Dybo's law, or Dybo–Illich-Svitych's law, is a Common Slavic accent law named after Soviet accentologists Vladimir Dybo and Vladislav Illich-Svitych. It was posited to explain the occurrence of nouns and verbs in Slavic languages which are invariantly accented on the inflectional ending. The latter is seen as an innovation from the original Proto-Balto-Slavic accent system, in which nouns and verbs either had invariable accent on the root, or "mobile" accent which could alternate between root and ending in the inflectional paradigm.^[1]

This article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these template messages) This article needs attention from an expert in Linguistics. The specific problem is: incomprehensible to a non-expert. (March 2020) This article needs additional citations for verification. (March 2020)

According to the law, the accent was shifted rightward from a non-acute syllable (i.e. a long circumflex syllable, or a short syllable) to the following syllable if the word belonged to the fixed accentual paradigm. This produced the difference between the later accent classes A and B. The length of the previously-accented syllable remains.

The preservation of the original length is the primary source of pre-tonic length in the later Slavic languages (e.g. Serbo-Croatian), because inherited Balto-Slavic vowel length had previously been shortened in pre-tonic syllables, without a change in vowel quality. This caused the phonemicization of the previously automatic quality variations between short and long vowels — e.g. short *o vs. originally long *a.

Word-final syllables with the Balto-Slavic acute register were shortened and then lost the acute, before the time Dybo's law operated. It could then be lengthened again by Van Wijk's law, producing a long non-acuted vowel. Therefore, when the accent shifted onto a final syllable, the new accent was either circumflex or short, but never acute.

When the new accent was circumflex or fell on a yer, the accent was subsequently shifted leftward again by Ivšić's law (a.k.a. Stang's law), resulting in a neoacute accent:

• Early Slavic *pírstu "finger" > MCS *pь̑rstъ > (Dybo's law) *pь̄rstъ̀ > (Ivšić's law) LCS *pь̃rstъ > obsolete Russian perst, gen sg perstá
• Early Slavic *kátu "cat" > MCS *kȍtъ > (Dybo's law) *kotъ̀ > (Ivšić's law) LCS *kõtъ> Russian kot, gen sg kotá
• Early Slavic *váljāˀ "will" > (loss of acute) MCS *vȏlja > (Van Wijk's law) *vȏljā > (Dybo's law) *vōljȃ > (Ivšić's law) LCS *võlja > Russian vólja, Slovak vôľa (note, Slovak ô [u̯o] specifically reflects neoacute)

When the new accent was short (either from an original short vowel or a shortened acute) and Van Wijk's law did not apply, then Ivšić's law had no effect, and the accent remained on the second syllable:

• Early Slavic *génāˀ "woman" > (loss of acute) Middle Common Slavic (MCS) *žȅna > (Dybo's law) Late Common Slavic (LCS) *ženà > Chakavian ženȁ
• Latin vīnum "wine" > MCS *vȋno > (Dybo's law) LCS *vīnò > Serbo-Croatian víno (with long rising accent, indicating neoshtokavian accent retraction from the following syllable)

The acute was preserved in non-final syllables, however. Thus, when shifting onto a non-final syllable, all three accents (acute, circumflex or short) were possible, depending on the state of the syllable before Dybo's law operated.^[2] Ivšić's law in turn operated on forms where the new accent was circumflex, but not where it was acute or short. Compare:

• Early Slavic infinitive *prásīˀtī > MCS *prósīˀtī > (Dybo's law) LCS *prosi̋ti.
• Early Slavic 3sg. present *prásīti > MCS *prósītь > (Dybo's law) LCS *prosȋtь > (Ivšić's law) *pròsitь.

Dybo's law was entirely prevented in cases of initial accent in words belonging to the mobile accent paradigm. In such forms, Meillet's law resulted in loss of the acute register on the root, so that all initial-accented mobile forms were in principle susceptible to Dybo's law. Jasanoff argues that such forms had a special "left-marginal accent", which was not affected by Dybo's law the way the "lexical accent" of fixed-accent paradigms was.^[3] Thus:

• Early Slavic *vádān "water" acc. sg. > MCS *vȍdǫ > LCS *vȍdǫ > Russian vódu, SC vȍdu
• Early Slavic *gálˀvān "head" acc. sg. > (Meillet's law) MCS *gȏlvǫ > LCS *gȏlvǫ > Russian gólovu, SC glȃvu

In the valence theory, followed by the Moscow accentological school but otherwise not generally accepted, Dybo–Illich-Svitych's law is not considered single one-time change, but rather a succession of changes. It is described as a series of rightward accentual shifts in various Late Proto-Slavic dialects, with successive removal of accent drift prohibitions.^[4] There are two prohibitions, common to all Late Proto-Slavic dialects:^[5]

• Prohibition on the right-side accent shift from syllables with a dominant acute – *ˈža̋ba̋ AP (a), *ˈja̋goda̋ AP (a).^[5]
• Prohibition on the shift of the accent on syllables with a dominant circumflex – *ˈslũžĩtь AP (b₁), *ˈnòsĩtь AP (b₁), *ˈnòžĩkъ AP (b).^[5]

Early Proto-Slavic (most likely Balto-Slavic) is also considered a shift of accent on internal syllables, as well as on some endings with a dominant aсutе – *žèˈna̋ AP (b), *tvòˈri̋ti AP (b₂), *slũˈži̋ti AP (b₁).^[5] This process refers to the Fortunatov–de Saussure's law.^[6]

Fortunatov–de Saussure's law is a sound law very similar to Dybo's that affected Lithuanian. Like Dybo's law, it caused a rightward shift of the accent from non-acuted syllables and a split in the original accentual paradigms. There are some differences, however:

• Dybo's law shifted the accent rightward regardless of what was in the next syllable, whereas De Saussure's law shifted it only when it was acuted.
• Dybo's law was blocked in mobile-accented words. Such words never had an acute in the first syllable as a result of Meillet's law, but they nonetheless retained the initial accent. Consequently, there was only a split into three paradigms, with the fixed-accented words splitting into paradigms a and b, but the mobile-accented words remaining unified under paradigm c.