This article includes a list ofgeneral references, butit lacks sufficient correspondinginline citations. Please help toimprove this article byintroducing more precise citations.(June 2018) (Learn how and when to remove this message)

Rendaku (連濁,Japanese pronunciation:[ɾendakɯ],lit. 'sequential voicing') is a pronunciation change seen in somecompound words inJapanese. When rendaku occurs, avoiceless consonant (such as/tksh/) isreplaced with a voiced consonant (such as/dɡzb/) at the start of the second (or later) part of the compound. For example, themorphemekami (paper) starts with the voiceless consonant/k/ when used as an independent word or as the first part of a compound word, but this is replaced with the corresponding voiced consonant/ɡ/ in the compound wordorigami, fromori (fold) +kami.

Rendaku is common in Japanese, but does not occur in all compound words. It is regularly blocked in certain circumstances. The most important phonological condition blocking rendaku is known as Lyman's law, which prevents rendaku if the second element of the compound word already contains one or more of the voiced obstruent consonant phonemes/dɡzb/. Rendaku is also systematically blocked if the second element of a compound is arecent loanword into Japanese from a foreign language. Furthermore, rendaku may fail to occur even in contexts where no definite blocking factor is present.

In theJapanese writing system, rendaku affects how a morpheme is spelled when using one of thekana syllabaries: it causes thedakuten ("voicing mark", written as ゛) to be added to the upper right corner of the kana character representing the first consonant and vowel in the second element of the compound. For example,kami andorigami are spelled inhiragana asかみ andおりがみ, respectively. Rendaku is not marked in writing when a morpheme is spelled usingkanji (logographs taken fromChinese characters). For example,kami (paper) is written with the kanji character紙, which is unchanged when used in the spelling oforigami (折り紙).

Linguistically, since rendaku involves aspects of both pronunciation (phonology) and word structure (morphology), it is categorized as amorphophonological phenomenon.

Rendaku replaces avoicelessobstruent consonant with a voiced consonant sound. For example, the voiceless alveolar plosive[t] becomes the voiced alveolar plosive[d] in the context of rendaku. In this case, the new consonant retains the same manner and place of articulation as the original consonant. However, rendaku can also cause additional changes depending on the sounds involved.

In the context ofJapanese phonology, some consonant sounds that seem distinct to English speakers, and that have distinct spellings inHepburn romanization, are analyzed asallophones (contextual variants of aphoneme). The following table describes the effects of rendaku in the standard variety of Japanese, using both phonemic transcriptions (marked by slashes, such as/s/) andphonetic transcriptions (marked by square brackets, such as[ɕ]).

Depending on accent, the voiced velar nasal[ŋ]may be used in Japanese as an alternative to the voiced velar stop[ɡ] when the consonant occurs in the middle of a word. These two sounds are typically analyzed as allophones of a single phoneme/ɡ/. In accents that use[ŋ] like this,[k] is replaced with[ŋ] in the context of rendaku.

The voiceless affricates[tstɕ] are commonly analyzed as allophones of/t/. The phonemic analysis of their voiced outcomes is complicated by a phonetic merger. Originally,[tstɕ] corresponded to voiced affricate sounds[dzdʑ], whereas[sɕ] corresponded to voiced fricative sounds[zʑ]. However, the historical distinction between[dzdʑ] and[zʑ] has been lost in the Tokyo-based standard (though not in all regional varieties of Japanese). The merged sounds may be pronounced either as voiced affricates or voiced fricatives, regardless of etymology (seeyotsugana). As a result of this merge, the phonetic sequence[(d)zɯ] represents a neutralization of historical/du/ and/zu/, and[(d)ʑi] represents a neutralization of historical/di/ and/zi/.

In the context of rendaku, dialects with the merger may be analyzed as possessing an underlying phonemic distinction between/dudi/ and/zuzi/ that becomes neutralized on the phonetic level as[dzɯdʑi].^[2] When morphemes that begin with the moraechi (ち/チ) andtsu (つ/ツ) undergo rendaku, the resulting moraeji andzu are generally spelled with the kanaぢ/ヂ andづ/ヅ, rather than the identically pronouncedじ/ジ andず/ズ. This is not a strict rule, however, and is relaxed in certain older compounds or names, especially those that are not easily recognized as compounds.

It is possible to characterize rendaku in terms of its effect on a morpheme's Japanesekana spelling: it adds thedakuten (voicing mark) to the first kana of the affected morpheme.^[3] The relevantgraphemes are shown in the tables below (excludingyōon digraphs, which are formed by taking an i-column kana and placing a smallya, yu, oryo kana after it).

The following table shows an example of rendaku for each major allophone of the eligible consonant sounds:

Rendaku occurs most frequently in compounds where the second element is a single morpheme ofnative Japanese origin that does not already contain a voiced obstruent phoneme. However, even though rendaku is usual in this context, it does not invariably occur: there are numerous exceptions pronounced without rendaku. Some compound words are pronounced with rendaku by some speakers, but without it by other speakers.

Rendaku is blocked when a voiced obstruent phoneme is already present in the second element of the compound. This rule, called Lyman’s law, is highly reliable, with only a small number of exceptions.^[9]

Rendaku is also blocked if the second element is a recent loanword from a language other than Chinese (gairaigo). This rule likewise has few exceptions.^[10]

Rendaku does not affect mostSino-Japanese elements, but this tendency is less consistent. A substantial minority do undergo rendaku as the second element of a compound. It has been speculated that Sino-Japanese elements that can undergo rendaku might have become "vulgarized", that is, adopted into the same category of vocabulary as native Japanese morphemes.^[11][12]

Other rules have been proposed regarding circumstances where rendaku is either blocked or favored, but often, it is only possible to identify tendencies rather than inviolable rules.

Lyman's law is the constraint thatrendaku cannot occur if the second element already contains a voiced obstruent phoneme (that is, if it contains any of the consonant phonemes/dɡzb/, including theirallophones such as[dʑ] and[ŋ]; all of these sounds are written with thedakuten and called "muddy sounds" (濁音,dakuon) in Japanese). This is considered to be one of the most fundamental of the rules governingrendaku.

yama +kado →Yamakado (surname)山門（やまかど） ("mountain" + "gate" → place name). Sincekado already contains a voiced obstruent consonant phoneme/d/, its initial consonant/k/ does not undergo rendaku, so the form *Yamagado やまがど does not exist (* indicates a non-existent form).

hitori +tabi →hitoritabi ("one person" + "travel" → "traveling alone"), not *hitoridabi. Sincetabi already contains a voiced obstruent consonant phoneme/b/, its initial consonant/t/ does not undergo rendaku.

yama +kaji →yamakaji ("mountain" + "fire" → "mountain fire"), not *yamagaji. Sincekaji already contains a voiced obstruent consonant phoneme/z/ (in the form of its allophonej[(d)ʑ]), its initial consonant/k/ does not undergo rendaku.

The precise formulation of Lyman's law varies between analysts. In particular, there are conflicting viewpoints on the size of the phonological domain in which it applies. One formulation states thatrendaku is blocked by the presence of voiced obstruent consonant "within a morpheme"; this may be interpreted as a consequence of a more general constraint that applies also to the underived form of native Japanese morphemes, which do not generally contain more than one voiced obstruent phoneme.^[13] An alternative view is that Lyman's law applies whenever a voiced obstruent consonant occurs anywhere within the second element of a compound, and so might also include cases where this element is composed of more than one morpheme. Rendaku occasionally (although relatively infrequently) causes voicing of the initial consonant of aSino-Japanese (kango) lexeme that is written with two kanji, and is in etymological terms composed of two Sino-Japanese roots;^[14] it is debatable whether such lexemes count as one morpheme or two from asynchronic perspective.^[15]

In modern Japanese, the presence of a voiced obstruent phoneme in the first element does not normally block rendaku, as demonstrated by examples such assode (そで, sleeve) +kuchi (くち, mouth) →sodeguchi (そでぐち, cuff).^[16] Nevertheless, it has been proposed that in certain circumstances, Lyman's law might be (or might once have been) sensitive to the presence of a voiced obstruent in the first element of a compound. Compound personal names ending in the elementta 'rice field' seem to usually show rendaku when the final mora of the first element contains/tksmn/ (e.g.Katada, Fukuda, Asada, Hamada, Sanada), but never show rendaku when the final mora of the first element contains/dɡzbj/ (e.g.Kadota, Nagata, Mizuta, Kubota, Hayata) (and usually do not show rendaku when it contains/rw/). The pattern of voicing seen in compounds like these may in part be a residue of an older version of the law that operated in Old Japanese.^[17] Examination of Old Japanese compounds suggests that Old Japanese had a constraint against two consecutive syllables starting with aprenasalized consonant (the source of modern Japanesedakuon/voiced obstruent phonemes), but over time, this constraint came to be replaced with the modern tendency for adakuon consonant to block rendaku only when it occurs in the second element of the compound.^[18]

Some formulations of the law treat it as applicable only in cases where thedakuon/voiced obstruent phoneme is the second consonant of the final element of the compound.^[4] In the case of Old Japanese, there is not enough evidence to conclude whether the Old Japanese version of Lyman's law applied to morphemes containing a prenasalized consonant in their third or later syllable.^[19] In modern Japanese, however, there is evidence that Lyman's law applies in general to morphemes of this form, as indicated by the lack of rendaku in examples such astsuno (つの, horn) +tokage (とかげ, lizard) →tsunotokage (つのとかげ, horned toad) orko (こ, child) +hitsuji (ひつじ, sheep) →kohitsuji (こひつじ, lamb).^[20] There are only a handful of exceptions, such asnawa (なわ, rope) +hashigo (はしご, ladder) →nawabashigo (なわばしご, rope ladder), where voicing occurs despite the presence of adakuon/voiced obstruent consonant in the second element of the compound.^[21][22]

Although this law is named afterBenjamin Smith Lyman, who independently propounded it in 1894, it is really a re-discovery. TheEdo period linguistsKamo no Mabuchi^[23][24] (1765) andMotoori Norinaga^[25][26] (1767–1798) separately and independently identified the law during the 18th century.

Another important factor affecting the likelihood of rendaku is the etymological source orlexical stratum of the second element of the compound. Rendaku frequently affectswago (native Japanese lexemes), infrequently affectskango (Sino-Japanese vocabulary), and very rarely affectsgairaigo (recent loanwords, such as borrowings from English). On the other hand, the lexical stratum of the first element of the compound is not relevant.^[27]

One possible reason for the resistance of Sino-Japanese morphemes to rendaku is the greater potential for it to cause homophony in this context. Native Japanese morphemes very rarely start with a voiced obstruent consonant, but this does not apply to Sino-Japanese; therefore, rendaku of Sino-Japanese morphemes is more likely to neutralize a contrast between distinct morphemes.^[28][12]

Sino-Japanese vocabulary is built from Sino-Japanese roots, which have a restricted phonological structure (one or two moras long). At least half or more of Sino-Japanese terms are "binoms" consisting of two roots, but some roots can be used on their own as words (mononoms).^[29] Each word is written with as manykanji (Chinese characters) as it has roots.

When used as the second element of a compound, most Sino-Japanese lexemes do not undergo rendaku. However, a minority (around 20% of mononoms, and 10% of binoms) do show rendaku in at least some compounds.^[30] No criteria have been identified that predict with 100% accuracy when this occurs.

In terms of phonology, rendaku is blocked by Lyman's law and so never affects a binom where the second root starts with a voiced obstruent phoneme; e.g.kōzō (構造) inshakaikōzō (社会構造, social structure),honzan (本山) insōhonzan (総本山, head temple).^[31][a] Based on a study of how native speakers pronounced novel compounds,Low 2009 concluded that rendaku is statistically less likely to affect a Sino-Japanese binom where the first root ends in the moraic nasal/N/ and the second root end in a voiceless obstruent, although it is not fully blocked in this context (the study's observed rate of rendaku in this context was 1.3%, compared to Vance 1996's dictionary-based estimate of rendaku affecting 10% of all Sino-Japanese binoms).^[35]

Sino-Japanese roots that start with voiceless obstruents may have variant pronunciations starting with voiced obstruent phonemes for other reasons aside from rendaku.

One reason is the existence of different readings of Sino-Japanese roots, corresponding in general to different time periods of borrowing. Two of the most important types of reading are termed Go-on and Kan-on.^[38] Certain Kan-on readings starting with/htsk/correspond regularly to Go-on readings starting with/bdzɡ/: these represent different adaptations of Early Middle Chinese voiced obstruent sounds, and so the existence of these alternative pronunciations is unrelated in origin to rendaku.

An example is the root地 'ground, land', which has both a Kan-on pronunciationchi[tɕi] and a Go-on pronunciationji[(d)ʑi]; for this reason, the use of the voiced pronunciation in compounds such asroji (路地, lane, alley) cannot necessarily be attributed to rendaku, since theji version of this root can also be found at the start of a word,^[39] e.g. injishin (地震, earthquake).

In addition, a minority of lexemes composed of two Sino-Japanese roots display a type of sequential voicing, affecting only roots in second position, that is indistinguishable in effect from rendaku, but probably has a distinct origin in terms of morphology. The relevant context is forms such as東国tōgoku (eastern provinces), which appears to display rendaku on its second element国koku (country). In this case, voicing is hypothesized to be the result of the preceding Sino-Japanese root東 originally ending in a nasal sound, which caused the following consonant to become prenasalized and voiced.^[39] Prenasalization and voicing of a consonant after a nasal sound (/ũ,ĩ,N/) is hypothesized to have been an active phonological rule up throughEarly Middle Japanese; inLate Middle Japanese,/ũ,ĩ/ were denasalized and voicing after/N/ ceased to be automatic.^[40]

Gairaigo vocabulary is usually not affected by rendaku when it occurs as the second element of a compound, as illustrated by the contrast betweengarasudana 'glass shelf', from native Japanesetana 'shelf', andgarasukēsu 'glass case', from foreignkēsu 'case': rendaku does not occur and is not expected to occur in the latter compound word, since its non-initial element belongings to foreign vocabulary.^[41] (In contrast, as shown bygarasudana, the use of agairaigo word as the first element of a compound does not prevent rendaku of a following native element.) Examples where rendaku affects agairaigo element of a compound are highly exceptional.^[b]

Rendaku affects compounds involving various parts of speech, such as Noun + Verb, Adjective + Noun, Verb + Noun, Verb + Adjective; however, it is rare in compounds of the form Verb + Verb.^[43]

Rendaku is very frequent in words formed byreduplication, other thanmimetic words, which are immune to rendaku.^[44]

Examples:

ひと + ひと → ひとびと (人々) (iteration)

hito +hito →hitobito ("person" + "person" → "people")^[45]

とき + とき → ときどき (時々) (iteration)

toki +toki →tokidoki ("time" + "time" → "sometimes")^[45]

Rendaku tends not to occur in non-reduplicative compounds which have thesemantic value of "X and Y" (so-calleddvandva or copulative compounds), as exemplified byyama +kawa >yamakawa "mountains and rivers", as opposed toyama +kawa >yamagawa "mountain river".^[9][46][47]

In compounds containing more than two elements (or rather, compounds where one element is itself a compound), thebranching structure of the compound may affect the application of rendaku. For example,todana, a compound of the morphemesto "door" andtana "shelf", retains its initial voiceless/t/ when used as the second element of the compoundfukurotodana "small cupboard". In examples like this, where the second element contains a voiced consonant as a result of rendaku, the lack of voicing at the start of the second element of the larger compound can potentially be explained as a consequence ofLyman's law.^[48]

Otsu 1980 proposed that rendaku is blocked in general in the left-branching elements of aright-branching compound, even in cases where Lyman's law does not apply. However, other linguists have questioned the validity or necessity of formulating such a constraint.^[49][50] The branching constraint is intended to explain the contrasting behavior of examples such as the following:

mon + [shiro +chō] >monshirochō, not *monjirochō ("family crest" + ["white" + "butterfly"] > "cabbage butterfly")

but

[o +shiro] +washi >ojirowashi (["tail" + "white"] + "eagle" > "white-tailed eagle")

This constraint does not apply to all words where the second element is composed of more than one morpheme. As discussed above, rendaku can affect Sino-Japanese "binoms" composed of two Sino-Japanese roots. Assuming the branching constraint is valid, it is possible it does not prevent rendaku in that context because Sino-Japanese binoms do not have the morphological status of compound words in the context of rendaku.^[49] In any case, there seem to be some counterexamples to the branching constraint, such asōburoshiki 'big talk', fromō- 'big' +furoshiki, fromfuro ' bath' +shiki 'carpet', ormachibikeshi 'fire brigade for common people', frommachi 'town' +hikeshi, fromhi 'fire' +keshi 'to extinguish'.^[51] One proposal accounts for such apparent counterexamples by postulating a distinction between "loose" compounds, which are hypothesized to be immune to rendaku per the branching rule when used as the right-hand element of a larger compound, and "strict" compounds, which can undergo rendaku the same way as single morphemes. However, it is not clear that there are reliable non-circular criteria for predicting whether a compound is "loose" or "strict".^[52] Other examples of rendaku affecting "multi-root" elements that are themselves composed of smaller elements include the following:^{[citation needed]}

おぼろ +つきよ → おぼろづきよ

oboro +tsukiyo →oborozukiyo ("haze" + "moonlit night" → "hazy moonlit night")

Here,tsukiyo is composed oftsuki ("moon") andyo ("night")

いろ +ちゃや → いろぢゃや

iro +chaya →irojaya ("lust" + "teahouse" → "brothel teahouse")

Here,chaya is composed ofcha ("tea") andya ("shop");cha by itself generally doesn't undergo rendaku, butchaya frequently does

The branching constraint analysis could be considered a violation of the Atom Condition, which states that "in lexical derivations from X, only features realized on X are accessible." An alternative view proposes that the process applies cyclically.

[nuri +hashi] +ire >nuribashiire ([lacquered chopstick] case, "case for lacquered chopsticks")

nuri + [hashi +ire] >nurihashiire (lacquered [chopstick case], "lacquered case for chopsticks")

This could be seen as the voicing between hashi and ire staying unrealized but still activating Lyman's Law.^[53]

Ito and Mester 2003 proposed a third account distinct from both Otsu 1980's branching-based constraint and Ito and Mester 1986's cyclical account. This hypothesis holds that the relationship between branching and rendaku is not direct, but is mediated byprosodic structure: the lack of rendaku in right-branching compounds such ashatsu kao awase is analyzed as a consequence ofkao coming at the start of a prosodic word.^[54] Per Ito and Mester 2007, whether a compound is treated as one or as multiple prosodic words is affected by the length of the second element of the compound: if the second element is longer than four moras (or two bimoraic feet), then the compound is required to have the prosody of a phrase rather than a single word.^[55]

Despite a number of rules which have been formulated to help explain the distribution of the effect ofrendaku, there still remain many examples of words in whichrendaku manifests in ways currently unpredictable. Some instances are linked with a lexical property as noted above but others may obey laws yet to be discovered.Rendaku thus remains partially unpredictable, sometimes presenting a problem even to native speakers,^{[citation needed]} particularly inJapanese names, whererendaku occurs or fails to occur often without obvious cause. In many cases, an identically written name may either have or not haverendaku, depending on the person. For example,中田 may be read in a number of ways, including bothNakata andNakada.

In some cases,rendaku varies depending on syntax. For instance, the suffixtōri (〜通り, "road, following"), fromtōru (通る, "to go, to follow"), is pronounced as-tōri (〜とおり) following theperfective verb, as inomotta-tōri (思った通り, "as I thought"), but is pronounced as-dōri (〜どおり, withrendaku) when following a noun, as inyotei-dōri (予定通り, "as planned, according to schedule") or, semantically differently – more concretely –Muromachi-dōri (室町通, "Muromachi Street").

It has been hypothesized that morphemes are inherently either more susceptible or more resistant to being voiced when used as the second element of a compound; in other words, that the propensity of a morpheme to undergo rendaku is lexically specified. Some morphemes show voicing in all compounds in which they occur. At the far end of the spectrum, a small percentage of morphemes appear to be immune to rendaku (for reasons not explained by the phonological, morphosyntactic, and lexical-stratum factors described above that regularly block rendaku). Other morphemes are intermediate and undergo voicing in some but not all compounds.Irwin 2016 found that out of elements that occurred at least 5 times in second position in a database of frequent compound words, the most common behavior (42%) was to undergo voicing in all compounds, and each range of greater resistance to rendaku contained successively fewer elements (35% had voicing rates from 0.667 to 0.999; 12% had voicing rates from 0.334 to 0.666; 7% had voicing rates from 0.001 to 0.333, and 5% were never voiced).^[56]

Rosen 2016 suggests that the first element of compound words also has a lexically variable propensity to trigger rendaku, hypothesizing that rendaku occurs in words where the combination of these two voicing weights (the rendaku-triggering weight of the first element and the rendaku-undergoing weight of the second element) exceeds a threshold value, and fails to occur when this value falls below the threshold.^[57]

Rendaku seems to be avoided in non-verbal elements that start with[h] or[ç] + vowel +/m/ + vowel, such ashama, hamo, hima, hime, himo, hema, which has been explained as an effect of a preference against havinghomorganic consonants at the start of adjacent syllables: rendaku would replace[h] or[ç] with[b], which is bilabial like[m].^[58][59] Rendaku is seen before vowel +/m/ + vowel inkoibumi 'love letter' fromfumi 'letter': this is consistent with the hypothesis, since[ɸɯmi] contains[ɸ], which is already homorganic with[m] (and so avoiding rendaku would not make any difference).^[60] (An alternative explanation supposes thatfumi, as well as other words that undergo rendaku to/b/, starts with an underlying labial consonant that is phonologically distinct from the/h/ found at the start of lexemes that do not undergo rendaku.^[61])

A study observed this tendency as an active factor affecting the frequency with which Japanese speakers preferred rendaku vs. non-rendaku variants ofnonce words.^[62]

For whatever reason, this avoidance is not seen in verb roots (including deverbal nouns), which undergo rendaku even in cases such asmusibamu 'to corrupt'.^[58]

No comparable effect is seen with other consonant sounds that undergo rendaku, since they retain the same place of articulation after the change.

Although rendaku does not occur in all compounds, it is common and studies of Japanese native speakers show that it occurs often in newly formed compounds. It has been suggested that its productivity in new compounds is caused by analogy with existing compound words.^[63]

Some studies have examined thephonological development of rendaku. Studies from the late 1990s and early 2000s indicate that children under five and a half years old show limited use of rendaku in any context. The use of rendaku in established compounds becomes frequent by the age of six, but children below this age do not show reliable use of rendaku in unfamiliar compounds. Children over the age of six use rendaku frequently both in familiar and unfamiliar compounds.^[64]

In linguistics, rendaku has been analyzed in terms of variousphonological theories. It has been popular to identifyLyman's Law, the restriction against applying rendaku to a morpheme that already contains a voiced obstruent phoneme, as an example of a more general theoretical concept known as theobligatory contour principle (or "OCP" for short). This principle was originally formulated (by Leben 1973) to refer to the phonology of tone, and referred to a hypothesized constraint against having consecutive identical tones in the underlying representation of a morpheme. Later phonologists have interpreted the principle as a more general constraint, using it to refer also to bans on identical adjacent specifications of various non-tone features. Thus, Lyman's Law has been interpreted as an example where the obligatory contour principle (OCP) applies to voicing specifications and rules out multiple occurrences of voicing within a morpheme.^[65]

The voiced obstruent consonants of modern Japanese go back to prenasalized voiced obstruents ofOld Japanese.^[66] Rendaku may have originated from the fusion of consonants with preceding nasal sounds derived from reduction of the genitive postpositional particleno (の)^[67] or the dative postpositional particleni (に): for example, according to this hypothesis, a form such asyamadori (やまどり, copper pheasant) might go back to an originalyama-no-tori "mountain-GEN bird".^[68] This explanation could help account for why rendaku is not found consistently in all compound words: if some compounds were originally formed withno orni, but others were formed with simple juxtaposition of two roots, then rendaku would be expected to have arisen only in the first category of compounds, but not in the second.^[69][70]

Whatever its origin, by the Old Japanese period rendaku had already become grammatically distinct from constructions withno orni, as shown by the occurrence of forms such asnadori "your bird", where a phrase withno would be ungrammatical, since the genitive of the pronounna "you" was always formed with the particlega.^[71]

In the Kahokudialect of Tōhoku, the phonemes/tktstɕ/ have voiced allophones[dɡdzdʑ] when they come between voiced vowels, and the phonemes/bdɡzdʑ/ have prenasalized or nasal allophones[ᵐbⁿdᵑɡ~ŋⁿdzⁿdʑ] when they come between vowels: for example,/mato/ 'target' is pronounced[mɑdo] (with a second consonant that is voiced but not prenasalized) whereas/mado/ 'window' is pronounced[mɑ̃ⁿdo] (with both voicing and prenasalization of the second consonant).^[72] Thus, rendaku in this dialect, when it occurs, typically involves phonetic prenasalization of the second element of the compound, as is speculated to have been the case in prehistoric Japanese: e.g./hama/ 'beach" +/kuri/ 'chestnut' →/hamaɡuri/[hɑmɑ̃ᵑɡɯɾɨ] 'clam'.^[73] However, the use of prenasalization at the start of the second element of a compound word is not uniform, and depending on the speakers and the words pronounced, significant variations were observed.^[74] There was a relationship between the rate of prenasalization and the speakers’ age: older individuals use prenasalized pronunciations at a higher rate than younger individuals. On the other hand, differences in the speakers’ gender and socioeconomic status did not affect the rate of prenasalization.^[74] It is speculated that dialect mixing with standard Japanese may have increased the use of pronunciations without prenasalization.^[75] In the case of/b/[ᵐb], loss of prenasalization may also have additionally been facilitated by the fact that the resulting sound[b] did not merge with any preexisting phoneme, since/p/ does not occur intervocalically in non-mimetic vocabulary (in contrast, in forms like [te̝ɡɑmɨ], the absence of prenasalization makes it possible to interpret the consonant[ɡ] as merely the voiced intervocalic allophone of/k/, which would result in the phonemic transcription/tekami/, with the same consonant phoneme as the independent word/kami/).^[76]

A Tōhoku dialect spoken in Iwate Prefecture has been reported to have a phonotactic constraint prohibiting prenasalized obstruents from occurring in two successive syllables; this rule produces regular denasalization in[kɑ̃ᵐbɑbɨ] 'ceremonial fire' as opposed to[hɑnɑ̃ᵐbi] 'fireworks'.^[77]

• Consonant mutation
• Lenition
• Sandhi

• Dexter, Kristen (14 August 2018)."Rendaku: Why Hito-Bito isn't Hito-Hito".Tofugu.
• Endō, Kunimoto (1981). "Hi-rendaku no Hōsoku no Shōchō to Sono Imi: Dakushiin to Bion to no Kankei kara"非連濁の法則の消長とその意味―濁子音と鼻音との関係から―.Kokugo Kokubun国語国文.50–3. (Japanese citation:遠藤邦基(1981)「非連濁の法則の消長とその意味―濁子音と鼻音との関係から―」(『国語国文』50-3))
• Frellesvig, Bjarke (2010).A History of the Japanese Language. Cambridge University Press.
• Fukuda, Suzy E. (2002). "The acquisition of rendaku: A linguistic investigation".Aoyama Journal of General Education.43:15–59.
• Fukuda, Suzy E.; Fukuda, Shinji (1999). "The operation of rendaku in the Japanese specifically language-impaired: A preliminary investigation".Folia Phoniatrica et Logopaedica.51 (1–2):36–54.doi:10.1159/000021480.
• Hirayama, Manami (2005). "Place asymmetry and markedness of labials in Japanese".Toronto Working Papers in Linguistics.24:125–168.
• Irwin, Mark (April 2005). "Rendaku-based Lexical Hierarchies in Japanese: The Behaviour of Sino-Japanese Mononoms in Hybrid Noun Compounds".Journal of East Asian Linguistics.14 (2):121–153.doi:10.1007/s10831-004-6306-9.ISSN 0925-8558.S2CID 55842710.
• Irwin, Mark (2016). "The Rendaku Database". In Vance, Timothy J.; Irwin, Mark (eds.).Sequential Voicing in Japanese: Papers from the NINJAL Rendaku Project. Studies in Language Companion Series. Vol. 176. John Benjamins.doi:10.1075/slcs.176.ISBN 978-90-272-5941-7.S2CID 64058981.
• Itō, Shingo (1928).Kinsei Kokugoshi. Ōsaka: Tachikawa Bunmeidō.
• Ito, Junko; Mester, Ralf-Armin (1986)."The Phonology of Voicing in Japanese: Theoretical Consequences for Morphological Accessibility"(PDF).Linguistic Inquiry.17. RetrievedApril 21, 2024.
• Ito, Junko; Mester, Armin (2003).Japanese Morphophonemics: Markedness and Word Structure. MIT Press.
• Ito, Junko; Mester, Armin (2007)."Prosodic Adjunction in Japanese Compounds"(PDF).MIT Working Papers in Linguistics.55:97–111.
• Kawahara, Shigeto (2015), "Can we use rendaku for phonological argumentation?",Linguistics Vanguard,1 (1):3–14,doi:10.1515/lingvan-2015-0001
• Kawahara, Shigeto (2018), "Phonology and orthography: The orthographic characterization of rendaku and Lyman's Law.",Glossa: A Journal of General Linguistics,3 (1):1–24,doi:10.5334/gjgl.368
• Kawahara, Shigeto; Ono, Hajime; Sudo, Kiyoshi (2006), "Consonant Cooccurrence Restrictions in Yamato Japanese",Japanese/Korean Linguistics,14:27–38
• Kubozono, Haruo (2005)."Rendaku: Its Domain and Linguistic Conditions"(PDF). In Jeroen van de Weijer; Kensuke Nanjo; Tetsuo Nishihara (eds.).Voicing in Japanese. Studies in Generative Grammar. Berlin, New York: Mouton de Gruyter. pp. 5–24.doi:10.1515/9783110197686.1.5.ISBN 978-3-11-018600-0. Archived fromthe original(PDF) on 2004-09-12.
• Kumagai, Gakuji (2020),Testing the OCP-labial effect on Japanese rendaku
• Labrune, Laurence (2012),The Phonology of Japanese, Oxford, England: Oxford University Press,ISBN 978-0-19-954583-4
• Low, James (2009).Issues in Rendaku: Solving the Nasal Paradox and Reevaluating Current Theories of Sequential Voicing in Japanese(PDF) (Senior thesis in linguistics thesis).Pomona College. Archived fromthe original(PDF) on 2016-03-03.
• Martin, Samuel E. (1987).The Japanese Language Through Time. New Haven: Yale University Press.ISBN 0-300-03729-5.
• Miyashita, Mizuki; Irwin, Mark; Wilson, Ian; Vance, Timothy J. (2016-06-13). "Rendaku in Tōhoku Japanese: The Kahoku-chō Survey". In Vance, Timothy J.; Irwin, Mark (eds.).Sequential Voicing in Japanese. Studies in Language Companion Series. Vol. 176.doi:10.1075/slcs.176.ISBN 978-90-272-5941-7.ISSN 0165-7763.
• Ohno, Kazutoshi (2000), "The lexical nature of rendaku in Japanese", in Nakayama, Mineharu; Quinn, Charles J. Jr. (eds.),Japanese/Korean linguistics 9, Center for the Study of Language and Information, Stanford University, pp. 151–164
• Ota, Mitsuhiko (2015), "L1 phonology: phonological development", in Kubozono, Haruo (ed.),Handbook of Japanese Phonetics and Phonology, Berlin: De Gruyter, pp. 681–718,doi:10.1515/9781614511984.681,ISBN 978-1-61451-252-3
• Otsu, Yukio (1980). "Some aspects of rendaku in Japanese and related problems".MIT Working Papers in Linguistics: Theoretical Issues in Japanese Linguistics 2:207–227.
• Rice, Keren (2002)."Sequential voicing, postnasal voicing, and Lyman's Law revisited*"(PDF). University of Toronto. Archived fromthe original(PDF) on 2016-03-03. Retrieved2015-03-19.

• Rosen, Eric (2016),"Predicting the unpredictable: Capturing the apparent semi-regularity of rendaku voicing in Japanese through harmonic grammar"(PDF),Proceedings of the Berkeley Linguistics Society,42:235–249
• Shibatani, Masayoshi (1990).The Languages of Japan. Cambridge; New York: Cambridge University Press. pp. 173–175.ISBN 0-521-36918-5.
• Suzuki, Yutaka (2004). "'Rendaku' no Koshō ga Kakuritsu suru made: Rendaku Kenkyū Zenshi"「連濁」の呼称が確立するまで―連濁研究前史―.Kokubungaku kenkyū国文学研究.142. Japanese citation:鈴木豊(2004)「「連濁」の呼称が確立するまで―連濁研究前史―」(『国文学研究』142)
• Uwano, Zendō (2015), "Bidakuon 2-dai",NINJAL Colloquium Series, Tokyo
• Vance, Timothy J. (1982), "On the Origin of Voicing Alteration in Japanese Consonants",Journal of the American Oriental Society,102 (2):333–341,doi:10.2307/602532,JSTOR 602532
• Vance, Timothy J. (1987).An Introduction to Japanese Phonology. Albany, N.Y.: State University of New York Press.ISBN 0-88706-361-6.
• Vance, Timothy J. (2015), "Rendaku", in Kubozono, Haruo (ed.),Handbook of Japanese Phonetics and Phonology, Berlin: De Gruyter, pp. 397–442,doi:10.1515/9781614511984.397,ISBN 978-1-61451-252-3
• Vance, Timothy J. (2022).Irregular Phonological Marking of Japanese Compounds. De Gruyter.
• Weijer, Jeroen van de (2005). Kensuke Nanjo; Tetsuo Nishihara (eds.).Voicing in Japanese. Berlin, New York: Mouton de Gruyter.ISBN 978-3-11-018600-0.
• Yamaguchi, Yoshinori (1988). "Kodaigo no Fukugō ni Kansuru Ichi-kōsatsu: Rendaku o Megutte"古代語の複合語に関する一考察―連濁をめぐって―.Nihongo-gaku日本語学 [Japanese Linguistics].7–5. (Japanese citation:山口佳紀(1988)「古代語の複合語に関する一考察―連濁をめぐって―」(『日本語学』7-5))

• Suzuki Yutaka."Exceptions to Lyman's Law"(PDF). Archived fromthe original(PDF) on 2011-08-13.
• Vance, Timothy J.; Irwin, Mark, eds. (2016).Sequential Voicing in Japanese: Papers from the NINJAL Rendaku Project. Studies in Language Companion Series. Vol. 176. John Benjamins.doi:10.1075/slcs.176.ISBN 978-90-272-5941-7.S2CID 64058981.

Look upCategory:Japanese terms with rendaku in Wiktionary, the free dictionary.

Wiktionary has a category onJapanese terms with rendaku.

• The Japanese Lexicon: A Rendaku EncyclopediaArchived 2016-08-25 at theWayback Machine –National Institute for Japanese Language and Linguistics

• Japanese phonology
• Japanese writing system terms

• Articles containing Japanese-language text
• Pages with plain IPA
• Articles containing Late Middle Chinese-language text
• Articles containing Portuguese-language text
• Articles with short description
• Short description is different from Wikidata
• Articles lacking in-text citations from June 2018
• All articles lacking in-text citations
• Pages with Japanese IPA
• All articles with unsourced statements
• Articles with unsourced statements from December 2024
• Articles with unsourced statements from January 2019
• Articles containing Old Japanese-language text
• CS1 uses Japanese-language script (ja)
• Webarchive template wayback links