We have just seen how the meaning of a simple sentence is determined from the meanings of its parts. Another common way of thinking about this process uses the concept of a thematic role(Gruber 1965, Fillmore 1968, Jackendoff 1987), a part of a word’s meaning which indicates the role that some individual plays in the action which that word describes.
Verbs and other heads of phrases may be associated with such thematic roles as agent, patient, theme, location, source, and goal. These roles are assigned to the subject and any arguments of the head. For example, the verb hitis associated with the roles <agent, patient>. (The list of thematic roles associated with a particular word is called a thematic grid.) The the- matic role of “agent” is assigned to Mary, the subject, indicating that Mary was the “doer” of the action of hitting, while the thematic role of
“patient” is assigned to the ball, indicating that the ball was acted on as part of the action of hitting. Here are some other thematic roles:
Theme: something which moves, literally or metaphorically, as part of an action
Source: the location/individual from which movement occurs Goal: the location/individual to which movement occurs Location: the location at which something happens Experiencer: someone who experiences something
Instrument: something an agent uses to make something happen Cause: something that causes something to happen
Stimulus: something that causes an experience
Here are some examples of verbs with their thematic grids (enclosed in angled brackets):
(5) Allen sent the book to Mary.
send: <agent, theme, goal>
(6) The corn ripened.
ripen: <theme>
(7) The sun ripened the corn.
ripen: <cause, theme>
Sometimes it can be tricky to come up with the right thematic roles to describe the meaning of a particular verb. For example, ripenhas the role
“theme” in (6)–(7) because we metaphorically view the change from unripe
to ripe described by this verb as a kind of movement. Notice that the verb ripenhas two different thematic grids in (6) and (7), demonstrating an impor- tant point about thematic roles. Many verbs have more than one thematic grid, and verbs can be classified according to which thematic grids they have as part of their definitions. For example, other verbs are like ripenin having both the <theme> grid and the <cause, theme> grid (e.g. break, redden).
Linguists are interested in studying which verbs are associated with which thematic grids – both within a single language like English and across lan- guages – because these patterns can tell us more about how these meanings of words are represented in our minds. For example, the pattern illustrated with ripenin (6)–(7) shows that our minds can make use of a “piece” of mean- ing that can be expressed as CAUSE. Verbs like ripen, with the meaning in (6), can be affected by a CAUSEVprocess, to create a new meaning, like ripenin (7), which can be represented as CAUSEripen<theme>. Some languages have a morpheme meaning CAUSE, but in English this piece of meaning can be added without changing the pronunciation of the word. Words which, like ripen, have multiple related meanings are polysemous, and processes like CAUSEVcan help to explain how they become polysemous.
Linguists who are interested in the meanings of words, and the rela- tions among words’ meanings, study lexical semantics. Thematic roles provide one very popular framework for investigating lexical semantics, in particular the lexical semantics of verbs, but not the only one. Lexical semantics is very interesting to syntacticians, because the meaning of a word often influences how it fits into syntax; for example, the fact that ripencan have two different patterns of thematic roles explains why it can be used grammatically either with or without an object. Another example is the Warlpiri sentence discussed above. The fact that ‘big’ is expressed as a noun – as are other meanings, like ‘how many’ and ‘in the middle’, which don’t seem “noun-like” to English speakers – implies that the gram- mar of this language will be very different from that of English. The Warlpiri example shows why lexical semanticists study the meanings of words in multiple languages, both to find out how word meanings are similar across languages and to determine how they may differ.
Similar meanings can be expressed very differently in different lan- guages. In the English sentence I like pasta,the subject Irefers to the per- son who likes something, and the object pastarefers to the thing I like.
The English verb like has the thematic grid <experiencer, stimulus>. In many languages, the person who likes something appears in the sentence as an indirect object, as in the Italian sentence (8).
(8) Mi piace la pasta.
to-me pleases the pasta
‘Pasta is pleasing to me.’
Italian doesn’t express the liking of things in the same way as English does. There is no verb equivalent to likesin Italian; piacerecomes closest, but it expresses the relationship between “liker” and “liked thing” quite differently. The Italian verb has a thematic grid <stimulus, experiencer>.
In this way it is similar to the English verb please, as in Pasta pleases me.
Piacereand pleaseare not equivalent, however; the experiencer is expressed as an indirect object with the Italian verb and a direct object with the English one.
English and Italian express the concept of “love” differently as well. In English, the verb loveis often used in a way similar to like; we can say I love pastaas well as I love my wife.The Italian verb for ‘love’ (amare) has the same thematic grid as the English verb love, <experiencer, stimulus>, and the
‘‘lover’’ is the subject in both languages:
(9) Amo mia moglie.
I-love my wife.
However, the meaning of amare is narrower and more serious than its English counterpart. To say Amo la pastain Italian would be quite strange, indicating a serious emotional attachment to pasta. There is no simple correspondence between like and love and the corresponding verbs in Italian. They are similar in some ways and different in others, and this is the typical situation when one compares the meanings of words across different languages.
Box 4.1 Semantics in the law
Semantic issues can be important in real life. Solan (2002) discusses an interesting example of how semantics can plan a role in the inter- pretation of legal statutes. He writes: “Raymond Moskal, who lived in Pennsylvania, would buy used automobiles, set back the odometers, send the inaccurate mileage readings to Virginia along with other required information, and receive new titles from Virginia with the incorrect mileage. He would then sell the cars for inflated prices to unsuspecting customers. He was prosecuted and convicted for violat- ing a statute that prohibits the interstate transportation of ‘falsely made’ securities.” In short, Moskal got real titles that contained false information.
The law in question was the following (it is made clear elsewhere that car titles count as “securities”):
Whoever, with unlawful or fraudulent intent, transports in inter- state or foreign commerce any falsely made, forged, altered, or counterfeited securities or tax stamps, knowing the same to have been falsely made, forged, altered, or counterfeited . . . Shall be fined under this title or imprisoned not more than ten years, or both.
(18 USC § 2314 (2001))
The US Supreme Court agreed that Moskal could be punished under this law, but Justice Scalia dissented for two reasons based on the mean- ing of the phrase falsely made. One reason had to do with the historical meaning of the phrase falsely madein legal documents and the other had to do with its ordinary meaning. Justice Scalia showed that in the
100 years up to 1939, when the statute was written, legal documents had used falsely madeto mean ‘forged’ or ‘counterfeit.’ Thus, it seems that the meaning of this crucial phrase had changed, at least within the world of law, between the time the law was written and the time it was applied to Moskal. Scalia’s other argument was that the phrase falsely made, in its ordinary meaning, includes only things that are counterfeit, not real documents that are made to contain false information.
After looking at the usage of the phrase in a database of contempo- rary English, Solan concluded that Scalia’s ordinary meaning argu- ment is wrong. He shows that falsely madetypically means ‘made to include false information,’ as in “[W]hen falsely made, this accusation [child abuse] can be enormously destructive.” In other words, a falsely made accusation means that that accusation contained false informa- tion, and Solan assumes by analogy that a falsely made car titlewould be a car title containing false information.