So far, we’ve been talking about individual speech sounds as if they’re all separate from each other. But we know of course that we don’t articulate individual segments when we speak – we don’t produce the word book as [b] – [ʊ] – [k]. When we’re speaking, our articulators are always moving – they’re moving away from from the position for the sound they just made, and preparing to make the sound that’s coming up.

You can feel this really easily by saying a couple of words. I want you to prepare to say this word, but don’t actually say it: just put your mouth in the position to say the word key. Pay attention to how you’re holding your mouth. What do you notice?  Now get ready to say this word, but don’t actually say it, just freeze in the position: cool. What position are your articulators in? Both key and cool start with the voiceless velar stop [k^h], so if we articulated speech segments individually, we’d expect our mouths to be in the same position for both words. But the vowels in each word are quite different: [i] in key is high and front, and [u] in cool is high, back and rounded.  So when we produce that [k] sound, our mouths are already preparing for the next vowel. This is called coarticulation:  the articulation of every speech sound is shaped by the sounds that come before and after it. When we’re doing detailed, narrow phonetic transcription, we can include details about coarticulation and other articulatory processes.

Probably the most common articulatory process is assimilation.  You can guess from its name that it involves sounds becoming more similar to each other. Sounds often become more similar to what’s coming up in the word. Here’s an example; say the words cat and can. They both have the vowel as the nucleus, but for can, when we produce that [æ] we’re already anticipating the upcoming nasal so we’ve already got the velum lowered to allow air into the nasal cavity. So the vowel gets nasalized too — it gets assimilated to the following nasal. We transcribe a nasal vowel with the diacritic for nasalization, like this:[æ̃]. Because this nasalization is in anticipation of an upcoming nasal consonant, we call this process anticipatory assimilation: the vowel is becoming more similar to the sound that follows it.  In some books, you might see this called regressive assimilation, since the nasal property of the [n] is moving backwards or regressing onto the vowel.

Assimilation can go in the other direction too:  sometimes the properties of one speech segment persevere into the next segment.  Say these two words out loud: bleed, please. The two [l] sounds in these two words are a little different from each other.  For bleed, the vocal folds are vibrating for the voiced [b] and they keep vibrating to produce the voiced [l].  We know that [l] is usually voiced so there’s nothing remarkable about that. But for please, the vocal folds are held apart for the voiceless [p^h].  We start making the [l] before the vocal folds start to vibrate, so the [l] becomes voiceless in this context. We say that the [l] following a voiceless stop is devoiced, and it gets transcribed with the diacritic for voicelessness, like this: [l̥]. In this case, the voiceless property of the [p] is persevering; it’s sticking around to have an influence on the [l], so we call it perseveratory assimilation. You might also see this called progressive assimilation because the voicelessness of the first sound progresses, or moves forward, onto the following sound. One thing to note about the diacritic for voicelessness: it only gets used when a sound that is ordinarily voiced becomes voiceless in one of these articulatory processes.  An [l] is usually voiced, so if it gets devoiced it gets the diacritic. But a sound like [h] or [s] is already voiceless, so it wouldn’t make any sense to transcribe it with the diacritic.

So assimilation can be anticipatory, where a speech sound is influenced in anticipation of the sound that’s about to be spoken after it, or perseveratory, where a sound is influenced by properties persevering, or lingering, from the sound that was just spoken.