The first exception to the LLI rule is directly linked to the distinction between old and new since it concerns repetitions. As previously explained, information which is already known by the speakers is left out of focus. So, in order for the last lexical item to bear the nucleus it also has to satisfy the condition of being new. This means that repeated words are routinely deaccented even if they are the last lexical item in the IP. This case is so clear that Downing (2006) further refines the LLI principle by saying that the nucleus is placed on the last lexical non-anaphoric item.
Let’s see some examples:
(11) 
The national interest is more important than any personal interest.
(12) The book is loosely based on my family’s history, or rather my father’s history.
(13) I don’t think I could walk that far, never mind run that far.
In none of the above sentences, the last accent is placed on the last lexical word but on the immediately previous one. This is due to the fact that the three items –interest, history and far– have already been uttered before and constitute old information.
It is difficult to overestimate the importance of this feature of English information structure. It is at the core of the language and every student who wants his or her English to sound natural should take it into account.
The following dialogue from a novel can give an idea of how central the deaccentuation of repeated items is in English. The interesting fact is that, despite it being written language, the author uses italics to signal where the accent goes.
(14)
-If you’re going to act like a child, maybe we’ll just have to treat you like a child.
-This isn’t what happens next –said Billy.
-We’ll see what happens next.
(Vonnegut, Kurt; Slaughterhouse Five)
As can be observed, the italics mark the nucleus of the IP. The part of the utterance which is being repeated is left out of focus. What is remarkable here is the author’s interest in being sure that the reader will get the intonation of the dialogue right in her or his mind. This practice is not uncommon in English literature, but seems outlandish in Spanish.
Now let’s have a look at some recorded examples taken from real life.
In (15) the British writer Julian Barnes reads an excerpt of the short story The Man of the World, written by Frank O’Connor.
(15) He had a way, when any of them joined us, of resting against the wall with his hands in his trouser-pockets and listening to them with a sort of well-bred smile, a knowing smile, that seemed to me the height of elegance (Julian Barnes; Leicester, UK).
If we measure the pitch of the sequence smile-knowing-smile, this is what we find:
Smile (1): 120.4 Hz.
Knowing: 123.9 Hz.
Smile (2): 75 Hz.
Between the first and the second time the noun smile is uttered there is a drop in frequency from 120.4 hz. to 75 hz. In between, the adjective knowing stands out with the highest frequency of all (123.9 hz). The measurement consistently reflects Barnes’ thoughtful, delicate reading. There isn’t any question about his intention of stressing the adjective and deaccenting the last smile, as though acknowledging that this is the only way of giving Frank O’Connor’s literature its proper meaning.
The same process can be unmistakably appreciated in (16), where the Canadian journalist Eleanor Wachtel asks a question including the word Spain three times, each of them corresponding to each of the different IPs the excerpt can be divided into. On the first occasion Spain carries the main accent of the IP, but on the following two Spain is left out of focus. It is worth noting that here again the deaccentuation is crystal clear, almost a text-book example.
(16) Your first novel, The South, is set in Spain, / and your most recent short story published in The New Yorker is set in Spain. / What kind of literary inspiration does Spain give you[1]? (Eleanor Wachtel; Montreal, Quebec, Canada).
[1] See also wh-questions ending with a verb.