Linking R

Linking R and intrusive R are sandhi or linking phenomena involving the appearance of the rhotic consonant (which normally corresponds to the letter ⟨r⟩) between two consecutive morphemes where it would not normally be pronounced. These phenomena occur in many non-rhotic varieties of English, such as those in most of England and Wales, part of the United States, and all of the Anglophone societies of the southern hemisphere, with the exception of South Africa. These phenomena first appeared in English sometime after the year 1700.

By definition, non-rhotic varieties of English only pronounce /r/ when it immediately precedes a vowel. This is called r-vocalisation, r-loss, r-deletion, r-dropping, r-lessness, or non-rhoticity.

For example, in non-rhotic varieties of English, the sound /r/ does not occur in a word such as tuner when it is spoken in isolation, before an intonation break (in pausa), or before a word beginning with a consonant. Even though the word is spelled with an ⟨r⟩ (which reflects that an /r/ was pronounced in the past), non-rhotic accents do not pronounce an /r/ when there is no vowel sound to follow it. Thus, in isolation, speakers of non-rhotic accents may pronounce the words tuner and tuna identically as /ˈtjuːnə/ (or /ˈtuːnə/ with the yod-dropping that typically occurs in the non-rhotic dialects of the northeastern United States, or /ˈtʃuːnə/ with the yod-coalescence that occurs in Southern Hemisphere English).

In contrast, speakers of rhotic dialects, such as those of Scotland, Ireland, and most of North America (except in some of the Northeastern United States and Southern United States), always pronounce an /r/ in tuner and never in tuna so that the two always sound distinct, even when pronounced in isolation. Hints of non-rhoticity go back as early as the 15th century, and the feature was common (at least in London) by the early 18th century.

...
Wikipedia