With combinations of morphemes, on the other hand, it is often possible to know exactly what the meaning of a particular complex form will be, provided we know the meaning of the parts. For example, if we know the meaning of the affix `un' and we know that a particular mormpeme A is an adjective, then we know that the complex form `unA' (provided there is such a form---this is a separate, morpho-syntactic question) means something like ``an adjective whose meaning is the opposite or inverse of A''.
It's not always the case that the meanings of complex words are this transparent, however. Consider the case of compounds. Although the meaning of many compounds is transparently derived from the meanings of their constituent words (`dog house', `birthday party', etc.), this is not always so. The compound `bellboy' does not refer to boys who look like bells (compare `goatboy'), it refers to someone who works in a hotel (and may or may not be called by ringing a bell).
A very nice illustration of the fact that you can't always predict the meanings of combinations of morphemes comes from Dakota (Sioux). In Dakota, the morpheme ha means `night' and the morpheme wakha means `holy'. These morphemes can be combined into the compound hawakha. What's interesting is that this compound is ambiguous. On the one hand, it can mean `holy night', which is what we'd expect if we just combined the meanings of the two roots. On the other hand, it can also mean `northern lights', a meaning that is not transparent at all.
A second situation in which we may not know the meaning of a complex word is in cases of lexical or structural ambiguity. The string of words `large stamp collection' can either refer to a collection of large stamps or a stamp collection that's large in size. Which reading we get is determined by how we put the words together, so we need to know something in addition to the words (the syntax) in order to get to meaning.
The final part of the claim---that we can always predict the meaning of a sentence if we know the words it contains and how they are put together (what the grammatical relations between them are)---can quite easily be shown to be false. On the one hand, we clearly need contextual information to figure out what the interpretations of expressions like pronouns are. On the other hand, since meaning includes stuff other than just what the words in the sentence give us, such as implicatures, the words by themselves just won't be enough. For example, to know that B's sentence in the following discourse means something like `Titanic wasn't such a great movie', we need to know that it was uttered in the context of A's question.
A: How did you like Titanic?
B: I got a lot of homework done.
B's words, by themselves, won't give us this aspect of the meaning of B's utterance, though.
Readings 1 and 2 arise from a structural ambiguity that is dependent on whether we analyze `her' as the indirect object and `dog biscuits' as a constituent which supplies the direct object, or whether we analyze `her dog' as a constituent which supplies the indirect object and `biscuits' as the direct object. The other readings arise from a pragmatic ambiguity, dependent on how the context tells us to fix the reference of the pronouns.
(ii) I'd like you to take out the garbage.
Reading 1a: I'd be happy if you knocked down the garbage pail(s)
Reading 1b: I want you to knock down the garbage pail(s).
Reading 2a: I'd be happy if you took the garbage outside.
Reading 2b: I want you to take the garbage outside.
Reading 3a: I'd be happy if you took the garbage out of whatever it's in.
Reading 3b: I want you to take the garbage out of whatever it's in.
Other readings: The different interpretations we get depending on the values of the pronouns, again.
The contrast between the (a) and (b) readings in 1-3 is pragmatic, depending on whether we interpret `like' literally, or whether we take into account the implicature that whenever someone says ``I'd like you to X'' they mean ``I want you to X''. The contrasts between 1 and 2-3 involve a lexical (and possibly syntactic as well) ambiguities: Reading 1 (for those who get it) comes from an idiomatic interpretation of `take out', reading 2 and 3 involve the more standard interpretations `take'. The contrast between reading 2 and reading 3 is also a kind of lexical ambiguity, that is dependent on whether we take `out' to mean something like `out to' vs. `out of'. Note that there's a The `other readings' are probably best characterized as pragmatic, because they shift from context to context, depending on who the speaker of the sentence is and who the addressee is.
(iii) Susan didn't buy two books.
Reading 1: There are two (specific) books that Susan didn't buy.
Reading 2: It's not the case that Susan bought exactly two books.
This is a case of what deSwart calls ``semantic ambiguity'', because (iii) does not appear to be structurally ambiguous, none of the lexical items are polysemous, and the two interpretations do not seem to be dependent on contextual factors (though context might tell us which reading is the intended one). Some people might notice that it appears to be possible to further divide reading 2 into two ``subreadings'':
Reading 2a: Susan bought less than two books.
Reading 2b: Susan bought more than two books.
(iv) Jane told me that she was very unhappy.
Reading 1: `she' = Jane
Reading 2: `she' = someone other than Jane
Other readings: speaker-dependent interpretations of `me'
All of these ambiguities are pragmatic in nature, as they are dependent on how the context tells us to interpret the pronouns.
(i)a presupposes and entails (i)b. That (i)a entails (i)b is shown by the fact that asserting (i)a and denying (i)b results in a contradiction:
(i)c. ??That John was assaulted scared Mary, but John wasn't assaulted.
That (i)a presupposes (i)b is shown by the fact that the entailment (i)b remains when (i)a is denied or turned into a question.
(i)d. That John was assaulted didn't scare Mary. [??But John wasn't
(i)e. Did (the fact) that John was assaulted scare Mary? [??...even though John wasn't assaulted?]
(i)b neither presupposes nor entails (i)a, as shown by the fact that you can assert (i)b and deny (i)a without contradiction:
(i)f. John was assaulted, but this [= that John was assaulted] didn't scare Mary.
(ii)a. Is John not aware that Mary is pregnant?
(ii)b. Mary is pregnant.
(ii)a presupposes (ii)b, since (ii)a is a question and (ii)b remains as an inference. This is illustrated by (ii)c, which shows that it's infelicitous to utter (ii)a and then deny (ii)b.
(ii)c. Is John not aware that Mary is pregnant? ??She isn't!
(ii)a does not entail (ii)b, however, because questions don't have truth values (in the same way that statements do). Since entailment is defined in terms of truth values, it follows that questions don't have entailments.
(ii)b neither presupposes nor entails (ii)a, as shown by the fact that we can assert (iib) and deny (ii)a:
(ii)d. Mary is pregnant, and John is aware of this [= that Mary is pregnant].
(iii)a. Some Italian is a violinist.
(iii)b. Bernardo is an Italian violinist.
(iii)a doesn't entail (iii)b, but (iii)b entails (iii)a, assuming that (iii)b doesn't mean that Bernardo plays an Italian violin (compare `Bernardo is an Italian violin player', which can have this interpretation). The fact that (iii)c is a contradiction shows that this is indeed an entailment; the fact that (iii)d is not a contradiction shows that (iii)a does not entail (iii)b.
(iii)c. Bernardo is an Italian violinist, but there are no Italian
(iii)d. Some Italian is a violinist, but Bernardo is not an Italian violinist.
(iv)a. If I discover that Mary is in New York, I will get angry.
(iv)b. Mary is in New York.
(iv)a are (iv)b not related by entailment or presupposition. That (iv)a doesn't presuppose (iv)b is fairly clear: intuitively, if I know that (iv)b is true, then my utterance of (iv)a would be quite strange. (How could I `discover' that Mary is in New York?) The negation test supports this claim: the negation of (iv)a does not require (iv)b to be true, nor does the question form:
(iv)c. It's not true that if I discover that Mary is in New York, I
will get angry.
(iv)d. Will you get angry if you discover that Mary is in New York?
There's no entailment relation, either, as shown by the fact that (iv)e is not contradictory. (Note that the wording of (iv)e ``counts as'' an instance of `(iv)a and not (iv)b', because I have asserted (iv)a and denied (iv)b.)
(iv)e. If I discover that Mary is in New York, I will get angry, so it's a good thing that Mary is not in New York.
The absence of entailment/presupposition relations in the other direction can be demonstrated in the same way.
(v)a. It's possible that Clinton will return to Arkansas.
(v)b. Clinton was in Arkansas before.
(v)a both entails and presupposes (v)b. The entailment is illustrated by the contradictory sentence (v)c; the presupposition is shown by the fact that the inference from (v)a to (v)b remains in the scope of negation and in questions:
(v)c. ??It's possible that Clinton will return to Arkansas,
even though he has never been there before.
(v)d. It's impossible that Clinton will return to Arkansas.
(v)e. Is it possible that Clinton will return to Arkansas?
(i) Keynes rhymes with brains.
(ii) He is a linguist.
(iii) He is a personal pronoun.
(iv) The sentence he is a personal pronoun has five words.
(v) Snow is white is true if and only if snow is white.
Return to the class homepage.