Voicing assimilation

Lesson opening

This is the first part of the lesson. The rest, the exercises and the reviews are in the app.

Voicing assimilation: Serbian writes what English only says

Say have to at normal speed: “I hafta go.” The v has quietly turned into an f, because the t right after it is voiceless and your mouth gets ready for it in advance. English spelling ignores this and keeps the v of have, so the word stays recognisable. Serbian does the opposite: it writes what it says. English writes vodka; Serbian writes votka, because in Serbian that d before k really is pronounced t. That is the whole lesson in one word. Where two consonants meet, the first one adjusts to its neighbour: it becomes voiced or voiceless to match. This is called jednačenje po zvučnosti (“matching by voicing”), and unlike English, Serbian writes the result into the word. Once you know the law, you will stop wondering why sladak (“sweet”) turns into slatka, or why Srbin (“a Serb”) gives the adjective srpski.

Voicing pairs: who turns into whom

The law rests on seven pairs of consonants in which every voiced sound has a voiceless twin made in the same place in the mouth (b/p as in bin/pin, z/s as in zip/sip). You met this map in “The consonant inventory”; here it is again, because assimilation runs exactly along it:

Consonants outside these pairs take no part in the game: the sonorants j, l, lj, m, n, nj, r and the sound v (always voiced, with no voiceless twin, as “The consonant inventory” showed) neither force a neighbour to change nor change themselves. That is why nothing happens before the v in svet (“world”) or tvoj (“your”). The voiceless c (the ts of cats) and h have no voiced twin either, but they devoice a consonant before them like any voiceless sound: vrabac → vrapca (b before c), iz + hrana → ishrana (“nutrition, diet”). The consonants that change are always from the seven pairs above.

Direction: the consonant on the right wins

In Serbian, assimilation works from right to left: the second consonant decides, and the first one adapts to it. The tongue prepares for the next sound in advance, so the one before it gives way. English does exactly this in have to → “hafta”; it just does not write it down. Here are the models:

Both directions are here. In slatka (“sweet”, feminine), vrapca (“of a sparrow”) and ispod (“under”) a voiced consonant became voiceless before a voiceless one. In svadba (“wedding”) it is the other way round: the voiceless t became voiced before the voiced b and turned into d. One rule covers both: “fall into line with your right-hand neighbour”.

What it looks like in everyday words

Here is a handful of everyday words where assimilation is already built into the spelling. You will meet them all the time:

Cheat sheet

The cheat sheet — everything essential from the lesson on one page. In the app it opens at any time.

Voicing assimilation: write as you speak

Voicing assimilation (jednačenje po zvučnosti): where two consonants meet, the first falls into line with the second. Voiceless before voiced becomes voiced; voiced before voiceless becomes voiceless. Direction right to left: the right-hand neighbour decides. Serbian spelling writes the result: English writes vodka, Serbian votka.

Seven voicing pairs

Sonorants j, l, lj, m, n, nj, r and the sound v take no part: they neither change nor make a neighbour change (svet, tvoj). Unpaired voiceless c and h devoice the consonant before them (vrapca, ishrana).

Follow the right-hand neighbour

Everyday words

In adjectives two laws team up: the fleeting a drops out, the consonants touch, assimilation follows (težak, but teška).

Prefixes iz-, raz-, bez-, uz-

Assimilation happens only when neighbours differ in voicing. A suffix can rewrite the root: Srbin, Srbija, but the adjective in -ski is srpski (b → p before s), the key to the whole -ski class (except d: gradski). Loanwords obey too: top (cannon) + -džija → tobdžija (gunner), p → b before dž; likewise ćevabdžija. A preposition is a separate word and keeps its spelling: iz kuće (said “is kuće”).

The main exception: d before s and š

d before s and š keeps its spelling, though it loses voice in speech: gradski, ljudski, predsednik, odseliti, odštampati.

What Serbian does not do

A final voiced consonant is not devoiced (unlike German Rad = Rat): grad ends in a real d, not “grat”; nož in the s of pleasure, not “sh”; Beograd in a voiced d. The law works only where consonants meet.