specific cases are associated with specific roles.

As an aside, notice that we've stripped the terms "subject" and "object" of their previous values. They no longer refer to full-blown noun cases. Also, they're not related to the thematic roles that are determined by the verb's argument structure. We're left with two terms "subject" and "object" that refer to some syntactic categories, perhaps not built into the verb but part of sentence structure. Given the approach we took above, we might put it this way: in English, subjects (syntax) are associated with agent or experiencer roles (morphosyntax) that are filled by nominative case nouns or pronouns (morphology).

For now, let's set case aside by saying that cased nouns are qualified to fill thematic roles. For example, English nouns that fill the role of patient are in the accusative or object case, but not every noun that's in the accusative fills the role of patient. Grammatically, they're two distinct things. That will suffice for a preliminary distinction between cases and roles. Join me next time to build on cases and roles as we turn to the topic of alignment.

Practice Exercise

Return to the same verbs you worked with in the previous activity, and answer the following questions:

1) Identify the case of every noun/pronoun (you may use the words "subject", "direct object" and "indirect object" if you prefer).

2) Identify the thematic role of every noun & pronoun. Use the three roles we named above.

3) One of the verbs is associated with a thematic role we haven't described. Which argument of which verb is that? What name might suit that "missing" role?

Alignment

Roles and cases revisited

In the last lesson, I proposed that the noun's cases and the thematic roles of arguments are two different things. Now I'd like to illustrate that distinction with some examples and make this really clear as we start to discuss alignemnt.

When I used Basque examples in the last section, I mentioned that some languages consistently treat subjects differently depending on the verb's transitivity. Specifically, subjects of transitive verbs don't look like subjects of intransitive verbs.

I used that as preliminary proof for the notion that verbs actually assign roles to their arguments, but not necessarily the case of the nouns that will fill those roles. In other words, roles are meaning-based - we expect to find "agent" and "experiencer" (subject-type) roles and "patient" (object-type) roles in a language. In contrast, we expect that different languages will fill those roles with different cases. The cases a particular language uses to fill those thematic roles reveal that language's morphosyntactic alignment.

If that starts to sound too abstract or technical, let me illustrate with some examples. First, let's abbreviate the agent as A, the patient as O and the experiencer as S (just to be difficult). Take the time to reread my definition of thematic roles above to make sure you're clear on the distinction between 1) case & role and 2) the three roles known as "agent", "patient" & "experiencer" (often abbreviated A, O & S).

I'd like to return to the Iberian Peninsula (where Basque is spoken), and compare examples from Basque and Catalan (a language closely related to Occitan/Provençal and, more distantly, to French and Spanish). Let's pick one intransitive verb and one transitive verb from Basque, and the same intransitive verb and transitive verb from Catalan.

Two different alignments

We're going to compare a few basic sentences in Basque and Catalan to get a sense of the verb's role in selecting roles filled by different cases. Consider the following verbs in Basque and Catalan.

Basque erori (1 = S)
'to fall'
ikusi (1 = A, 2 = O)
'to see'
  
caure (1 = S)
'to fall'
veure (1 = A, 2 = O)
'to see'

Not only do the Catalan and Basque verbs have the same meaning, they also share the same number and type of arguments. With respect to argument number and assigned thematic roles, the transitive see and intransitive fall are grammatically identical so far. Now, let's start to fill those roles with real nouns. First, the transitive verb see:

Basque ikusi da
'has seen'
Jonek Peio ikusi du.
'John has seen Peter.'
Peiok Jone ikusi du.
'Peter has seen John.'
  
Catalan ha vist
'has seen'
En Joan ha vist el Pere.
'John has seen Peter.'
En Pere ha vist el Joan.
'Peter has seen John.'

Nothing remarkable here. Understandably, both languages expect the noun filling the agent A and the noun filling the patient P to have different cases. Basque uses suffixes, while Catalan uses particles to differentiate those cases. The subject and object are clearly and distinctly marked. At this point, we're tempted to revert to our old notion of "subject case" vs. "object case". That is, until we consider the intransitive verb fall:

Basque erori da
'has fallen'
Jone erori da.
'John has fallen.'
  
Catalan ha caigut
'has fallen'
En Joan ha caigut.
'John has fallen.'

Something different happens when we compare the intransitive verb fall to the transitive see in the two languages. Catalan does what we've come to expect - it fills the experiencer role of fall and the agent role of see with a noun in the same case. Only the patient role of see is treated differently in Catalan. Basque, on the other hand, fills the experiencer role of fall and the patient role of see with a noun in the same case. Only the agent role of see is treated differently in Basque.

We end up with noun roles filled with nouns from different cases in different languages. We say that the two languages differ in alignment. Basque aligns the experiencer and patient roles, and separates the agent role. Catalan aligns the experiencer and agent roles, and separates the patient role. If we looked at more examples from both languages, we'd find that they conform to this pattern.

language alignment
Basque S = O, A
(treats nouns filling experiencer roles like nouns filling patient roles; treats agent roles differently)
Catalan S = A, O
(treats nouns filling experiencer roles like nouns filling agent roles; treats patient roles differently)

Of course, there are names for these two alignments. The Catalan alignment gets called "nominative-accusative", while the Basque alignment is "ergative-absolutive". English is an example of another accusative language. Tibetan is another ergative language. These two aren't the only ways languages have of handling alignment.

Practice Exercise

1) Create three sentences in English with the pronoun she. In the first sentence, make her the subject of the transitive verb see. In the second, make her the object of the same verb see. In the third sentence, make her the subject of the intransitive verb fall.

2) Does English have ergative-absolutive alignment or nominative-accusative alignment? Use the pronoun in the examples you just created along with the explanations you read above to justify your answer.

3) Stop and recall the section in Intro to the Grammar of Sentences that introduced you to rules and rule formulation. Try to write a productive rule that adequately expresses how Basque treats a noun when it's the subject of a transitive verb, the object of a transitive verb and the subject of an intransitive verb.

Answers to the practice exercises

Practice Exercise (Verbs & nouns revisited)

1) see (1 = he, 2 = everything); say (1 = Joe, 2 = that, 3 = to me); speak (1 = you, 2 = multiple languages, 3 = every day); walk (1 = I, 2 = down the stairs, 3 = with my broken leg)

2) All look like core arguments except argument 3 of speak and arguments 2 & 3 of walk (arg 3 of speak and args 2 & 3 of walk are optional - they may be removed without impacting the grammar of the sentence).

3) see (1, 2, 3); say (1, 2, 3); speak (1, 2); walk (1)

Practice Exercise (Valency & transitivity)

1) see = transitive; say = ditransitive; speak = transitive; walk = intransitive

2) see = divalent; say = trivalent; speak = divalent; walk = monovalent

Practice Exercise (Roles & cases)

1) he = nominative/subject, everything = accusative/object; Joe = nom/subject, that = acc/object, to me = dative/indirect object; you = nom/subject, multiple languages = acc/object; I = nom/subject

2) he = agent, everything = patient; Joe = agent, that = patient, to me = ?; you = agent, multiple languages = patient; I = experiencer

3) The indirect object "to me" of "Joe said that to me" doesn't fit into any of the roles we've discussed so far. Roles often associated with indirect objects are the "target" when the action is directed at the noun ("to me" in "he says it to me") or "benefactor" when the action is for the sake of the noun ("to me" in "he gives it to me").

Practice Exercise (Alignment)

1) a) She sees the dog.
b) The dog sees her.
c) She fell yesterday.

2) She is used as the subject of intransitive and transitive sentences. Her is used for the object of transitive sentences. Like Catalan, English expects nouns filling the A and S roles one way (nominative she), and O separately (accusative her). This is characteristic of a nominative-accusative language.

3) Answers will vary, but take this as an example:

Basque ergative-absolutive case rule
Look at the verb's transitivity value.
If the verb is transitive, select a subject noun that is in the ergative case ("subject" case) and an object noun that is in the absolutive case ("object" case).
If the verb is intransitive, select a subject noun that is in the absolutive case ("object" case).

Using the concept of roles, we can build a rule that doesn't have to look at the verb's transitivity. This one might be more straightforward:

Basque ergative-absolutive case rule
Look at the thematic role being assigned.
If the role being assigned is agent, select a noun that is in the ergative case ("subject" case).
If the role being assigned is patient or experiencer, select a noun that is in the absolutive case ("object" case).