This chapter provides a description of the various dialects of Ryukyu-substrate Japanese, that is, informal varieties of Japanese with substratal influence from Ryukyuan languages. Ryukyu-substrate Japanese has developed along with the language shift process over the last century as each generation adopts or discards certain features of their local language. Since younger generations have not acquired their local language, they assign new functions to linguistic items heard in the speech of their elders, or otherwise create new ones based on Ryukyuan word formation rules. The text summarizes the most important studies of geographical and generational variation in phonological, grammatical and lexical contact effects on the Japanese spoken in Amami, Okinawa, Miyako, Yaeyama and Yonaguni.
Over the last century, language contact between Japanese and Ryukyuan languages has led to the emergence of new varieties of Japanese. The Japanese adopted by Ryukyuan people during the course of language shift, far from being a faithful replica of a mainland variety, has continued to retain elements of local languages. The language varieties resulting from language shift are essentially dialects of Japanese, typically used in informal contexts. They are, however, of particular interest to linguists because, unlike mainland dialects, they show strong influence from languages other than Japanese – in this case Ryukyuan languages – in the form of a trace substratum. This substratum is in a constant process of renegotiation as young Ryukyuans continue to borrow and often assign new functions to words from a local language they never acquired.
In the international literature on language shift, there has been a lack of research on trace substratum effects on replacing languages. According to Heffernan (2006: 642–643), the few existing studies of other language shift cases have revealed that the emergence of new dialects is a product of the following factors: (1) variability such as the simplification of phonological contrasts; (2) the production of innovative forms, most likely due to a lack of standardized norms; and (3) the adoption of new variants by younger generations as a marker of local identity. These tendencies match the findings of research on Japanese varieties spoken in the Ryukyus.
There are subtle and complex differences among the substrate-influenced Japanese varieties across the islands on all levels of language analysis, from phonetics to discourse. Manifestation and awareness of substratal effects from Ryukyuan languages vary according to generation, geographical location, level of education, occupation, gender, environmental factors and individual experience (Takaesu 1994: 247; Takaesu 2005: 266). Differences are most marked between generations, however, and the geographical variation is becoming less noticeable among younger people. Substrate-influenced varieties now appear to be leveling towards those spoken in three cities: Naze in Amami, Naha in Okinawa and Ishigaki in Yaeyama (Nagata 1999: 226). Of these, the Naha variety is the most important and has a strong influence on others across the archipelago (Nagata 1996: 105).
This chapter focuses on generational and geographical variation among the Japanese varieties spoken in the main island groups – Amami, Okinawa, Miyako, Yaeyama and Yonaguni. The discussion summarizes the findings on substrative effects reported mainly in studies by Nagata Takashi, Motonaga Moriyasu and Takaesu Yoriko, which remain the most comprehensive to date. It follows their lead in terms of organization, with separate treatment of non-standard phonological, grammatical and lexical features.
The literature has tended to focus on the variety of Japanese spoken on Okinawa Island and the surrounding islets. This has most often been known by its name in the Okinawan language, uchinaa yamatuguchi, where uchinaa means “Okinawa”, yamatu means “Yamato”, i.e., mainland Japan, and -guchi corresponds roughly to “language”. Until recently, it has been rendered as “Okinawan Japanese” wherever it is mentioned in the very few studies written in English (e.g., Matsumori 1995; Ōsumi 2001). The reader should be careful not to confuse uchinaa yamatuguchi with uchinaaguchi, the Okinawan name for the obsolescing language spoken in the Okinawan island group.
One definition of uchinaa yamatuguchi frequently cited in the literature is from Takaesu (1994: 246), who describes it as “Standard Japanese subject to phonological, grammatical and lexical interference from local dialect that has emerged during the shift process.” Anderson (2015: 489) adds that it “comprises Okinawa-accented Japanese combined with up to 3% insertions of well-known Uchinaaguchi-related lexemes”, thus distinguishing substrate-influenced Japanese from (attempts at) code-switching with a higher proportion of Okinawan. The substratum also contains a small number of English loans due to the presence of US bases on the islands (Ōsumi 2001: 88–89; Karimata 2008: 59–62).
The geographical name “Okinawa” in the term “Okinawan Japanese” is potentially ambiguous. As pointed out by Karimata (2008: 63), strictly speaking it should denote Okinawa Island and the surrounding islets but for the sake of convenience it has also been interpreted as encompassing the whole of Okinawa Prefecture (e.g., Takaesu 2004). Such ambiguity has thus far been unproblematic since much of the research on substrate-influenced Japanese has been conducted in the Naha region and studies of the varieties of Japanese spoken in other parts of Okinawa Prefecture such as Yaeyama are few in number. Although not part of Okinawa Prefecture, the Amami island group is part of the Ryukyu Archipelago with its own language (Amamian), and the Amamian-influenced variety of Japanese spoken there goes by its local name, ton-futsūgo or hansuu-futsūgo (literally “Sweet Potato Japanese”).
In this chapter, the English terms used in Anderson (2015) are adopted to ensure clarity in respect of geographical variation and to reflect precisely the linguistic nature of these varieties of Japanese: “Ryukyu-substrate Japanese” is an umbrella term covering all substrate-influenced varieties of Japanese spoken across the archipelago, while local varieties specific to the island groups are referred to as “Amami-substrate Japanese”, “Okinawa-substrate Japanese”, “Miyako-substrate Japanese”, “Yaeyama-substrate Japanese” and “Yonaguni-substrate Japanese”. 1
Since age is the factor that seems to determine the nature of substratal effects most noticeably (Nagata 1996: 134), it is important to identify generational cohorts of speakers who display similar language repertoires, features and behaviors. This kind of categorization of speakers was first attempted by Yabiku (1963) and other scholars have also suggested generational cohorts (e.g., Agarie 1983; Takaesu 1994; Nagata 1996; Ōsumi 2001; Zayasu 2017), but those appear to be based on reported language repertoire and proficiency or the author’s subjective view. Anderson (2009) suggests four generational subgroups on the basis of text analysis of transcribed language in context. The table below summarizes the characteristics of each subgroup and the proportion of Okinawan-related words used in their Okinawa-substrate Japanese (the lower percentages shaded gray) and when code-switching (the higher percentages). 2
Source: Anderson 2009: 228, 238.
Interestingly, these four subgroups map almost exactly onto Hokama’s well-known four eras of Japanese language education:
Source: Hokama 1964: 65.
The four generational subgroups defined above will be referred to in the next section using the following abbreviations: full speakers (FS), rusty speakers (RS), semi-speakers (SS) and non-speakers (NS) of Ryukyuan languages.
Phonological differences from Standard Japanese (henceforth, SJ) are most evident in productive bilinguals who acquired a Ryukyuan language as their mother tongue. 3 As in many western dialects of Japanese, the SJ compressed /ɯ/ vowel is rounded closer to [u] across the generations and in all island groups with the exception of Amami. Since Ryukyuan short /u/ corresponds to Ryukyu-substrate Japanese /u/ and /o/ and Ryukyuan short /i/ corresponds to /i/ and /e/, hypercorrection may occur among productive bilinguals, yielding sowaɾu (SJ: sɯɰaɾɯ, “sit”) and semeɾu (SJ: ɕimeɾɯ, “comprise”), although this kind of L1 interference had already become rare decades ago (Motonaga 1984: 373). SJ long /oː/ is lowered to [ɔː], except in Amami (as far south as Okinoerabu), where it can be diphthongized to [oɯ], or [ou] and lengthening can be inconsistent, yielding hodoɯ (SJ: hoːdoː, “report”) and soɯɡoɯ (SJ: soːɡo, “mutual”) (Nagata 1996: 21–22). In Ryukyu-substrate Japanese, the long /eː/ in Sino-Japanese words is diphthongized to [ei], as in heiwa (SJ: heːɰa, “peace”), but this is thought to be influence from Kyushu Japanese (Takaesu 1994: 248). SJ single-mora words are lengthened by FS and RS everywhere except Amami due to substratal influence in words such as haː (SJ: ha, “tooth”) and kiː (SJ: ki, “tree”) (Nagata 1996: 24).
Across the Ryukyus, the pronunciation of Japanese among FS in particular is described as generally hard and constricted-sounding (Motonaga 1984: 371) or having a “choking” quality (Nagata 1999: 222). The extent to which this phenomenon manifests in certain environments allows a hearer to tell where a speaker is from. For example, under the influence of phonemic contrasts in the Amamian and Yonagunian languages, bilingual speakers of Amami-substrate Japanese and Yonaguni-substrate Japanese may use fortis (unaspirated, tense) variants of voiceless plosive and affricate phonemes such as /k/, /t/ and /tɕ/, as compared with their lenis (weakly aspirated, lax) counterparts. In Amami-substrate Japanese, fortis [kˀi] and [kˀɯ] morae are particularly noticeable (Nagata 1996: 22), while some examples in Yonaguni-substrate Japanese are ɸutˀa (SJ: ɸɯta, “lid”), nikˀu (SJ: nikɯ, “meat”) and tɕˀiː (SJ: tɕi, “blood”) (Nagata 1996: 48). It is probably the extra energy given to consonant production with tense glottal folds that underlies the tendency in most Ryukyu-substrate varieties to articulate tense semivowels, use fortis [pˀ] in foreign loanwords, e.g., pˀaɴ (SJ: paɴ, “bread”) and consistently denasalize intervocalic word-medial /ɡ/, e.g., kaɡami (SJ: kaŋami, “mirror”) (Motonaga 1984: 372).
Another phenomenon contributing to the general impression of “constrictedness” is the glottalization of word-initial vowels. This is one of the main substratal features separating northern (Amami and Okinawa) and southern (Miyako, Yaeyama and Yonaguni) varieties of Ryukyu-substrate Japanese (Takaesu 1994: 251). In SJ, glottalized and unglottalized word-initial vowels appear in free variation but in northern Ryukyu-substrate Japanese it is a consistent substrative effect, particularly noticeable in the productive bilingual generations. Items with this feature, such as ʔoto (SJ: oto ~ ˀoto, “sound”), contrast with those which were historically written with word-initial morae wi and wo. The latter are pronounced with so-called “gentle voicing onset”, that is, a slight glide with no glottalization, as in ʲiɾu (SJ: iɾɯ, “exist”) and ʷotoko (SJ: otoko, “man”). In all island groups, northern and southern, both the e mora and historical we mora are realized as [ʲe] word-initially and word-medially, e.g., ʲeiɡa (SJ: eːɡa, “film”) and maʲe (SJ: mae, “before”) (Nagata 1996). This glide on /e/ has the effect of palatalizing all preceding consonants, most noticeably the /se/ and /ze/ morae, which become [ɕe] and [dʑe] respectively, as in ɕeɴɕei (SJ: seɴseː, “teacher”) and kadʑe (SJ: kaze, “wind”). This phenomenon, too, tends to be limited to FS and, to a lesser extent, RS (Nagata 1996: 443). As well as palatalization, labialization may occur across the Ryukyus and is typically associated with FS. This phenomenon obtains specifically in Sino-Japanese words containing the mora /kwa/ or its voiced counterpart /ɡwa/, as in ʔokʷaɕi (SJ: okaɕi, “sweets”) and ɡʷaikoku (SJ: ɡaikokɯ, “foreign country”). In Amami-substrate Japanese, /kwa/ can be tense and unaspirated [kˀʷa] (Nagata 1996: 23).
In the speech of elderly speakers of Okinawa-substrate Japanese, particularly in rural areas like Kumejima (Motonaga 1984: 372), the SJ /ɕ/ and /s/ phonemes have merged into a single phoneme with two allophones, [ɕ] occurring before /i/ and /e/ and [s] occurring before /a/, /u/ and /o/, yielding saɕiɴ (SJ: ɕaɕiɴ, “photograph”) and sudʑiɴ (SJ: ɕɯdʑiɴ, “husband”). SJ voiced counterparts /dʑ/ and /z~dz/ can merge to /dʑ/ in elderly Okinawa-substrate Japanese speakers, yielding dʑuboɴ (SJ: zɯboɴ, “trousers”) and dʑɔːkiɴ (SJ: zoːkiɴ, “dustcloth”) (Takaesu 1994: 249). In northern Ryukyu-substrate Japanese, SJ affricates in morae [tsɯ] and [dzɯ]~[zɯ] are sometimes pronounced [tɯ] and [dɯ] (Amami) or [tu] and [du] (Okinawa), respectively (Nagata 1996). Further south in Miyako and Yaeyama, FS may pronounce SJ [tsɯ] with a high central unrounded vowel [ɨ] with fricativization (usually romanized as ï) due to substratal influence from the local languages, as in tsɨki (SJ: tsɯki, “moon”) and natsɨ (SJ: natsɯ, “summer”). The same vowel may also be seen in the SJ [sɯ] and [ɕi] morae, sometimes realized as [sɨ] in Miyako and Yaeyama (Hateruma), respectively (Nagata 1996).
Another noticeable feature of productive bilinguals’ Okinawa-substrate Japanese is the reversal of voicing features such that SJ voiced consonants can become voiceless word-initially or word-medially, e.g., taɾe (SJ: daɾe, “who”) and kaɾata (SJ: kaɾada, “body”) (Takaesu 1994: 250). Conversely, SJ voiceless consonants can become voiced, e.g., ɡani (SJ: kani, “crab”) and nuɡu (SJ: nɯkɯ, “omit”). Similar to this phenomenon is the redistribution of SJ phonemes /d/ and /ɾ/ in patterns that mirror local varieties of Okinawan. In Naha, alveolar tap [ɾ] appears in all environments, producing words such as koɾomo (SJ: kodomo, “child”) and ɾekiɾu (SJ: dekiɾɯ, “can do”), while in Shuri and Itoman the two sounds tend to be in complementary distribution, with [ɾ] appearing only word-medially and [d] appearing word-initially, as in daɴpu (SJ: ɾaɴpɯ, “lamp”) and daku (SJ: ɾakɯ, “comfortable”) (Motonaga 1984: 373). In the Ishigaki variety of Yaeyama-substrate Japanese, the influence of the local variety of Yaeyaman affects the pronunciation of SJ /ɾ/, which is realized as an alveolar approximant [ɹ] similar to that found in many varieties of English. This feature appears only in the speech of FS and RS, who may have non-standard pronunciation of words like ʲeɹi (SJ: eɾi, “collar”) (Nagata 1996: 40).
Ryukyu-substrate Japanese pitch accent and intonation are quite different from the standard. Prosodic features remain the clearest indicator of non-standardness and are persistent in younger generations (Heffernan 2006: 644). On the sentence level, so-called “High Rising Terminal” (HRT) intonation (similar to “upspeak” in some varieties of English) is a pan-Ryukyuan substratal feature that is stable across all generations. On the word level, a range of different pitch accent types obtain in the Ryukyuan languages, and FS tend to maintain the pitch accent patterns of their L1, even when speaking Japanese. 4 By contrast, RS and SS tend towards the so-called “accentless” or “one-pattern” types common in parts of southern Kyushu rather than adopting a standard pitch accent system (Nagata 1996: 58). 5 In the accentless type, all phonological words have a low tone until the final syllable, at which point the pitch rises. Hence, SJ áme (“rain”) and amé (“candy”) in isolation are both amé in Ryukyu-substrate Japanese. If there is an enclitic case marker, the high tone shifts to the grammatical particle, as in ame=ɡá (rain/candy=NOM). Phonology is based on syllables, not on morae, so when words end with a long vowel or /ɴ/, high pitch is maintained across the whole syllable nucleus and coda, as in kɔːtɕɔ́ː́ (“headmaster”) and ʲeiɡakáɴ́ (“cinema”) (Motonaga 1984: 374).
This section deals with non-standard meaning-making in Ryukyu-substrate Japanese arising from any transfer relating to the function/structure morphemes of Ryukyuan languages. It leaves aside content/lexical words for discussion in the next section. Some function words/morphemes are Japanese ones that have been repurposed while others have been transferred directly from Ryukyuan languages. 6 The transcription below is romanized as there is no need for phonetic comparison, and most examples are taken from Motonaga (1984), Takaesu (1994) and Nagata (1996). 7
First, let us consider non-standard particle usage. In SJ, nominative case marker ga and accusative case marker o undergo deletion when binding particles such as wa and mo are used, but in some Ryukyuan languages, such as Okinawan, nominative case markers ga and nu may precede and join with binding particles ya and n to form the contracted combinations gaa, gan, noo and nun. This has influenced Okinawa-substrate Japanese to the extent that the following sentences are possible:
According to Nohara (2005: 75), these particle combinations are becoming rare in Okinawa, even among FS, while Zayasu (2017: 308) finds that younger generations in Okinawa have completely lost this feature. Furthermore, it is absent across all generations in Ishigaki Yaeyama-substrate Japanese due to the lack of any such particle usage in Yaeyaman (Zayasu 2017: 316). In Miyakoan, an accusative topic marker ba can follow accusative case marker yu, and the ba has been retained in Miyako-substrate Japanese to highlight contrast between objects (Nagata 1996: 80):
In Amamian, ba is an accusative case marker, which can be transferred directly into Amami-substrate Japanese and used to emphasize an object (Nagata 1996: 70):
In Ryukyuan languages, a focus particle du (thought to derive from historical binding particle zo) is used for emphasis, rather like English cleft sentences or “not until” constructions. In order to retain this resource in their Japanese, the first productive bilinguals repurposed the Japanese ga particle so that it could follow case markers such as ga, ni, de, to and kara. Recent data from Zayasu (2017: 323) suggest that sentences similar to the following from Motonaga (1984: 375) still obtain across generations in parts of Yaeyama at least, but may have disappeared altogether in parts of Okinawa:
Across the Ryukyus, the case marker kara is used in a wider variety of ways than in SJ. In Okinawa-substrate Japanese, the -te kara form is used more frequently than in SJ as a non-sequential conjunctive expression, and the particle ni may be added. This construction is thought to be influenced by the conjunctive -aa ni form in Okinawan (Takaesu 1994: 275):
Furthermore, when dynamic actions take place in a location, kara is used where the location would be a direct object marked by o in SJ:
Zayasu (2017: 269) reports that older SS tend to use SJ o when the street is relatively close to the speaker. Another non-standard use of kara is as an instrumental case marker for means of transport where de would be used in SJ:
This non-standard use of kara is common across the Ryukyus but perhaps less so in Yaeyama. Zayasu (2017: 261) finds that it is not generally used by young people, although some older SS use it when referring to methods of transport which are not self-driven such as public transport or lifts from others. Although Nagata (1996: 70) claims it to be straightforward substratal influence (Amami-substrate Japanese kara having come from Amamian hara, for instance), Hirayama (1984: 75) warns against overstating the link to Ryukyuan languages, pointing out that the phenomenon is also observed in many mainland dialects. A similar instrumental usage of kara to mark information sources is commonly observed in younger generations and is claimed to be less common in FS and RS (Zayasu 2017: 261): 8
Bare case marking is much more common in Ryukyu-substrate Japanese than in SJ. One exception to this is Amami-substrate Japanese, in which ga and chi (from Amamian) may replace standard purposive and lative ni, respectively (Nagata 1996: 70):
However, omission of nominative ga, accusative o, genitive no, lative e, and locative ni often results in the production of sentences like the following in Ryukyu-substrate Japanese (Takaesu 1994: 254):
Note that in the last example above, the interactional particle ne(e) is used (particularly by females) as an interrogative marker where no would be used in SJ. It generally carries a falling or dipping tone and is common across the Ryukyus (Motonaga 1984: 376). This non-standard usage of ne(e) appears not to be substratal influence, however, since it bears no resemblance to the various interrogative markers used in Ryukyuan languages (Zayasu 2017: 343). On the other hand, non-standard usage of ne(e) with the volitional form to signal a speaker’s intention can be attributed to substratal influence from Ryukyuan languages. This use of ne(e) allows a speaker to avoid sounding abrupt and instead to relax the listener psychologically by unconsciously involving them in decision-making (Uchima 2002: 33). Hence, in Ryukyu-substrate Japanese, kakoo ne(e) (“I’ll write if that’s okay”) tends to signal intention while kakoo (“let’s write, shall we?”) would be used to invite an addressee to participate in a joint activity (Zayasu 2017: 345). It is thought that the use of ne(e) in the former constitutes substratal influence from “softening” interactional particles in Ryukyuan languages, e.g., kaka yii in Okinawan and kaka raa in the Ishigaki dialect of Yaeyaman both translate as “I’ll write, if that’s okay” (Takaesu 2002: 153; Zayasu 2017: 347).
Like ne(e), interactional particles (IP) sa(a) and yo(o) also have a broader range of functions in Ryukyu-substrate Japanese. In SJ, sa is usually used at the end of a dependent clause to hold the floor and maintain a listener’s attention, whereas yo is used to add exclamatory emphasis and signal an appropriate point for turn-taking at the end of an independent clause. These standard usages are also possible in Ryukyu-substrate Japanese, but the functions may also be reversed, such that yo takes the place of sa and sa replaces yo (Miyahira and Petrucci 2014):
Perhaps related to this particular use of yo is the non-standard standalone expression da kara yoo, which is used as a backchanneling device meaning “exactly!” or the recently emerged American slang expression “I know, right?!” As for sa(a), the sentence-final particle ne(e) can be added to involve the listener by seeking agreement, just as with yo ne in SJ. Note that, in the following example, Ryukyu-substrate Japanese sa(a) directly follows the adjectival noun without the need for the copula as in SJ iya da yo nee (Anderson 2009: 138).
To an extent, sentence-final expressions are specific to particular generations and geographic areas. Miyako-substrate Japanese has retained some unique features, perhaps because of Miyako’s relative isolation from migrants and tourists (Nagata 1996: 35). One example is the sentence-final expression sai ga, which resembles an exclamatory English tag question, e.g., kirei sai ga (SJ: kirei ja nai ka, “it’s pretty, isn’t it?!”) (Motonaga 1984: 376; Nagata 1996: 82). Nohara (1996) examines sentence-final expressions in Okinawa-substrate Japanese. He discusses the Okinawan-derived expression annii, which fulfills a similar function to sai ga, although in this case it is only observed in younger generations. For example, a young Okinawan might say yaa furaa annii (SJ: omae baka ja nai ka, “You’re an idiot, aren’t you?!”), based on the Okinawan expression ʔyaa furaa ʔarani used by older speakers. Other Okinawan-derived expressions only observed in young people’s speech are the assertive particle yashi in sake nonderu yashi (SJ: sake nonderu yo, “I’m drinking sake!”) and emotive particle with final vowel lengthening yasshii in aitsu atama ii yasshii (SJ: aitsu atama ii naa, “Wow, that guy’s clever!”), apparently derived from Okinawan copula plus sentence-final particle with geminate consonant ya ssaa. Young speakers of Amami-substrate Japanese do not use yashi or saa (Long 2013: 94).
Modal particles (MODP) are also used differently from SJ. The grammaticalization process of the explanatory particles no (da) and wake (da) is more advanced in Ryukyu-substrate Japanese than in SJ. They appear to function as sentence-final particles, having lost their nominalizing function in many environments. Instead of being followed by copula da/desu as in SJ, they may be preceded by copula da or de aru (sometimes contracted to daaru) (Takaesu 1994). Thus, such combinations as da no ni and da wake are permitted in Ryukyu-substrate Japanese, in contrast to SJ in which adnominal copula na must precede the particles in place of da, and de aru is a literary form. This weakening of the nominalizing function allows both no and its contracted form n to appear in the same phrase in different syntactic slots as in the following example, which would be rendered in SJ as hodohodo de ii n ja nai ka with a final question particle (Fujiki 2004: 15):
A similar meaning is conveyed by wake in statements and questions. Originally a noun meaning “reason” or “grounds” and serving a conclusive function in expressions of cause-effect, wake is undergoing grammaticalization and serving a more general explanatory function in SJ, too, but its modal use has spread across the Ryukyus from the Naha area and is particularly common among young females (Nagata 1996). Hence, in Ryukyu-substrate Japanese, the following example is less likely to be used for explicitly linking an action with a reason (“That’s why I saw a movie yesterday”) than it would be in SJ and instead serves to involve the hearer (Motonaga 1984: 376):
Explanatory wake may also appear in questions that elicit an explanation, where no (ka) would be more likely to be used in SJ. These questions carry rising intonation and are translatable into English as declarative constant polarity tag questions:
Explanatory wake is a direct translation of Ryukyuan baa, which is used in a similar way in Ryukyuan languages, e.g., ʔichuru baa yoo in Okinawan translates to iku wake saa (“S/he’s going, you see!”) in Okinawa-substrate Japanese. This baa morpheme was borrowed by SS to be used as an alternative to wake with a local slang flavor (Motonaga 1984: 376):
Like wake, baa may be preceded by the copula da in the speech of SS and NS. A combination of particles such as X da baa yo na has a similar meaning to SJ X na n da yo ne, but baa conveys something of the emphatic youth slang flavor of “totally/totes” in “I like totally died”. This usage of baa is looked upon unfavorably by older bilinguals and is considered to be “incorrect” language (Fujiki 2004: 31). It is common in Okinawa but less so in Amami and Miyako (Long 2013: 94; Nohara 1998: 261).
One modal particle common to all varieties of Ryukyu-substrate Japanese is non-standard usage of hazu, syntactically and semantically reflecting Okinawan haji, Miyakoan pazï, Amamian and Yaeyaman (Ishigaki) hazï, and Yaeyaman (Iriomote) hachi (Nagata 1996). In SJ, hazu is a partially grammaticalized noun meaning “expectation”. It is followed by the copula and expresses something that “should be” or “is supposed to be”, based on objective grounds such as a timetable or prior arrangement, e.g., SJ: shiken wa ashita no hazu da, “The test should be tomorrow” (according to the timetable). In Ryukyu-substrate Japanese, hazu is a fully grammaticalized particle which may follow the copula, and the basis for a speaker’s conjecture may be purely subjective (Takaesu 1994: 277).
Like wake, this use of hazu is very common among young female speakers across the Ryukyus (Kinjō and Shō 2000: 31). SJ expressions daroo, yoo da, soo da and rashii are generally covered by Ryukyu-substrate Japanese hazu, although SJ quotative particle tte is often used for hearsay, rendered as cchi in Amami-substrate Japanese under the influence of Amamian (Nagata 1996: 67). As in SJ, this quotative particle is also used for expressing thoughts among other functions (Long 2013: 91):
One modal particle which may be used in a non-standard way by bilinguals in Miyako is beki, a direct translation from Miyakoan carrying the sense of an arrangement or plan rather than the SJ meaning of “obligation” or “inevitability” (Motonaga 1984: 378):
Other kinds of non-standard modality and aspect are reflected in verb suffixes or auxiliaries. 10 One inflection used particularly often in narratives is the -yotta form. In Kyushu, -yotta expresses the progressive aspect, but this is not the case in Ryukyu-substrate Japanese; rather, it is a direct evidential marker indicating the speaker’s certainty of information acquired through witnessing an event taking place (usually visually). 11 The agent is most often a third party but may also be the speaker talking about seeing themselves in a dream (Yabiku 1987: 122) or recollecting their own past habits (Takaesu 2004: 320; Zayasu 2017: 382). The -yotta form is used cross-generationally and, in terms of substratal influence, semantically reflects the Okinawan evidential -utan form. For example, okiyotta (“I saw somebody wake/waking up”) in Okinawa-substrate Japanese corresponds to ʔukiyutan (= ʔukiitan) in Okinawan (Takaesu 1994: 266). Zayasu (2017: 398) discusses the possibility that Ishigaki-substrate Japanese -yotta might be influenced by the Ishigaki Yaeyaman evidential –tta form, but points out that it could simply have spread from Okinawa. Whatever the case, it appears not to have reached as far as Iriomote in Yaeyama by the mid-1990s (Nagata 1996: 100). Consider the following example:
The -yotta form can also be used in the 1st person to emphasize the subjectivity of verbs of perception (Zayasu 2017: 387):
When used with verbs of cognition and emotion, the -yotta form emphasizes the speaker’s perception of a third party’s mental state (Zayasu 2017: 388):
For older speakers, the -yotta form appears not to be appropriate in stative situation types where a process of change cannot be witnessed (Zayasu 2017: 389):
However, these kinds of sentences are permissible in the speech of some (mostly younger) speakers, and the first example may have a mirative interpretation (surprise at a sudden discovery) (Zayasu 2017: 393):
Furthermore, young people have been observed to use -yotta with stative verbs of existence, for which the -te iru form is inappropriate in both SJ and Ryukyu-substrate Japanese (Zayasu 2017: 391):
In young people’s speech, such expressions can also have a mirative interpretation where someone makes an unexpected appearance (Zayasu 2017: 393):
In Anderson (2015: 489) it was claimed that Ryukyu-substrate Japanese lacks a non-past form -yoru as there is in Kyushu. This may have been true when Motonaga (1979: 44) originally made that observation, but a non-past form has since emerged as a new linguistic resource, first mentioned in Takaesu (1994: 266). The -yoru form indicates that a speaker witnesses an action or change in the process of taking place or about to begin (Takaesu 2004: 319). This non-past form is interesting because it is not drawn directly from the substratum but rather derives from an existing substrate-influenced form.
The Ryukyu-substrate Japanese -te aru form (also contracted to -taaru) differs from the standard in terms of aspect and modality. In SJ, -te aru is used most often as a resultative with nominative case, best translated into English as passive “X is done” or “X has been done” and rarely as a perfect with accusative case or in an active sense of “Y has done X” (Jarkey 2003). The two uses are exemplified below:
In Ryukyu-substrate Japanese, the latter usage is the norm, with optional deletion of the accusative marker (Takaesu 1994: 267). The SJ nominative resultative, where the agent of a transitive verb is backgrounded and the patient focused on, is generally expressed in Ryukyu-substrate Japanese using the passive -rareru form, as in mado ga akerarete iru (or its contracted form akerareteeru) (Takaesu 2004: 326). In contrast to SJ, the Ryukyu-substrate Japanese -te aru construction (or its contracted form -taaru) often indicates indirect, inferential evidentiality, translatable as “Y must have done X”. Therefore, it is not usually used with a first person subject (Takaesu 2004: 324; Takaesu 2005: 267; Karimata 2006: 54). The past form -te atta (-taatta) could be translated as past perfect with evidential adverb “Y had presumably done X”. The -te aru and -te atta forms are used cross-generationally and, in terms of substratal influence, semantically reflect evidential forms in Ryukyuan languages which fulfill a similar role such as Okinawan -een/-eetan and Ishigaki Yaeyaman -een/-eeda (Zayasu 2017: 363). Like the -yotta form, however, -te aru is not used in all areas of the Ryukyus, being absent from Amami-substrate Japanese (Nagata 1996: 68).
Furthermore, perhaps because the contracted -taaru variant starts with -ta, it can be attached to the past tense morpheme of adjectives and the copula in the following ways (Karimata 2006: 55):
Note that, in the second example above, -te aru indicates mirativity (surprise at a sudden discovery) (Karimata 2008: 57), albeit different in quality from that expressed by -yotta. The -te aru form may also lend a perfect interpretation to states resulting from actions expressed by intransitive verbs for which the -te iru form would be used in SJ (Zayasu 2017: 361).
As with the -yotta form, some (mostly younger) speakers have been observed using -te aru with stative verbs of existence, for which the -te iru form is inappropriate in both SJ and Ryukyu-substrate Japanese (Zayasu 2017: 362):
In addition to -te aru, Okinawa-substrate Japanese has another perfect form -te nai, this time indicating completive or evaluative modality similar to SJ -te shimatta (Takaesu 1994: 267). It is coincidental that -te nai and its past counterpart -te nakatta constructions look like negative forms of SJ -te aru or -te iru; in fact, they derive from the -ti neen/neentan forms in Okinawan. Interpretation depends on context and intonation, and speakers will use adverbs such as moo (“already”) or mada (“still”) to clarify meaning, yet misunderstandings do still occur (Karimata 2006: 55). This semantic conflict with SJ perhaps explains why SS and NS avoid the substrate-influenced forms. The examples below show how they are used by productive bilinguals (Takaesu 2002: 158):
Two more examples of non-standard modal auxiliaries are -te minai and -te aruku (Takaesu 1994: 268). The -te minai form marks experiential aspect and is equivalent to SJ X shita koto ga nai (“haven’t done X before”). The -te aruku form marks frequentative aspect. The verb aruku means “walk” in SJ but in Okinawa-substrate Japanese it can also mean “commute” or “go back and forth” under the semantic influence of its Okinawan cognate ʔacchun. The grammaticalized variant has the sense of “go around doing X” or “do nothing but X”, making the following expressions possible:
Other Ryukyu-substrate Japanese verb inflections closely resemble those of western mainland dialects, particularly Kyushu varieties. Negatives may have the standard -nai/-nakatta endings but in informal situations -n(katta) or -ran(katta) suffixes may be used. The -ran endings yield non-standard verbs like seran (SJ: shinai, “not do”), but these are absent from Miyako-substrate Japanese, where senai is used instead (Nagata 1996: 74). Causatives tend to have -(r)asu suffixes (or -(s)asu in Miyako) as opposed to SJ -(s)aseru endings, but these do not appear to be substrate-influenced. In the Ryukyus, causatives are used in place of SJ benefactive -te morau constructions, so ikashita (“made somebody go”) is preferred to itte moratta (“had somebody go for me”) (Takaesu 2002: 159).
Imperative structures in Ryukyu-substrate Japanese uniformly take -(r)e endings on Type I consonant-stems under the influence of Ryukyuan -(r)i inflection. SJ -ro suffixes do not obtain on Type II vowel-stems. For example, most varieties have kake (SJ: kake, “write!”), tabere (SJ: tabero, “eat!”) and se(re) or shi(re) (SJ: shiro, “do!”). Miyako-substrate Japanese is an exception to this rule. For older bilingual speakers, Type II verbs take a -ro suffix, producing standard form okiro (SJ: okiro, “get up!”) (Nagata 1996: 74), most likely due to substratal influence from Miyakoan ukiru, rather than any special tendency towards SJ. In addition to the standard suru na negative imperative, Okinawa-substrate Japanese has a less abrupt-sounding -kee form, e.g., mado akenkee (SJ: mado o akeru na, “Don’t open the window!”) (Takaesu 2004: 311). This may derive from the -ki/-kee particle used with negative imperatives in Okinawan.
The volitional form is indicated by SJ -(y)oo suffixation in Okinawa and Miyako. However, Amami and Yaeyama inflection of Type II vowel-stems end in -roo, taberoo (SJ: tabeyoo, “let’s eat”) (Nagata 1996). In addition, the use of the form differs from SJ, and it can indicate intentional or hortative moods depending on context. Usage may also differ by generation or geographical location, and in some cases -yoo can serve one of the functions while -roo serves the other, or -roo may indicate a more informal register. The negative hortative form in Okinawa-substrate Japanese is quite unlike any functionally similar SJ expression and features a -koo suffix, e.g., mado akenkoo (SJ: mado o akeru no yosoo, “let’s not open the window”). According to Takaesu (2004: 310), this -koo ending shows the influence of Okinawan -ka suffixes in sanka (“let’s not do”) and yumanka (“let’s not read”).
Takaesu (1994: 263) reports that, in Okinawa-substrate Japanese, both imperative and volitional forms may take on imperfective aspect with the addition of -te-oku, which often lacks the preparative sense it has in SJ and instead emphasizes the durativity of the verb. This allows for the following contracted forms, which are sometimes difficult to translate into SJ:
Source: Takaesu 1994: 263.
In descriptions of substratal influence in Ryukyu-substrate Japanese, scholars have tended to treat content words (vocabulary) separately from function morphemes (grammar). Substrative effects on words with lexical content have been categorized in different ways (Motonaga 1984; Takaesu 1994) but it appears that, broadly speaking, transfer results in three main types of surface form: (1) direct transfer of Ryukyuan forms (some of which happen to have an identical SJ cognate); (2) coinage of hybrid forms which exist in neither Ryukyuan languages nor Japanese; and (3) direct translation of Ryukyuan meanings into their Japanese cognates. More specifically, (1) may include items that have undergone phonological change; (2) involves the creation of a neologism by fusing morphemes from different codes together with or without phonological change; and (3) involves semantic transfer from Ryukyuan languages with the addition of new senses to a Japanese cognate’s standard semantic domain. 12 Examples of these three categories are given below.
This field would benefit from more qualitative studies in the area of substratal influence from languages other than Okinawan, that is, from Amamian, Miyakoan, Yaeyaman and Yonagunian. Such studies could contrast substrative effects with code-switching among bilingual speakers of these languages, an area that has not yet been investigated. Some studies have documented semi- and rusty speakers’ use of Okinawan insertions in Okinawa-substrate Japanese (Kawamitsu 1992; Anderson 2009; Sugita 2014), but there is a need for research using recorded natural conversation that captures a wider range of expressions in use, such as those listed in Mabuigumi (1989, 1990), Haapuudan (2003) and Fujiki (2004).
Another under-researched area is the relationship between superstratal interference in the retreating languages and substrative effects in the replacing language. In other words, attention needs to be paid to the issue of whether some hybrid expressions start life as superstratal interference in Ryukyuan speech before passing into local varieties of Japanese and remaining in the substratum. Accordingly, there is a need for more studies of language loss (societal attrition) from Ryukyuan languages such as phonological conformity to Japanese, the loss of unique morphosyntactic features, the restriction of lexis to higher frequency items and register reduction to informal styles.
Finally, the study of Ryukyu-substrate Japanese needs to be broadened and made more accessible to an international readership. Studies of Ryukyu-substrate Japanese in emigrant communities, such as those in South America, are entirely absent from the literature. The Ryukyuan case has provided rich data, which would be of interest to scholars involved in other language shift cases around the world. It is hoped that, in the future, international cooperation between linguists will allow for cross-linguistic studies to compare findings relating to substratal influences on replacing languages during language shift.
I wish to thank Satoshi Mizutani for his assistance in sourcing material from Japan, Nerida Jarkey for imparting her knowledge of Japanese aspect and mood and Hideki Arakaki from the Kosamedō bookshop, whose third-party credit card payment services for international orders have been invaluable. I am especially grateful to my friends Raymond Gatt and Patricia Carew for their support during my research.
The pan-Ryukyuan variety of Japanese and substrate-influenced variants spoken in the Sakishima Islands remained nameless until Karimata (2006: 58) coined the terms “Ryukyu Creole Japanese” and “Miyako Creole Japanese”.
“Okinawan-related” means any substrate-influenced item with a surface form that differs from standard Japanese (wake would not count, for example).
Transcription is broad unless distinguishing between standard and non-standard pronunciation.
Local language pitch accent is said to be better preserved in Amami-substrate Japanese than in other substrate-influenced varieties, being stable even among the SS generation (Nagata 1996: 24).
Owing to the influence of television and school education, younger generations are able to imitate Tokyo pitch accent patterns when called upon to read aloud, for example (Nagata 1996: 58).
Ryukyu-substrate Japanese also has non-standard features which are used elsewhere in Japan and cannot be considered as substratal influence. These include irregular -i adjectives, -na adjectives and verbs which have been reanalyzed as regular -i adjectives, yielding iikunai (SJ: yokunai “not good”), sukikunakatta (SJ: suki ja nakatta “didn’t like”), kirekuaru (SJ: kirei da “pretty/clean”), chigaukunai (SJ: chigawanai “not different”), genki mitakatta (genki mitai datta “seemed in good health”).
Since the standard Hepburn romanization system cannot adequately represent Ryukyuan varieties because of rephonemization, it has been slightly adapted for this chapter. The morae pronounced [ti], [tɕi], [di] and [dʑi] in Ryukyuan words are transcribed consistently as <ti>, <chi>, <di> and <ji>. When [ɸ] appears before vowels other than /u/, it is transcribed as <f>.
I observed this usage of kara to mark “information source” in the speech of a Miyakoan male FS born in 1954 when collecting data for my thesis (Anderson 2009), so it is by no means restricted to young people’s language. The actual recorded phrase was terebi kara (“on the television”).
Kinship terms from Kyushu dialect such as uchi (1SG FEM) and anta (2SG) are commonly used in the Ryukyus.
Across the Ryukyus, two ablative modal forms exist: potential -(r)areru (an intrinsic “-able/-ible” property) and dynamic -kireru (an individual’s capability, often used in the negative). In contrast to SJ, the potential form is identical to the passive, e.g., kakareru (SJ: kakeru “can write”). The -kireru form is, however, thought to originate from Kyushu dialect and is therefore not connected to substratal influence.
Takaesu (2004: 322) documents other uses for the -yotta form such as hypothetical “would have done” and prospective “about to do”.
This non-standard usage of Standard Japanese (SJ) items does not include the everyday use of literary expressions or those which would be considered as stylistically inappropriate in SJ, such as ichinichi, ninichi … (SJ: tsuitachi, futsuka … “first, second” … day of the month), ichimee, nimee … (SJ: hitori, futari … “first person, second people” …), de arimasu (equivalent to SJ copula desu, rarely used by elderly full speakers), kangetsukai (SJ: tsukimi “moon-viewing”) and a tendency to use humble forms in place of honorifics.
ʔakisamiyoo has many variants such as ʔagijabe, ʔagije, ʔageˀ, hasshabiyoo, hassayoo, hasse, etc. (Fujiki 2004).
These hybrid words with Ryukyuan inflectional suffixes are perhaps best classed as Japanese-superstrate Okinawan (yamatu uchinaaguchi). A full discussion is beyond the scope of this chapter, and there is a need for further investigation into the contexts in which these items are used, i.e., which speaker subgroups use them and which matrix language they tend to be inserted in.
Some unique Ryukyuan loanwords from American English obtain on account of the long-standing US military presence, e.g., tuunaa (“tuna”), aisuwaaraa (“iced water”) and koohii shaapu (“coffee shop”) (Karimata 2012: 24).
The “count” sense of yomu most likely derives from that of Okinawan yunun/yumun but also obtains in Shikoku and parts of Osaka (Hirayama 1984: 72). Likewise, non-standard deixis is not unique to the Ryukyus and also appears in mainland dialects of Tohoku, rural Kanto and Kyushu (Hirayama 1984: 74).
Examples are the use of kureru (“give”) in everywhere except Miyako when the recipient is a listener or third party (SJ: ageru/yaru “give to outgroup”), and the use of kuru (“come”) across all island groups for movement seen from the deictic perspective of the addressee (SJ: iku “go”).