1) Words are generally word in context: this means that they are surrounded by other sounds, that generally influence the way in which they are performed, even by the same speaker.

2) Speech accomodation (see Giles's works): it has been demonstrated that speakers tend to mimic the prosodic & phonetic features of their conversational counterpart. This means that we change the way we utter a word according to whom we are speaking with.

3) Social cognition: close to the previous point, but slightly different. Thanks to our interactional ability (that is, the "social" variant of the linguistic ability), we are able to modulate our way to produce sounds. We all know when is appropriate or inappropriate to yell, to whisper and which effect may produce shortening or strenghtening sounds.

As for the "pure" acoustic reasons, I think it may be useful to think of speech as a motor action, that needs a coordinate effort of muscles and complex inputs from the primary motor cortex to be performed, exactly as throwing a ball: is the way you throw a ball always the same, and will the ball always fall in the same place? No. Similarly, the action "pronouncing the word: apple" may yield different outcomes.

It is  not always easy to produce the same sound exactly the same every time. Sometimes it is just the case of producing a very closely related approximate which in turn can lead to different pronunciations.

1) Words are generally word in context: this means that they are surrounded by other sounds, that generally influence the way in which they are performed, even by the same speaker.

2) Speech accomodation (see Giles's works): it has been demonstrated that speakers tend to mimic the prosodic & phonetic features of their conversational counterpart. This means that we change the way we utter a word according to whom we are speaking with.

3) Social cognition: close to the previous point, but slightly different. Thanks to our interactional ability (that is, the "social" variant of the linguistic ability), we are able to modulate our way to produce sounds. We all know when is appropriate or inappropriate to yell, to whisper and which effect may produce shortening or strenghtening sounds.

As for the "pure" acoustic reasons, I think it may be useful to think of speech as a motor action, that needs a coordinate effort of muscles and complex inputs from the primary motor cortex to be performed, exactly as throwing a ball: is the way you throw a ball always the same, and will the ball always fall in the same place? No. Similarly, the action "pronouncing the word: apple" may yield different outcomes.

I cannot answer your question, but recommend a read of "The Working Brain, An Introduction to Neuropsychology" by A R Luria, the work much praised by the famous Oliver Sacks in his autobiography. Lucia's book was published in 1973, and so is pre-CTs and MRIs and PETs. But he has a lot to say about language and pronunciation and relevance to brain function. The book is cheap via Amazon.

If you are interested in acoustic reasons, which, as one of the answers above mentioned, are connected with motor actions of speech organs, have you considered works by K. Stevens (articulatory-acoustic aspect of speech)? Especially Acoustic Phonetics (1998).

I think it is useful to do two kinds of distinction to correctly deal with this topic:

1. between a developmental perspective (language acquisition vs language completely acquired)

2. between a typical vs. atypical perspective (different phonological skills among primary and secondary language disorders,  in association  or not with other kind of difficulties).