Information status

In English, the location of the nucleus is strongly affected by whether the words contain old or new information. The general rule is that we accent new information, but not old information. That is, we deaccent(=remove potential accents from ) old information.

If all the information in the utterance is new, then we can accent all the lexical items. So the nucleus is placed (as expected) on the last lexical item:

〉 Yes madam?
I'd like a gin and 'tonic
However, in practice we tend to downgrade potential accents between the first one and the last. (This is discussed in
5.9
). So in practice we often say:
Yes madam?
I'd like a gin and 'tonic
and we apply this principle in the examples.

As long as the last lexical item contains new information, the lexical item is accented, and thus bears the nucleus. However, if the last lexical item contains old information (= something already mentioned), then it is not accented. Rather, it is deaccented. So the nucleus goes earlier, namely on the last item that does contain new information:

In this example, tonic has already been mentioned, and is therefore old information. As a result, it gets deaccented. Thus the place of the nucleus normally signals the end of the new information in an intonation phrase.

We generally avoid placing a nucleus on an item which repeats something that has been said earlier: we do not accent a repeated item (a 'given' item, old information). So we say, for example:

〉 Who doesn't want to dance?
》 'Bill doesn't want to dance
It would sound strange to say:
〉 D'you object to dogs?
》 No I adore 'dogs.

〉 Who doesn't want to dance?
Bill doesn't want to 'dance.

In the correct versions, we see that the repeated items (dogs, want to dance) are not accented. Rather, they are deaccented: they lose the accent they might otherwise have had. This is because the information they convey is not new.

We deaccent repeated words even if, strictly speaking, they contain new information. Thus we tend to say:

a green chair and a 'blue chair
Tina Rodman and 'Jane Rodman
72'52 (= seven two 'five two)
where the final, deaccented, item would not actually be predicable from the context (even though the intonation, once we reach the nucleus, makes it predictable for the hearer). After all, the speaker might have been going to say:
a green chair and a blue 'curtain
Tina Rodman and Jane 'Stuart
725'6(= 'seven two five 'six)