Assimilation is a connected speech process in which a sound changes to become more similar to a neighboring sound. It is driven by articulatory efficiency: the speech organs anticipate the next sound and begin adjusting early, or carry over a feature from the previous sound. The result is smoother, faster production at the cost of the original sound's identity.
Regressive (anticipatory) assimilation: the most common type in English. A sound changes under the influence of the sound that follows it. The mouth "looks ahead."
Progressive assimilation: a sound changes under the influence of the sound that precedes it. Less common in English but regular in morphology: the plural -s is pronounced /s/ after voiceless consonants (cats /kæts/) and /z/ after voiced consonants (dogs /dɒgz/).
Coalescent assimilation: two adjacent sounds merge into a single new sound that shares features of both. This is the most noticeable type for learners:
Assimilation is gradient, not categorical. It occurs more frequently in:
It occurs less in careful speech, emphatic pronunciation, or when a speaker is consciously articulating for clarity.
For listening, learners need to recognise assimilated forms. A Vietnamese learner hearing /wʊdʒuː/ needs to parse it as "would you"; if they only know the citation forms, this string is opaque. Exposure through natural-speed audio with transcript support is the most effective method.
For production, coalescent assimilation (/t+j/, /d+j/) is the highest-priority target because it is pervasive, rule-governed, and learnable. Drilling common phrases such as "What do you...?", "Would you...?", and "Don't you...?" as chunks with their assimilated forms builds automaticity. Place assimilation (/n/ → /m/, /n/ → /ŋ/) is lower priority for production because it happens naturally as fluency increases.
Assimilation interacts closely with Elision. In a cluster like "handbag" /hæmbæg/, the /d/ is elided and the /n/ assimilates to /m/, with both processes operating simultaneously. Teaching them together through Connected Speech gives learners a more accurate model of how English actually sounds.