Halliday & Matthiessen (2014: 416):
The grammar enables us to explain why phrasal verbs have evolved to the extent that they have done in modern English. The leading edge is formed by those of type (i), the adverbial ones, which are particularly widely spread. Typically these have non-phrasal, one-word synonyms, or near-synonyms; yet the phrasal form tends to be preferred, and is strongly favoured in the spoken language. Why is this?
Suppose we have a two-participant clause, active in voice, in which the main item of news is the Goal. The Goal comes at the end, and this is where the prominence – the information focus – typically falls. We can express the process either phrasally or nonphrasally – there is nothing very much to choose between the two:
they cancelled the meeting they called off the meeting
Suppose however that I want the focus of information to be the Process rather than the Goal. At this point a significant difference arises. If I say
they cancelled the meeting
the result is that the information focus is now non-final; this is a marked, strongly foregrounded option, and therefore carries additional overtones of contrast, contradiction or unexpectedness. I may not want these overtones; but the only way I can avoid them is to leave the focus unmarked – i.e. at the end. This means that the Process, not the Goal, must come last. … but in English it is impossible – I cannot say they the meeting cancelled – unless the Process is split into two parts. This, therefore, is what happens, with a phrasal verb: it splits the Process into two parts, one functioning as Predicator and the other as Adjunct, with the Adjunct coming in its normal place at the end:
they called the meeting off