User:Grover cleveland/r stuff


 * Stage 1: before R-dropping
 * letter: /lɛtər/ -> [lɛtər]
 * the letter is: /ðə lɛtər ɪz/ -> [ðə lɛtər ɪz]
 * comma: /kɒmə/ -> [kɒmə]
 * the comma is: /ðə kɒmə ɪz/ -> [ðə kɒmə ɪz]
 * Stage 2: underlying forms still contain /r/, which is realized as zero unless immediately before a vowel
 * letter: /lɛtər/ -> [lɛtə]
 * the letter is: /ðə lɛtər ɪz/ -> [ðə lɛtərɪz]
 * comma: underlying form: /kɒmə/ -> [kɒmə]
 * the comma is: underlying form: /ðə kɒmə ɪz/ -> [ðə kɒmə ɪz]
 * Stage 3: underlying forms no longer contain /r/. A realization rule inserts [r] after /ə/ when followed by a vowel
 * letter: /lɛtə/ -> [lɛtə]
 * the letter is: /ðə lɛtə ɪz/ -> [ðə lɛtərɪz]
 * comma: /kɒmə/ -> [kɒmə]
 * the comma is: /ðə kɒmə ɪz/ -> [ðə kɒmərɪz]

Stage 1 represents a fully rhotic accent. At Stage 2, speakers do not prounounce /r/ except before a vowel, but underlying forms still contain /r/. A child exposed to the output of stage 2 is likely to lose the underlying /r/ in words like letter, and explain the linking R in phrases like the letter is as the result of an R-insertion rule. There are two reasons for this:
 * In all contexts other than word-finally, the child has no reason to postulate underlying /r/ except before a vowel. It is therefore more regular to postulate that underlying /r/ can only ever occur before a vowel.
 * In principle, the child's hypothesis of an R-insertion rule ought to be contradicted by examples like the comma is, where there is no [r] in stage 2 output. However, words like comma are very rare in the everyday English a child is exposed to during language acquisition, and are far outweighed by words of the letter type where [r] is present.