Final devoicing and affrication of /t/ before /r/

[Rudolph C Troike]

        There is a fortis/lenis difference between the beginning and
ending of syllables in English, which for voiced fricatives translates
into a weaker voicing effect in syllable- and word-final position.
Perceptually this makes it harder to hear the distinction, and speakers
coming from a non-English background may carry forward a lack of actual
distinction in voicing but preserve a fortis/lenis (although less fortis
than in syllable-initial position) contrast. [This can continue into
subsequent generations.]
        /r/ is devoiced after initial /t/ (and other initial voiceless
stops) by the aspiration of the initial stop [compare <treat> and
<street>]. If the tongue-tip assimilates to the retroflex position of the
/r/ [for those varieties of English which have this /r/], the point of
articulation becomes post-alveolar instead of alveolar, and approximates
the acoustic effect of /ch/ [tsh], especially owing to the voiceless [r],
which is articulatorily and acoustically close to the [sh].

Ttelling ESL/EFL learners to pronounce initial /tr/ as if it were /chr/ is
an old TESL trick to get a pronunciation that is more English-sounding and
more intelligible than the transferred /t/ + flapped [r] found in many
other languages.

        Rudy Troike