Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 23 February 2017

Japanese Linguistics

Summary and Keywords

The rigor and intensity of investigation on Japanese in modern linguistics has been particularly noteworthy over the past 50 years. Not only has the elucidation of the similarities to and differences from other languages properly placed Japanese on the typological map, but Japanese has served as a critical testing area for a wide variety of theoretical approaches.

Within the sub-fields of Japanese phonetics and phonology, there has been much focus on the role of mora. The mora constitutes an important timing unit that has broad implications for analysis of the phonetic and phonological system of Japanese. Relatedly, Japanese possesses a pitch-accent system, which places Japanese in a typologically distinct group arguably different from stress languages, like English, and tone languages, like Chinese. A further area of intense investigation is that of loanword phonology, illuminating the way in which segmental and suprasegmental adaptations are processed and at the same time revealing the fundamental nature of the sound system intrinsic to Japanese.

In morphology, a major focus has been on compounds, which are ubiquitously found in Japanese. Their detailed description has spurred in-depth discussion regarding morphophonological (e.g., Rendaku—sequential voicing) and morphosyntactic (e.g., argument structure) phenomena that have crucial consequences for morphological theory. Rendaku is governed by layers of constraints that range from segmental and prosodic phonology to structural properties of compounds, and serves as a representative example in demonstrating the intricate interaction of the different grammatical aspects of the language. In syntax, the scrambling phenomenon, allowing for the relatively flexible permutation of constituents, has been argued to instantiate a movement operation and has been instrumental in arguing for a configurational approach to Japanese. Japanese passives and causatives, which are formed through agglutinative morphology, each exhibit different types: direct vs. indirect passives and lexical vs. syntactic causatives. Their syntactic and semantic properties have posed challenges to and motivations for a variety of approaches to these well-studied constructions in the world’s languages.

Taken together, the empirical analyses of Japanese and their theoretical and conceptual implications have made a tremendous contribution to linguistic research.

Keywords: mora, pitch-accent, loanwords, lexical strata, conjugation paradigms, compounds, Rendaku, Case particles, zero pronouns, scrambling, passives, causatives, relative clauses

1. Introduction

Japanese has become one of the most extensively studied non-Western languages in the field of modern linguistics over the past 50 years. It is a well-defined goal of linguists to give a detailed and coherent description of the language from the perspective of grammatical analysis. Without doubt, this goal has been achieved in Japanese linguistics, as evidenced in a series of major works, starting with Shigeyuki Kuroda’s Generative Grammatical Studies in the Japanese Language (1965), Susumu Kuno’s The Structure of the Japanese Language (1973), and Samuel Martin’s A Reference Grammar of Japanese (1987), and continuing with the many books and articles that have followed for further exploration. The scholarly investigation of Japanese has made significant typological and theoretical contributions to linguistics as a discipline, adding to the advancement of our knowledge of natural language and the nature of the language faculty.1 Clear articulation of similarities to and differences from other languages has properly placed Japanese on the typological map; in this context, Japanese has been serving as a critical testing ground for a wide variety of theoretical approaches.

Indicative of the rigor and intensity with which the linguistic research on Japanese has been conducted, and the degree of its productivity, are the number and range of introductory books and handbook series on Japanese linguistics published in the current century: following the lead of Shibatani (1990), the works of Iwasaki (2013), Tsujimura (2014a), and Hasegawa (2015) survey core phenomena under diverse linguistic approaches.2 Tsujimura (1999), Nakayama, Mazuka, and Shirai (2006), Miyagawa and Saito (2008), and a 12-volume handbook series to be published by De Gruyter Mouton between 2015 and 2017 (Shibatani & Kageyama, 2015–) offer overviews of specific topics of Japanese linguistics that place the language in the larger context of linguistic research. Additionally, journals and academic conferences that cater to, or are inclusive of, Japanese linguistics have elevated the stature of the field by confirming its importance to the general linguistics community. Included in the list of such venues are Journal of East Asian Linguistics, Journal of Japanese Linguistics, Japanese/Korean Linguistics (both the conference and its proceedings), and Workshop on Altaic Formal Linguistics (both the conference and its proceedings).

The limited space of this essay allows for only a brief sketch of selected topics in each core area of Japanese linguistics. The focus will be placed on phonetics, phonology, morphology, syntax, and semantics of Standard Japanese and their treatments primarily in the generative tradition. These sub-areas, however, overlap regarding specific issues and terminology; for instance, the section of morphology deals with both phonological and syntactic as well as semantic matters. Other important areas of Japanese linguistics including sociolinguistics and historical linguistics will not be covered here.

2. Phonetics and Phonology

Allophonic alternations are phonological processes that are arguably the most essential to the sound system of Japanese. Allophonic changes of consonants include t ➔ ts / __ u, t ➔ č / __ i, s ➔ š / __ i, z ➔ ǰ / __ i, h ➔ Φ‎ / __ u, and h ➔ ç / __ i. In addition, nasal /n/ in coda position is assimilated in the place feature of the following consonant. High vowels, /i, u/ (the latter being unrounded [ɯ]), become devoiced between voiceless consonants, or when they are word-final and preceded by a voiceless consonant. High vowel devoicing can be considered an assimilation phenomenon where the [-voice] feature of the immediately adjacent sounds is spread to the high vowels. However, devoicing of high vowels is further conditioned by additional factors, including interaction with accent and dialectal variation (Fujimoto, 2015).

The notion of mora is fundamental to describing and analyzing phonetic and phonological phenomena in Japanese. Contrastive with syllable, mora is instantiated by (C)V (e.g., ro.o.ka “hallway”), coda nasal (e.g., “examination”), and the first part of a geminate (e.g., “musical instrument”). Each instance of mora has been experimentally confirmed to be a timing unit: the three three-mora Japanese words above—rooka, kensa, and gakki—have very similar temporal length, and native speakers perceive each word to consist of three units. Those who are familiar with Japanese poetry may realize that the length requirements of haiku and tanka poetry—5-7-5 and 5-7-5-7-7, respectively—reflect that mora, rather than syllable, is the concept underlying the poetic meter. Speech errors that adult native speakers make and children’s phonological acquisition patterns have been shown to provide further evidence for the reality of mora. While phonetic and phonological phenomena in Japanese are governed primarily by mora, the role of syllable can be detected especially in dialects, as observed with phenomena like vowel coalescence (Kubozono, 2006).

Besides entering in the membership of mora, Japanese geminates—typically voiceless obstruents /pp, tt, kk, ss/ for native words—exhibit phonetic characteristics that are typologically noteworthy.3 First, geminates generally occur inter-vocalically in Japanese, while there are languages that can have geminates word-initially and/or word-finally.4 Second, the consonantal duration of geminates in contrast with their singleton counterpart is not uniform across languages, and Japanese geminates are placed at the longer end of the typological spectrum. Third, vowels immediately before geminates are longer than those preceding single consonants in Japanese, while the reverse situation is obtained in Finnish and Icelandic, among other languages. Vowels following geminates are shorter than those after single consonants in Japanese (Han, 1994), but observations internal to other languages and their typological comparisons regarding vowels after geminates are in need of further investigation.

Languages of the world can be classified according to the way in which prosodic prominence is registered. Contrastive with stress languages, like English and Russian, and with tone languages, such as Chinese and Igbo, Japanese instantiates a pitch-accent language where a word has a designated mora for accent, and the pitch contour (or intonation) of the word is determined based on the location of the accented mora. Examples include kákudo “angle”, omόtya “toy”, and takará “treasure”, where the accented vowel (i.e., mora) indicates pitch fall and resulting surface pitch contours are HLL, LHL, and LHH, respectively.5 Additionally, there are unaccented words like sakura “cherry blossom,” where no accent is placed on any mora for signaling pitch fall; sakura is pronounced with the LHH pitch contour as a result. The difference between final accented takará and accentless sakura can be seen when an accentless suffix is added: takará-ga (LHHL) vs. sakura-ga (LHHH). Accentuation patterns in Japanese vary depending on the parts of speech, and they show interesting interactions with word formation processes including suffixation and compounding. For instance, some particles and suffixes change the accentuation patterns of the words to which they are attached, while others do not trigger any change. There is a copious amount of literature on dialectal variation in accentuation patterns. Reflecting mora as a unit central to the sound pattern of the language, high and low pitch is assigned to each mora. However, there are indeed dialects whose accentuation patterns are better described in terms of syllable (Kubozono, 2006). This and other phenomena, like the one mentioned earlier, that are often captured as dialectal variation suggest that syllable, in addition to mora, is an operational linguistic unit in Japanese.

Loanword phonology deals with a large number of issues related to phonological adaptation when two languages are in contact. A fundamental question is how the sound system of a source language is adapted to that of a borrowing language; given that no two languages have identical sound patterns, diverging characteristics, whether they be segmental or prosodic, need to be negotiated in order for lexical borrowing to be complete. Loanword phonology is an important area of exploration in Japanese since it exhibits a somewhat peculiar set of properties that warrants a status as an independent lexical class. As Kubozono (2015) notes, however, it also reveals the structure of Japanese, in its core, and phonological principles underlying natural language that would otherwise not be made apparent. Due to the breadth of issues loanword phonology invokes, more space will be allocated to the topic in this section. Source languages for lexical borrowing in the Japanese context show wide distribution, ranging from Chinese, Dutch, Portuguese, German, Russian, Italian, to English, but words with Chinese origin have customarily been treated as forming an independent word class in its own right under the term “Sino-Japanese stratum” (see section 3).6

Phonetic substitutions are inevitable for phonological adaptation since the source languages typically have phonemes that are missing in the sound inventory of Japanese. Taking English as an example of a source language, the substitution of English interdental fricatives [θ‎, ð] with their alveolar counterparts [s, z] is commonly observed: three and smooth, which are common English loans in Japanese, are pronounced as [suɾi:] and [sumu:zu], respectively. Since Japanese lacks interdental sounds entirely, the substitution can be viewed as an attempt to find sounds of the closest place of articulation, while maintaining the manner and voicing features intact. Other examples of phonetic substitutions are found with English [f, v] being replaced by [Φ‎, b], respectively, as in [Φ‎aibu] for five. Arguably the most frequently discussed phonetic substitution concerns English /r/. Native speakers of Japanese perceive /r/ differently when it occurs syllable-initially or as part of syllable onset (e.g., room, Colorado, and pride) and when it is in coda position (e.g., calendar and pork); and as a result, /r/ in these two situations have non-uniform phonetic realization in loanwords. The syllable-initial /r/ in English is substituted by the Japanese flap, [ɾ], while in coda position, the English liquid is realized as a long vowel of the syllable: [ɾu:mu], [koɾoɾado], [puɾaido], [kaɾenda:], and [po:ku] are the adapted pronunciations of room, Colorado, pride, calendar, and pork, respectively. It is of note that English /l/ is also adapted as [ɾ] since Japanese only has one liquid consonant in its inventory.

One of the well-researched issues in loanword phonology concerns geminates when a source language like English does not contrast geminates and single consonants. It has been observed that lexical borrowing into Japanese can exhibit types of geminates that are not generally found in native Japanese words, as with geminates of voiced stops. Thus, what environment in a source language leads to the realization of geminates in Japanese is not a trivial question (Irwin 2011; Katayama, 1998). There are several crucial elements interacting with each other, and exceptions and idiosyncrasies are also detected. However, as far as English-based loanwords are concerned, word-final stops and affricates that are preceded by lax vowels tend to create a favorable environment for geminates in Japanese. This accounts for loans with geminates like [moppu] “mop,” [kokku] “cook,” and [mačči] “match,” as opposed to loans realized with single consonants like [taipu] “type,” [sutoɾaiku] “strike (in baseball),” and [sa:či] “search.” Loanwords such as [ɾabu] “love,” [basu] “bus,” and [ǰazu] “jazz” follow the same restriction: the word final consonants are fricatives in the source language, and thus they fail to geminate.7 Another geminate-inducing environment is created by certain voiceless obstruents such as [p, k, f, s, č] followed by word-final syllabic [l] in the source language: [appuɾu] “apple,” [waΦΦ‎uɾu] “waffle,” and [kyassuɾu] “castle” exemplify the phonological condition.

Many linguists have observed that loanwords are less restrictive in the choice of geminate consonants. The clearest case is found with geminated voiced consonants: [baggu] “bag,” [beddo] “bed,” [guzzu] “goods,” and [ǰaǰǰi] “judge” are some of the examples. (Hirayama, 2008; Kawahara, 2011b; Lovins, 1975). Some of these geminates in loanwords find their variants with their voiceless counterparts, while others do not. For instance, alternation is available for [baggu]~[bakku] “bag” and [beddo]~[betto] “bed,” but not for [webbu]~*[weppu] “web” and [ɾeddo]~*[ɾetto] “red.” The contrast in alternation between voiced and voiceless geminates is attributed to the presence of preceding obstruents: in the first two examples, the word-initial [b] allows for the alternation, while in the last two, the glide [w] and the liquid [ɾ] make the voiceless geminate option unavailable (Kawahara, 2011a; Nishimura, 2003). With the recent surge of experimental work in linguistic research in general, investigations on voicing of geminates in Japanese, in connection with and independent of loanword phonology, have made clear native speakers’ perception and production patterns concerning geminates. As Kawahara (2005) reports, voiceless geminates are more accurately recognized as such than voiced geminates, and voiced geminates that speakers produce have weakened acoustic properties to the extent that they are similar to their voiceless geminate counterparts. The strong disposition of native speakers for voiceless geminates suggests that such a principle underlies the availability, although limited, of the voiced~voiceless geminate alternation in loanwords. Crucially, the alternation is one-directional: voiced geminates find their voiceless alternants, as in [baggu]~[bakku] “bag” and [beddo]~[betto] “bed,” while we do not find instances where loanwords with voiceless geminates such as [kokku] “cook” and [mačči] “match” have free alternation with their voiced geminate counterparts *[koggu] and *[maǰǰi].8

One of the implications of mora (with the three types of instantiations discussed above) as the basic phonological unit, as opposed to syllable, is that consonant clusters and word-final consonants are not expected in Japanese except for clusters resulting from coda nasal and geminates. If a source language allows for consonant clusters or word-final consonants, they are made to fit the mora structure by vowel epenthesis. The earlier example [sutoɾaiku] “strike (in baseball)” has the consonant cluster in the onset and ends in a consonant; epenthetic vowels split the initial cluster [str], and a vowel is added after the coda consonant [k], both in the attempt to create (C)V sequences consistent with the mora structure of the language. Related to vowel epenthesis as a sound adaptation strategy in loanwords is clipping, which is a morphological process common for other word classes of the Japanese lexicon but is particularly prevalent in lexical borrowing. Frequent vowel epenthesis results in extended word length, and clipping is applied for shortening purposes. Examples include [ɾisutora] for “restructuring,” [depa:to] for “department store,” [pasokon] for “personal computer,” [poteči] for “potato chips,” [sumaho] for “smart phone,” and [buɾapi] for “Brad Pitt.”

Finally, prosodic adaptation may not always be observed cross-linguistically, but there seem to be various patterns when it occurs, and the degree of reference to the prosodic system of the source language may vary (Davis, Tsujimura, & Tu, 2012). McCawley’s (1968) original observation of the accent placement of loanword nouns from English has been the basis of further analyses following it. English is a stress language and is syllable-based. Japanese contrasts with English in these two respects, but accent placement of some loanwords is triggered by English stress, which is based on syllable structure, while that of others is determined irrespective of it. The most common accentuation pattern of loanwords that McCawley generalizes is that accent is placed on the syllable containing the antepenultimate mora, regardless of the stress in the source language: [tόmato] “tomáto,” [bitámin] “vítamin,” [yo:ɾόppa] “éurope,” [wašínton] “Wáshington,” and [eɾebé:ta:] “élevator.” Kubozono (2008) further demonstrates that the antepenultimate rule is not only dominant in loanwords, but also most productive in accented three-mora nouns of the native word class. This parallelism confirms that the strong constraint put on loanword phonology has a significant interaction with more general phonological principles underlying Japanese.9 Other loanwords whose accent placement has nothing to do with the stress of the source language are unaccented loanwords such as [ameɾika] “America,” [aiɾon] “iron,” and [kaunta:] “counter.” Loanwords whose accent maintains the same location as in the source language are found with [buɾú:] “blue,” [ɾésutoɾan] “restaurant,” and [kosumopόɾitan] “cosmopolitan,” among others.

Lexical borrowing in Japanese from English presents a case of prosodic adaptation between a pitch-accent language and a stress language, but it would be interesting to investigate a wide variety of cases of contact between a tone language and a stress language, or between a tone language and a pitch-accent language. In investigating contact between two languages with different prosodic systems, relevant issues have been pursued as to whether or how two distinct prosody types interact in lexical borrowing, and whether and how segmental and suprasegmental properties are negotiated. Research on the prosodic adaptation patterns of Japanese loanwords addresses many of these questions, placing the language in a broader typology while confirming and revealing even further the fundamental nature of the sound system intrinsic of Japanese.

3. Morphology

The lexicon in Japanese is organized into a four-way stratification, as originally proposed by McCawley (1968): native (also called Yamato), Sino-Japanese, mimetic,10 and foreign. The lexical stratification has proven to be important in its own right, but the way in which the strata interact with one another can inform us of essential properties of an individual stratum that are otherwise missed. The names of the strata may suggest that the division is made on an etymological basis, but McCawley’s intention, followed by others, is that the four-way division reflects a set of phonological characteristics particular to each stratum. In theoretical treatments, the relevant phonological characteristics—including the occurrence of singleton [p] and postnasal voicing—have been stated in terms of constraints (e.g., Optimality Theory), according to which the native and foreign strata are on opposite poles, the former being the most restrictive while the latter the least (Ito & Mester, 1995a, 1995b).

While the four-way stratification of the Japanese lexicon is driven by the phonological nature of each stratum, some have taken the position that native and mimetic strata should be united because mimetic words were not borrowed from other languages and hence are etymologically indistinguishable from native words. (Irwin, 2011; Labrune, 2012). Hamano (2000), in her detailed study of Old Japanese, notes that there are significant similarities between the two strata. For instance, of the five vowels of Japanese, /e/ is the least frequent in both native and mimetic vocabulary; and the occurrence of intervocalic /h/ is rare in both strata. On the other hand, Nasu (2015) points out that they do exhibit contrastive behavior at the morpheme-initial position. Native words resist [p] while mimetics allow it; and while voiced obstruents are uncommon in the initial position of bimoraic morphemes within native words (only 15.1% according to Nasu, 2015), they are much more frequent in the mimetic stratum (42.5%). The phonological similarities and contrastive behavior of native and mimetic words to date do not seem to be decisive in determining whether the Japanese lexicon reflects a three-way or four-way stratification. In this connection, it should be noted that the phonological characteristics that have been discussed in the literature of lexical stratification seem to focus on segmental properties, but it may be worth considering systematic patterning in aspects other than segmental phonology. For instance, building upon previous observations such as Hamano (1998) and Tamori and Schourup (1999) but reorganizing and extending them, Akita (2009) demonstrates that mimetic vocabulary in Japanese fits into 15 prosody-based templatic patterns (e.g., CVV-CVV, CV’N-CVN, CVCVQ’, CVCVri, CVCV-CVCV, where the apostrophe indicates the accent of the preceding segment, and “N” and “Q” are shorthand of coda nasal and a geminate), and claims that appearing in one of the 15 templates prototypically identifies a given word as mimetic. From the semantic perspective, Kita (1997) and Tsujimura (2014b) argue that mimetic vocabulary exhibits unique characteristics that are fundamentally distinct from non-mimetic words. While these templatic and semantic properties have not been considered in relation to the lexical stratification issue, they seem to make a reasonable contribution to the question of whether an independent lexical status should be given to mimetics, and ultimately to the refinement of the organization of the Japanese lexicon.

As an agglutinative language, Japanese displays inflectional morphology that is best instantiated by the conjugation paradigms of verbs, adjectives, and the copula. Unlike many languages with verbal inflection, predicative forms are not inflected for person, number, or gender. Instead, affixes are added to a root to form a paradigm. These inflectional affixes include, but are not limited to, markers for tense/aspect, negation, voice, mood, and modality. For example, a consonant-ending verb kak- “write” has the (partial) conjugation paradigm of kai-te (gerund), kak-u (non-past), kak-ana-i (non-past; negative), kai-ta (past), kak-ana-kat-ta (past; negative), and kak-eba (conditional). The corresponding paradigm for the adjective oisi- “delicious” is oisi-ku (gerund), oisi-i (non-past), oisi-ku-na-i (nonpast; negative), oisi-kat-ta (past), oisi-ku-na-kat-ta (past; negative), and oisi-ke-reba (conditional). The copula –da takes the paradigm of –de (gerund), -da (non-past), -ja-na-i (non-past; negative), -dat-ta (past), and -ja-na-kat-ta (past; negative). When nouns are used predicatively, they occur with the copula –da and its various paradigmatic forms (e.g., isu-da “it is a chair”; isu-ja-nakat-ta “it was not a chair”). There are two classes of adjectives in Japanese. One type is like oisi- “delicious,” whose inflectional paradigm has been illustrated above. The other is referred to as “adjectival nouns” due to their dual status as “adjectival” and “nominal.” They are “adjectival” in that they serve as modifiers for nouns that follow them and take adverbs like totemo and sugoku for intensifiers. But, they are “nominal” in that their conjugational paradigm patterns with that of the copula, on par with predicative nouns. Examples of adjectival nouns include kirei “pretty, clean,” benri “convenient,” and sizuka “quiet.”

Word formation is an unavoidable topic in morphology. Word formation processes employed in Japanese include prefixation, suffixation, reduplication, clipping, compounding, and borrowing. In particular, compounding occurs richly and exhibits wide variation in terms of the category and stratum of composite members, in relations between compound members, and in the degree of productivity. Within the native vocabulary, nouns, verbs, and adjectives are the primary parts of speech categories that serve as members of compounds, and their combinations exhibit an assortment comprising, although not limited to, N-N (e.g., natsu-matsuri “summer-festival,” hon-dana “book-shelf”), V-V (e.g., tachi-gui “stand-eat,” oboe-gaki “remember-write”), A-A (e.g., asa-guroi “shallow-black,” usu-akarui “thin-light”), N-V (e.g., hon-yomi “book-read[ing],” yama-nobori “mountain-clibm[ing]”), V-N (e.g., kiri-kuchi “cut-mouth,” yomi-mono “read-thing”), N-A (e.g., en-yasu “Yen-cheap,” hara-ita “stomach-painful”), and A-N (e.g., too-goe “far-voice,” yasu-zake “cheap-Sake”). Continuing to assume the four-way stratification of native (N), Sino-Japanese (SJ), foreign (F), and mimetic (M), there are ample examples of hybrid compounds, crossing over the strata: native ~ Sino-Japanese (e.g., yasu-ryokan “cheap [N]–inn [SJ]”), foreign ~ native (e.g., garasu-mado “glass [F]–window [N]”), mimetics ~ native (e.g., piri-kara “sting [M]–spicy [N]”), Sino-Japanese ~ foreign (e.g., denki-sutoobu “electric [SJ]–stove [heater] [F]”), Sino-Japanese ~ mimetic (e.g., kin-pika “gold [SJ]–shiny [M]”), and mimetic ~ foreign (e.g., pakku-man “chomp [M]–man [F]”).

Much of the literature on compounds to date has focused on compound verbs of the N-V and V-V patterns, and it is in that context that the relationship between composite members and the degree of productivity has been examined (Ito & Sugioka 2002; Kageyama, 1993). N-V compounds, many of which are nominalized, can be divided into to two types according to their morphosyntactic relations: N serves either as the head V’s argument or as its adjunct. The argument-V relation is observed with nominalized N-V compounds like kutsu-migaki “shoes-polish,” ame-huri “rain-fall,” and hige-sori “beard-shave,” while the adjunct-V relation is maintained in compounds like hi-yake “sun-burn,” te-gaki “hand-write,” and abura-itame “oil-stir fry.” The semantic nature of the adjuncts to the verb varies from source/reason (hi-yake) and instruments (te-gaki) to manner and resulting states. The division based on the argument-adjunct distinction of the N member has been claimed to have implications to productivity. Ito and Sugioka (2002), for one, point out that N-V compounds of the argument-predicate type are more productive, and newly coined compounds of this type would be easily understood, while N-V compounds with the adjunct-predicate relation are less productive, and new coinage is semantically less transparent.

V-V compounds have been even more closely scrutinized due to the enormous number of examples in the language as well as to the important implications that they bear. Kageyama (1993) carefully examined and analyzed V-V compounds into lexical and syntactic compounds based on their contrastive behavior vis-à-vis a battery of diagnostic tests. As the terms of “lexical” vs. “syntactic” suggest, lexical V-V compounds show a tighter connection between the members reflecting the status as single words, while syntactic V-V compounds allow for morphological interruption between the composite members. For instance, the passive morpheme –(r)are- can intervene between the two Vs in syntactic compounds like tabe-hajimerutabe-rare-hajimeru “eat-pass.-begin” and korosi-kakerukoros-are-kakeru “kill-pass.-about to,” but it is not possible with lexical V-V compounds like oshi-hiraku➔ *os-are-hiraku “push-pass.-open” and kiki-kaesu➔*kik-are-kaesu “ask-pass.-return.” Other tests include the replacement of so-suru ‘do so,” subject honorification with o-ni naru, and reduplicative forms. All these tests confirm contrastive properties between the two types of V-V compounds. The second member of syntactic V-V compounds tends to be aspectual in the semantic nature, including V–hajimeru “begin to V,” V-tsuzukeru “continue to V,” and V-owaru “finish V-ing,” but other verbs referring to excessiveness (V-sugiru), reciprocal actions (V-au), and possibility (V-uru) also serve as the second member of syntactic V-V compounds. To the extent that these second members maintain the aspectual or regularized (and expected) meanings, syntactic compounds with them are productively formed. Lexical V-V compounds, on the other hand, are far more difficult to generalize over because the semantic relations between the two show a wide variety. Furthermore, lexical VV-compounds often undergo semantic bleaching, and their meanings cannot be compositionally deduced based on the meaning of each composite in isolation. The earlier examples, oshi-hiraku “push-open” and kiki-kaesu “ask-return,” have relatively transparent meanings, putting together what each verb means: oshi-hiraku means open something by pushing, and kiki-kaesu means ask questions or listen to something again. The relatively transparent semantic relations between the two members further include means (e.g., tataki-tsubusu “hit-smash”), manner (e.g., kake-yoru “run-approach”), cause (e.g., kuzure-ochiru “collapse-fall”), and juxtaposition of similar descriptions (e.g., naki-sakebu “cry-scream”).11 However, there are numerous V-V compounds whose meanings cannot be compositionally predicted: ochi-tsuku (“fall-arrive”) means “settle down, calm down,” nomi-aruku (“drink-walk”) means “go bar-hopping,” and naki-otosu (“cry-drop”) means “use tears to get one’s way.” These meanings can be reached by metaphorical extension, but regular patterns are hardly captured.

In addition to the connected issues of transparency of meaning and productivity, the lexical vs. syntactic distinction has phonological implications, specifically concerning contraction and sequential voicing (Rendaku), the latter of which will be taken up later. Ito and Sugioka (2002) observe that some V-V compounds alternate with contracted forms that contain coda nasal or a geminate consonant. For example, humi-kiru (“step-cut”) “take off, take a bold course” ➔hungiru and humi-tsukeru (“step-attach”) “stamp down” ➔hunzukeru illustrate the first alternation; hiki-komu (“pull-be congested”) “bring in” ➔kikkomu and buchi-taosu “hit-knock down” ➔buttaosu are the example of a geminate consonant. They observe that V-V compounds that alternate with contracted counterparts are restricted to lexical compounds, as the examples above attest to. Furthermore, some V-V compounds that can be interpreted as either lexical or syntactic compounds allow for their contracted alternates only when they are considered to be lexical compounds: oi-kakeru is ambiguous between a lexical compound (“chase-run”) and a syntactic compound (“chase-about to”), but the contracted alternate okkakeru can only mean the former. This phenomenon, if more widely confirmed, offers phonological validation for the two-way division of V-V compounds.

Since a V-V compound consists of two non-identical verbs, an important question has been pursued in the literature as to how the argument structure of the compound is assigned and whether the process to the assignment is constrained in any way. (Fukushima, 2005; Kageyama, 1993; Matsumoto, 1996; Nishiyama, 1998). There is very little issue with compounds consisting of verbs of an identical argument structure type—transitive+transitive, unergative+unergative, and unaccusative+unaccusative—as the argument structure type of the whole is identical with that of composite members. However, compounding of verbs of non-homogeneous argument structures is constrained such that verbs with the external argument (i.e., transitive and unergative verbs) can be put together while compounding of either of these verbs with an unaccusative verb results in ungrammatical forms. Counterexamples of this generalization—compounds comprising a transitive verb and an unaccusative verb or those of an unergative verb and an unaccusative verbs—have been noted and discussed, but they are generally given independent reasons attributing to analogical processes, diachronic change, and coinage (Ito & Sugioka, 2002; Kageyama, 2013).

Exploration of the relation between the two sets of argument structures in V-V compounds has led Kageyama (2013) to claim that lexical V-V compounds need to be further grouped into two classes: thematic compound verbs and aspectual compound verbs. Building upon but reorganizing earlier classifications of lexical V-V compounds that take semantic relations between the compound members into consideration (e.g., Kageyama 1993, 1999; Matsumoto 1996, 1998; Fukushima 2005; Yumoto 2005), Kageyama (2013) recognizes the difference in semantic contribution that each composite verb makes to the meaning of the whole V-V compound. In thematic compound verbs such as tsuki-otosu (“push-drop”), koroge-ochiru (“roll-fall”), aruki-tsukareru (“walk-get tired”), and koi-shitau (“long for-yearn”), the meanings of V1 and of V2 are both relevant to the meaning of the compound, but the semantic weight is on V2 while V1 modifies it. This relation is demonstrated by the relevance of the meanings of both V1 and V2 when they are phrasally coordinated (with the gerundive form of V1): for tuki-otosu, tsuite otosu “push and drop ➔ drop by pushing,” and for aruki-tsukareru, aruite tsukareru “walk and get tired ➔ get tired after walking.” The main event denoted by the compound verb comes from V2, but the meaning of V1 is sustained to provide additional information. Thematic compound verbs, thus, reflect the relevance of the meaning of both composite members. This relation contrasts with aspectual compounds verbs like nui-ageru (“sew-raise”) “finish sewing,” sawagi-tateru (“make a racket-stand”) “make a great fuss,” and shimeshi-au (“suggest-meet”) “arrange in advance,” where the meaning of the compound is determined by V1 while that of V2 is bleached and simply provides aspectual information, such as completion, emphasis, and reciprocity, among others. Although the relation between V1 and V2 is different between the two types within lexical V-V compounds from the perspective of semantic contribution of composite verbs, it should be underscored that both thematic compound verbs and aspectual compound verbs still belong to lexical V-V compounds, contrasting with syntactic V-V compounds. Even though aspectual compound verbs are similar to syntactic V-V compounds in that the primary role of the V2 member is aspectual in nature, none of the diagnostic test that separates lexical compounds from syntactic compounds (e.g., insertion of the passive morpheme, substitution of soo-suru “do so,” subject honorification with o-ni naru, and reduplicative forms) shows that aspectual compound verbs are syntactic compounds.

The topic of sequential voicing, “Rendaku,” is perennially discussed in connection with compounds as it raises a number of individual and connected issues that have larger implications to the morphophonology and morphosyntax of Japanese.12 The phenomenon is generalized such that the initial voiceless consonant of the second member of a compound undergoes voicing. Examples include [maki]+[suši] ➔ [maki-zuši] “rolled sushi,” [me]+[kusuɾi] ➔ [me-gusuɾi] “eye drop,” [ki]+[to] ➔ [ki-do] “wooden door,” and [hana]+[či] ➔ [hana-ǰi] “nose bleed.” The Rendaku phenomenon is constrained in various realms of the grammar. Each of these constraints has met varying degrees of counterexamples, casting some doubt on the regularity and predictability of the phenomenon. Below, however, I will illustrate the range of data that have motivated the constraints.

Lyman’s Law prevents Rendaku from applying if the second member contains a voiced obstruent. Effects of this segmental constraint are shown in the contrastive pairs: [ya] “arrow” + [šiɾuši] “sign” ➔ [ya-ǰiɾuši] “arrow sign” vs. [šita] “below” + [šigoto] “work” ➔ [šita-šigoto] “preparatory work”; [ao] “blue” + [kaeɾu] “frog” ➔ [ao-gaeɾu] “green frog” vs. [aka] “red” + [kabu] “turnip” ➔ [aka-kabu] “beet”; and [aki] “autumn” + [soɾa] “sky” ➔ [aki-zoɾa] “autumn sky” vs. [tsukimi] “moon watching” + [soba] “soba” ➔ [tsukimi-soba] “soba in broth with a raw egg on top.” Lyman’s Law has been considered by some an instance of dissimilation, and its formal analyses have been accommodated by theories that are motivated for general dissimilation phenomena including the Obligatory Contour Principle (Ito & Mester, 1986).

Rendaku is constrained by the lexical strata in that the degree of its application varies depending on the stratum, ranging from the most prevalent application in the native class (examples above)13 to its strict prohibition in mimetics (e.g., pika-pika/*pika-bika; suta-suta/*suta-zuta; kari-kari/*kari-gari). Details of Rendaku application to Sino-Japanese words are quite complicated to give here due to the intricate historical background of Chinese borrowing and morphological structure of the original Chinese forms. Examples of sequential voicing in compounds that are formed with Sino-Japanese words at least for the second member are attested (e.g., [ko] “child” + [kaiša] “company” ➔ [ko-gaiša] “subsidiary company,” [ao] “blue” + [šašin] “photo” ➔ [ao-ǰašin] “blueprint,” and [tetsu] “iron” + [ko:ši] “grid” ➔ [tetsu-go:ši] “iron grid.” However, it seems that there are far more instances in which expected voicing does not actually occur, suggesting that Rendaku is not as prevalent in Sino-Japanese as in the native stratum. Equally restricted is the foreign stratum: [pa:to-taimu] “part-time,” [insutanto-ko:çi:] “instant coffee,” and [konsome-su:pu] “consommé soup” are not realized as *[pa:to-daimu], *[insutanto-go:çi:], and *[konsome-zu:pu] even though other conditions are met for Rendaku application. Some words that were borrowed quite some time ago and have been nativized, such as karuta “card” (borrowed from Portuguese) and kiseru “smoking pipe” (borrowed from Khmer), undergo Rendaku: [uta-garuta] “playing cards with waka on”; [mizu-giseɾu] “water pipe” (Irwin, 2011).

The semantic relations between the two members of a compound, represented by internal morphological structure, determine when Rendaku does and does not apply. When two members are of equal semantic weight, forming a conjoined structure termed “dvandva compounds,” Rendaku is invariably resisted. This is shown by [oya-ko] “parent and child,” [iki-kaeɾi] “coming and going,” and [kusa-ki] “grass and tree.” When a compound consists of three or more members, the hierarchical structure governs which members are targets for voicing. In brief, only the members on the right branch can undergo Rendaku. For example, three words, [o:] “big,” [tako] “octopus,” and [suki] “like” may be compounded into [[[o:] + [tako]] + [suki]] “love(er) of big octopus” or [[o:] + [[tako] + [suki]]] “great love(er) of octopus,” where the former has two right branches ([tako] and [suki]) while the latter has one ([suki]). This right branch condition yields [o:-dako-zuki] under the first meaning with two Rendaku applications, and [o:-tako-zuki] under the second with one Rendaku application. The right branch condition suggests that the phonological phenomenon of sequential voicing takes into account the internal structure of compounds, which reflects the semantic relationship among the members.

Our earlier discussion of N-V compounds noted that the N member can be either an argument or an adjunct of the V member. This dichotomy further interacts with Rendaku in that voicing is generally limited to the adjunct N-V compound type. The following contrasting pairs illustrate the difference: [šo:setsu-kaki] “novel-writing” (argument) vs. [te-gaki] “writing by hand” (adjunct), [zubon-tsuɾi] “trouser-hanging” (argument) vs. [sakasa-dzuɾi] “upside down-hanging” (adjunct), and [muši-toɾi] “bug-catching” (argument) vs. [ike-doɾi] “alive-catching” (adjunct) (Ito & Sugioka, 2002; Sugioka, 1986).

Contrastive of pervasive occurrences of the Rendaku phenomenon, its irregularity has also been noted, as supported by a number of apparently random failures of Rendaku applications (Vance, 1987, 2015). Recent scrutiny of larger data sets, however, reveals that cases in which Rendaku is expected to occur but actually does not indeed pattern with regularity, and that a systematic account for the regularity is found in the prosodic size of a compound and of its member (Rosen, 2001, 2003). As is already discussed by Martin (1987) and Vance (1987), among others, a small group of nouns (e.g., kasu “dregs,” katachi “shape,” himo “string,” and shita “below”) always resist Rendaku when they are the second member of a compound. Separate from these Rendaku “immune” words, and focusing on native nouns, there is another group of nouns that resist Rendaku. Rosen (2003) calls these nouns Rendaku “resisters,” pointing out that they are different from Rendaku immune nouns in that (a) Rendaku resisters undergo voicing in a small number of short N-N compounds, where each member is one or two moras, and (b) Rendaku always applies to them when they are in long compounds, namely compounds of five moras or longer. That is, when Rendaku resisters block voicing and when they do not is predictable given the prosodic size of a compound. To illustrate, kusa “grass,” as the N2 member of a compound, is a Rendaku resister: it mostly shows resistance to voicing in short compounds like [aki-kusa] “autumn grass,” [buta-kusa] “pig grass,” and [mizu-kusa] “water grass,” but a small number of instances that undergo voicing is attested, as in [no-gusa] “wild grass,” [mo-gusa] “water plants,” and [či-gusa] “great variety of flowering plants.” A parallel situation is observed with kuse “habit,” ki “tree,” and te “hand,” among others, when they serve as the second member of short native N-N compounds. In contrast, when these nouns form a long compound with a long N1 member (i.e., three moras or longer), they invariably undergo voicing: [çitsuji+kusa] ➔ [çitsuji-gusa] “sheep grass” and [tokiwa+ki] ➔ [tokiwa-gi] “evergreen.” The statistical distribution of Rendaku to these short and long compounds that Rosen summarizes is indicative of its relevance of prosodic size. For native N-N compounds he surveyed, 90% of the total 580 short compounds and 100% of the total 196 long compounds undergo voicing, pointing to the prevalence of the phenomenon. Rendaku immune nouns form 49 short compounds and 14 long compounds, but no voicing is attested in any of them. For Rendaku resisters, only 30% of the total 119 short compounds sees voicing, whereas 100% of 13 long compounds all undergo voicing. Rendaku has been discussed extensively regarding both its pervasiveness and its irregularity, but Rosen’s analysis—known as Rosen’s Rule—of the irregular Rendaku application as a systematic reflection of prosodic structure of compounds sheds new light on the phonological phenomenon from a different angle. At the same time, it calls for a morpho-phonological mechanism to mark Rendaku immune nouns and Rendaku resisters separately in the lexicon so that their individual paths to the voicing phenomenon are uniquely represented.14

4. Syntax and Semantics

The basic word order of Japanese is viewed as SOV (subject-object-verb), but constituents in a sentence can be rearranged without much semantic change as long as the verb maintains its final position (See the discussion of scrambling in this section). The subject and object of a transitive sentence are marked morphologically by the nominative (-ga) and accusative (-o) Case particles, respectively. The subject of an intransitive sentence, both the unergative and unaccusative types, is also marked with the nominative Case particle, -ga. Additionally, an indirect object in a ditransitive sentence is marked by the dative (-ni) Case particle. Many of ditransitive sentences are formed with verbs of change of possession (e.g., ageru “give” and ataeru “give”), verbs of transfer (e.g., okuru “send” and nageru “throw”), and verbs of transaction (e.g., uru “sell” and harau “pay”) (Kishimoto, 2001, Matsuoka, 2003). The nominative and accusative Case particles are often dropped, especially in casual conversations. This is one of the characteristics that separate the Japanese Case system from the more traditional Case patterns that are observed with languages like Latin, Russian, and Finnish (Tsujimura, 2014a).

The Japanese Case marking patterns suggest that the language belongs to the nominative-accusative, rather than the ergative-absolutive, system. However, there are a few other Case arrays, at least some of which may be triggered by the semantic type of verbs. First, a small number of stative verbs (e.g., dekiru “be able to do,” iru “need”) mark the subject and object with the dative and nominative Case particles, respectively; or both can be marked with the nominative Case (e.g., Kodomo-ga/ni sonna koto-ga dekiru-no [child-Nom/Dat such.a.thing-Nom] “can a child do such a thing?”). This Case array is also available to the potential form of verbs like kak-e-ru “can write” and mi-rare-ru “can see.” Second, in sentences with verbs whose subject can be considered to have the source role (e.g., okuru “send,” renraku-suru “inform”), the postposition –kara “from” can replace the nominative Case on the subject (e.g., Watashi-ga/kara sochira-ni tegami-o okurimasu [I-Nom/from you-Dat letter-Acc send] “I will send you a letter”). Third, a less frequently discussed pattern is the instances of “oblique subject” (Sells, 2008), where the subject that may be interpreted to represent a type of “institution” like a company, school, and the police is marked with the postposition –de “at” (e.g., Keisatsu-de kono jiken-o atukatteiru [the.police-at this case-Acc treat] “The police have been dealing with this case”). The same Case pattern is also observed with a plural subject in sentences like Haha-to watashi-de daigaku-o otozureta [mother-and I-at college-Acc visited] “My mother and I visited the college”; Gakusei-tachi-de bokoo-o otozureta [student-pl.-at alma.mater-Acc visited] “The students visited their alma mater” (Alfonso, 1974; Inoue, 2000; Martin, 1987; Sells, 2008).

Constituents of a sentence in Japanese do not all have to be overtly expressed as long as they are contextually understood. Since unexpressed constituents are interpreted as pronouns in the context, they are sometimes referred to as “zero pronouns.” The phenomenon, however, should be distinguished from unexpressed subjects in languages like Italian and Spanish, where the person and number features of an unexpressed subject are reflected on the verbal inflection. These languages have been treated in the syntax literature as “pro-drop” languages. While some scholars have treated Japanese as a pro-drop language, the nature of the unexpressed constituents is different enough between the two types of languages to warrant the term “zero pronouns” for Japanese as well as Chinese and Korean.

Flexible permutation of constituents, grammatical constructions marked by agglutinative morphosyntax, and head-final phrase structure are some of the characteristics of Japanese that have contributed to typological generalizations and validation of syntactic theories. Among the many themes of research investigation in Japanese syntax, scrambling is one of the most rigorously and extensively discussed syntactic topics in the generative tradition especially during the last quarter of the 20th century and even to date. The depth and breadth of the inquiry is reflected not only by issues that emerge directly from the scrambling phenomenon with significant theoretical and typological implications, but also by the range of other phenomena—syntactic or otherwise—that show linguistically interesting patterns in interacting with scrambling. Arguably the foremost theoretical contribution that the new wave of syntactic analyses of scrambling has made is to the configurationality of the basic structure of Japanese. The basic word order of SOV and its scrambled counterpart OSV were said to not exhibit significant semantic differences, and their syntactic and semantic statuses were treated on par between the two permutations. This non-configurational analysis, where its hall-mark “flat” structure assumes that the subject and the object are symmetrically structured, was challenged by the configurational view that Japanese has a hierarchical structure in which the subject c-commands the object. Based on evidence from weak crossover (Saito & Hoji, 1983), pronominal reference (Saito, 1985), and numeral quantifiers (Miyagawa, 1989), among others, the configurational approach demonstrated that scrambling is a movement operation applied to a hierarchically ordered structure. Scrambling, thus, has placed the syntactic structure of Japanese typologically under the same rubric as English, for instance, rather than singling it out to fall under languages of somewhat uncommon structural properties. Continuing pursuit of scrambling within the generative tradition, ranging from the government and binding theory to the Minimalist Program as well as to the theories that have come between them, has led to extensive discussions on such issues as short- vs. long-distance scrambling and optionality of the movement (Kitahara, 1994; Miyagawa, 1997, 2005, 2006; Saito, 2005; Takano, 1999; Yatsushiro, 2003).

An earlier discussion of scrambling as an instance of a syntactic movement operation includes the claim that in the ditransitive structure, the dative-accusative order is the basic word order, while the accusative-dative configuration results from scrambling (Hoji, 1985). The two word orders are exemplified respectively by Taroo-ga Hanako-ni chokoreeto-o agega [Taro-Nom Hanako-Dat chocolate-Acc gave] and Taroo-ga chokoreeto-o Hanako-ni ageta [Taro-Nom chocolate-Acc Hanako-Dat gave] “Taro gave Hanako chocolate.” Hoji’s argument is based on quantifier scope interpretations. In Taroo-ga dareka-ni dono-hon-mo ageta [Taro-Nom someone-Dat gave] “Taro gave someone every book,” “some” can take the scope over “every,” but not vice versa. In Taroo-ga dono-hon-moi dareka-ni ti ageta [Taro-Nom everybook-alsoi someone-Dat ti gave], on the other hand, both scope interpretations are available. The scope ambiguity in the accusative-dative order of the second sentence is attributed to the trace left behind by scrambling of the accusative NP. Support for this position has further been provided (e.g., Kishimoto, 2008; Takano, 1999; Yatsushiro, 2003), but an alternative view that both orders can be base-generated has also been proposed (Miyagawa, 1997; Miyagawa & Tsujioka, 2004). The debate over the basic word order for ditransitive verbs has direct relevance to general structural properties of Japanese, but at the same time, it has made a broader cross-linguistic contribution to the controversial topic of the dative alternation (or dative shift) in English and other languages. A large body of literature, inclusive of diverse theoretical views, has been devoted to the discussion of the double object construction (e.g., John gave Mary flowers) and the PP dative construction (e.g., John gave flowers to Mary) and their syntactic and semantic relations (Bowers, 1981; Bruening, 2001; Chomsky, 1955; Dryer, 1987; Harley, 2002; Larson, 1988; Oehrle, 1976; Richards, 2001). Double object verbs in Japanese do not exhibit the type of structural alternation as in English, restricting apparent variation only to the switch in word order between the dative-marked and accusative-marked NPs. The dearth of structurally obvious differences that substantiate two variable syntactic projections in Japanese has casted doubt on the relevance of the cross-linguistic comparison for the dative alternation between English and Japanese. Drawing on data from ditransitive idioms, Miyagawa & Tsujioka (2004), Kishimoto (2008), and Tsujioka (2011) exchange arguments over the question of whether (and when) base-generation is available to both or either of the dative-accusative order and the accusative-dative order. The significance of this debate about Japanese is that, beyond facilitating an explication of the syntactic nature of the ditransitive structure, it shows that scrambling plays a methodological role as a well-established apparatus for elucidating the way in which arguments of double object verbs are projected in syntax, and that it participates in, and contributes to, a cross-linguistic discussion of syntactic and semantic characterization of double object verbs and their arguments.

Syntax and semantics of passive and causative sentences, which are formed by morphological agglutination, have been given a variety of treatments from diverse perspectives. The most prominent characteristic of Japanese passives that has piqued linguists’ interest of investigation, regardless of theoretical orientation, is that there are two types of passives—direct and indirect passives—and that one of them, indirect or adversative passives, is valence-increasing. Direct passives in Japanese are treated roughly parallel to those in English as a valence-reducing process based on transitive verbs: for instance, an active sentence with the transitive verb, homeru “praise,” Sensei-ga Hanako-o hometa [teacher-Nom Hanako-Acc praised] “The teacher praised Hanako,” has its corresponding passive sentence Hanako-ga (sensei-ni) homer-are-ta [Hanako-Nom (teacher-by) praise-pass.-past] “Hanako was praised (by the teacher),” where the passive morpheme -(r)are- is suffixed to the verbal root. Indirect passives, in contrast, add an NP that semantically corresponds to an individual who becomes affected by the action denoted by the verb. As such, indirect passives are formed around either intransitive or transitive verbs: Taroo-ga kodomo-ni nak-are-ta [Taro-Nom child-by cry-pass.-past] “Taro was (adversely) affected by his child’s crying” is based on the intransitive verb nak- “cry,” and Taroo, which is not an argument of nak- “cry,” is an additional NP that signals the indirect passive construction. Indirect passives with transitive verbs share the same syntactic and semantic characteristics, as demonstrated by Taroo-ga sensei-ni kodomo-o homer-are-ta [Taro-Nom teacher-by child-Acc praise-pass.-past] “Taro was affected (i.e., pleased) that the teacher praised his child.” This transitive-based indirect passive sentence suggests that semantically indirect passives can have positive effects on the individual referred to by the subject NP despite the alternative term “adversative” passives. While the availability of the valence-increasing passive construction in Japanese has provided a new testing ground for theoretical approaches that had been founded on the more widespread direct passive construction across languages, Japanese is far from unique, in that it joins other languages of the world including Yupik, Thai, and Vietnamese in exhibiting similar constructions.

Just as passives are formally recognized by the allomorphs of -(r)are- to be suffixed to the verbal root, so are causatives by another set of allomorphs -(s)ase- to the verbal root. The literature on Japanese causatives has amassed discussions of the productive (or, syntactic, periphrastic) causatives with -(s)ase- and lexical causatives. Many of lexical causatives correspond to the transitive counterparts of morphologically related verb pairings like taos-/taore- “fell/fall,” ake-/ak- “open/open,” and naos-/naor- “repair/be repaired,” where the first of each pair is considered a lexical causative verb (see Jacobsen, 1992 for more examples). The productive causative construction with -(s)ase- is a valence-increasing process, whereby an NP corresponding to a causer is added, and is formed around either an intransitive or transitive verb. The causee object of a causative sentence based on an intransitive verb can be marked either by the accusative Case particle –o to induce a coercive interpretation (o-causative), Taroo-ga Hanako-o nak-ase-ta [Taro-Nom Hanako-Acc cry-caus.-past] “Taro made Hanako cry”; or by the dative Case particle –ni, which brings about the sense that the causation is by way of giving permission (ni-causative), Taroo-ga Hanako-ni nak-ase-ta [Taro-Nom Hanako-Dat cry-caus.-past] “Taro let Hanako cry.” The two types of marking the causee are reduced solely to the dative marking in causative sentences with transitive verbs in order to avoid the Double-o Constraint (Harada, 1973; Kuroda, 1978; Poser, 1981). The constraint disallows two occurrences of the accusative marking in a clause (i.e., one for the causee and the other for the object of the transitive verb): Taroo-ga kodomo-*o/ni yasai-o tabe-sase-ta [Taro-Nom child-*Acc/Dat vegetable-Acc eat-caus.-past] “Taro had his child eat vegetables.”

Semantically, the lexical and productive causative constructions have been distinguished such that the nature of causation is more direct in lexical causatives than productive causatives. Syntactically, they demonstrate for the most part contrastive behavior that is attributed to structural distinctions. There are several syntactic tests to highlight the differences, but chief among them is reflexive binding.15 It is widely accepted that the Japanese reflexive word jibun “self” takes only the subject as its antecedent. When jibun appears in the lexical causative construction with the verb nose- “get someone on,” the antecedent of jibun is unambiguously the subject of the sentence: Taroo-ga Hanako-o jibun-no kuruma-ni nose-ta [Taro-Nom Hanako-Acc self-Gen car-onto get.someone.on-past] “Taro got Hanako in self’s (=his) car.” With the productive causative construction with –(s)ase-, the reflexive word is ambiguously interpreted: in Taroo-ga Hanako-o jibun-no kuruma-ni nor-ase-ta [Taro-Nom Hanako-Acc self-Gen car-onto get.on-caus.-past], the antecedent of jibun is either Taro or Hanako. Two possible interpretations of jibun indicate that there are two subjects, pointing to a bi-clausal structure. Much of the discussion on the two types of Japanese causatives has centered on similarities and differences in structural and semantic properties as well as in the degree of regularity and productivity in deriving these causatives as part of the word formation mechanism. This issue has been approached from diverse theoretical frameworks including LFG (e.g., Matsumoto, 1996), HPSG (e.g., Manning, Sag, & Iida, 1999), the Minimalist Program (e.g., Harley, 1995), and Distributed Morphology (e.g., Harley, 2008). Setting theoretical preferences aside, continuing investigation of the Japanese causative construction and the breadth of approaches available in order to articulate an important range of issues that cross over syntax and morphology and interact with the principles that govern each component suggest that the causative structure built upon agglutinative morphology still leaves much to be pursued, awaiting fruitful results yet to emerge. In that light, evidence from specific language impairment and aphasics such as that discussed in Fukuda and Fukuda (2001) and Sugioka, Ito, and Hagiwara (2001), for example, help illuminate implications of the two kinds of causatives to brain activity.

Relative clauses have been widely examined cross-linguistically, and have brought much import to typological mapping of the world’s languages and to their theoretical implications. Among substantial syntactic factors that have been considered in the typological context is the head-initial vs. head-final nature of the filler-gap relation formed by relative clauses. English, for example, represents the head-initial relative clause type (the cookies [that Mary bought ___]), while Japanese is an example of the head-final relative clause type ([Hanako-ga ___ katta] kukkii). In language processing research, the head-initial relative clause languages like English demonstrate that subject relative clauses, where the gap is the subject of the relative clause (the woman [who ___ bought the cookies]), are asymmetrically easier to process than object relative clauses (the cookies [that the woman bought ___]). The asymmetrical processing patterns have been explained by the linear distance between the gap and the head: the linear distance of the filler-gap relation is shorter in subject relatives than object relatives (Gibson, 1998). The linear distance approach predicts that languages of the head-final relative clause type, like Japanese, show that the processing of object relative clauses ([josei-ga ___ katta] kukkii) [[woman-Nom ___ bought] cookies] would be easier than that of subject relative clauses ([___ kukkii-o katta] josei) [[___ cookies-Acc bought] woman]], since the linear distance of the filler-gap relation is shorter in object relative clauses—the reverse situation of English. Psycholinguistic experiments that measure the reading time of subject and object relative clauses in Japanese have confirmed that the results are actually similar to English, in that the reading time for subject relative clauses was shorter than that for object relative clauses (Miyamoto & Nakamura, 2003, Ueno & Garnsey, 2008). The parallel results in Japanese and English pose a challenge to the linear distance analysis, but instead support the structural distance account that, in both head-initial and head-final relative clause languages, subject relative clauses are easier to process than object relative clauses because the subject is in a hierarchically more prominent position than the object (O’Grady, 1997). The discussion of the subject-object asymmetry in processing of relative clauses further confirms that Japanese employs a configurational structure in which the subject asymmetrically c-commands the object, as argued based on scrambling and other phenomena.

To conclude, this article has highlighted only some of the key areas of research and findings on Japanese linguistics. It is expected that Japanese will continue to be an important testing ground for linguistic theories in the future.

Further Reading

Asahi, Y. A., Usami, M., & Inoue, F. (Eds.) (2017). Handbook of Japanese sociolinguistics. Berlin: De Gruyter Mouton.Find this resource:

Frellesvig, B. (2010). A history of the Japanese language. Cambridge, U.K.: Cambridge University Press.Find this resource:

Hasegawa, Y. (2015). Japanese: A linguistic introduction. Cambridge, U.K.: Cambridge University Press.Find this resource:

Irwin, M. (2011). Loanwords in Japanese. Amsterdam: John Benjamins.Find this resource:

Iwasaki, S. (2013). Japanese (rev. ed.). Amsterdam: John Benjamins.Find this resource:

Jacobsen, W., & Takubo, Y. (Eds.) (2017). Handbook of Japanese semantics and pragmatics. Berlin: De Gruyter Mouton.Find this resource:

Kageyama, T., & Kishimoto, H. (Eds.) (2016). Handbook of Japanese lexicon and word formation. Berlin: De Gruyter Mouton.Find this resource:

Kubozono, H. (Ed.) (2015). Handbook of Japanese phonetics and phonology. Berlin: De Gruyter Mouton.Find this resource:

Miyagawa, S., & Saito, M. (Eds.) (2008). The Oxford handbook of Japanese linguistics. Oxford: Oxford University Press.Find this resource:

Nakayama, M. (Ed.) (2015). Handbook of Japanese psycholinguistics. Berlin: De Gruyter Mouton.Find this resource:

Shibatani, M. (1990). The languages of Japan. Cambridge, U.K.: Cambridge University Press.Find this resource:

Shibatani, M., Miyagawa, S., & Noda, H. (Eds.) (2017). Handbook of Japanese syntax. Berlin: De Gruyter Mouton.Find this resource:

Tsujimura, N. (Ed.) (2005). Japanese linguistics (Vols. 1–3). London: Routledge.Find this resource:

Tsujimura, N. (2014a). An introduction to Japanese linguistics (3d ed.). Oxford: Wiley Blackwell.Find this resource:


Akita, K. (2009). A grammar of sound-symbolic words in Japanese: Theoretical approaches to iconic and lexical properties of mimetics. Doctoral diss., Kobe University, Kobe, Japan.Find this resource:

Alfonso, A. (1974). Japanese language patterns. Tokyo: Sophia University.Find this resource:

Bowers, J. (1981). The theory of grammatical relations. Ithaca, NY: Cornell University Press.Find this resource:

Bruening, B. (2001). QR obeys superiority: Frozen scope and ACD. Linguistic Inquiry, 32, 233–273.Find this resource:

Chomsky, N. (1955). The logical structure of linguistic theory. New York: Plenum.Find this resource:

Davis, S., Tsujimura, N., & Tu, J. (2012). Toward a taxonomy of loanword prosody. Catalan Journal of Linguistics, 11, 13–39.Find this resource:

Dryer, M. (1987). On primary objects, secondary objects, and antidative. Language, 62, 808–845.Find this resource:

Fujimoto, M. (2015). Vowel devoicing. In Kubozono, H. (Ed.), Handbook of Japanese phonetics and phonology (pp. 167–214). Berlin: De Gruyter Mouton.Find this resource:

Fukuda, S., & Fukuda, S. (2001). An asymmetric impairment in Japanese complex verbs in specific language impairment. Cognitive Studies, 8, 63–84Find this resource:

Fukushima, K. (2005). Lexical V-V compounds in Japanese: Lexicon vs. syntax. Language, 81, 568–612.Find this resource:

Gibson, E. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68, 1–76.Find this resource:

Greenberg, J. (Ed.) (1963). Universals of language. Cambridge, MA: MIT Press.Find this resource:

Hamano, S. (1998). The sound-symbolic system of Japanese. Stanford, CA: CSLI Publications.Find this resource:

Hamano, S. (2000). Voicing of obstruents in old Japanese: Evidence from the sound-symbolic stratum. Journal of East Asian Linguistics, 9, 207–225.Find this resource:

Han, M. (1994). Acoustic manifestation of mora timing in Japanese. Journal of the Acoustical Society of America, 96, 73–82.Find this resource:

Harada, S. (1973). Counter equi NP deletion. Annual Bulletin, 7, Research Institute of Logopedics and Phoniatrics, University of Tokyo, 113–147.Find this resource:

Haraguchi, T. (1977). The tone pattern of Japanese: An autosegmentla theory of tonology. Tokyo: Kaitakusha.Find this resource:

Harley, H. (1995). Subjects, events, and licensing. Doctoral diss., Massachusetts Institute of Technology, Cambridge, MA.Find this resource:

Harley, H. (2002). Possession and the double-object construction. In P. Pica & J. Rooryck (Eds.), Linguistic variation yearbook 2 (pp. 31–70). Amsterdam: John Benjamins.Find this resource:

Harley, H. (2008). On the causative construction. In S. Miyagawa & M. Saito (Eds.), The Oxford handbook of Japanese linguistics (pp. 20–53). Oxford: Oxford University Press.Find this resource:

Hasegawa, Y. (2015). Japanese: A linguistic introduction. Cambridge, U.K.: Cambridge University Press.Find this resource:

Hirayama, M. (2008). Asymmetries in voiced stop gemination. In H. Kubozono (Ed.), Asymmetries in phonology: An East-Asian perspective (pp. 77–92). Tokyo: Kurosio.Find this resource:

Hoji, H. (1985). Logical form constraint and configurational structures in Japanese. Doctoral diss., University of Washington, Seattle, WA.Find this resource:

Inoue, K. (2000). Sentences with non-nominative subjects in Japanese. Masters thesis, Kanda University of International Studies, Chiba, Japan.Find this resource:

Irwin, M. (2009). Prosodic size and rendaku immunity. Journal of East Asian Linguistics, 18, 179–196.Find this resource:

Irwin, M. (2011). Loanwords in Japanese. Amsterdam: John Benjamins.Find this resource:

Ito, J., & Mester, A. (1986). The phonology of voicing in Japanese: Theoretical consequences for morphological accessibility. Linguistic Inquiry, 17, 49–73.Find this resource:

Ito, J., & Mester, A. (1995a). Japanese phonology. In J. Goldsmith (Ed.), The handbook of phonological theory (pp. 817–838). Oxford: Oxford University Press.Find this resource:

Ito, J., & Mester, A. (1995b). The core-periphery structure of the lexicon and constraints on reranking. In J. Beckman, L. Walsh Dickey, & S. Urbanczyk (Eds.), Papers in optimality theory (pp. 181–209). Amherst, MA: University of Massachusetts, GLSA.Find this resource:

Ito, J., & Mester, A. (2008). Lexical classes in phonology. In S. Miyagawa & M. Saito (Eds.), The Oxford handbook of Japanese linguistics (pp. 84–106). Oxford: Oxford University Press.Find this resource:

Ito, T., & Sugioka, Y. (2002). Go no shikumi to go-keisei [Mechanism of words and word formation]. Tokyo: Kenkyūsha.Find this resource:

Iwasaki, S. (2013). Japanese (rev. ed.). Amsterdam: John Benjamins.Find this resource:

Jacobsen, W. (1992). The transitive structure of events in Japanese. Tokyo: Kurosio.Find this resource:

Kageyama, T. (1993). Bunpō to gokeisei [Grammar and word formation]. Tokyo: Hitsuji Shobō.Find this resource:

Kageyama, T. (1999). Keitairon to imi [Morphology and meaning]. Tokyo: Hitsuji Shobō.Find this resource:

Kageyama, T. (2013). Goiteki hukugōdōshi no shintaikei. In T. Kageyama (Ed.), Hukugōdōshi kenkyū no saisentan [New explorations into the mysteries of compound verbs] (pp. 3–46). Tokyo: Hitsuji Shobō.Find this resource:

Katayama, M. (1998). Optimality theory and Japanese loanword phonology. Doctoral diss., University of California Santa Cruz.Find this resource:

Kawagoe, I. (2015). The phonology of sokuon, or geminate obstruents. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 79–120). Berlin: De Gruyter Mouton.Find this resource:

Kawahara, S. (2005). Voicing and geminacy in Japanese: An acoustic and perceptual study. In K. Flack & S. Kawahara (Eds.), University of Massachusetts occasional papers in linguistics, Vol. 31. Papers in experimental phonetics and phonology (pp. 87–120). Amherst, MA: Graduate Linguistic Student Association.Find this resource:

Kawahara, S. (2011a). Aspects of Japanese loanword devoicing. Journal of East Asian Linguistics, 20, 169–194.Find this resource:

Kawahara, S. (2011b). Japanese loanword devoicing revisited: A rating study. Natural Language and Linguistic Theory, 29, 705–723.Find this resource:

Kawahara, S. (2015). The phonetics of sokuon, or geminate obstruents. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 43–78). Berlin: De Gruyter Mouton.Find this resource:

Kishimoto, H. (2001). The role of lexical meanings in argument encoding: Double object verbs in Japanese. Gengo Kenkyū, 120, 35–65.Find this resource:

Kishimoto, H. (2008). Ditransitive idioms and argument structure. Journal of East Asian Linguistics, 17, 141–179.Find this resource:

Kita, S. (1997). Two-dimensional semantic analysis of Japanese mimetics. Linguistics, 35, 379–415.Find this resource:

Kitahara, H. (1994). Restricting ambiguous rule-application: A unified analysis of movement. In M. Koizumi & H. Ura (Eds.), MITWPL 24: Proceedings of FAJL 1 (pp. 179–209). Cambridge, MA: MIT Working Papers in Linguistics.Find this resource:

Kubozono, H. (2006). The phonetic and phonological organization of speech in Japanese. In M. Nakayama, R. Mazuka, & Y. Shirai (Eds.), The handbook of East Asian psycholinguistics, Vol. 2. Japanese (pp. 191–200). Cambridge, U.K.: Cambridge University Press.Find this resource:

Kubozono, H. (2008). Japanese accent. In S. Miyagawa & M. Saito (Eds.), The Oxford handbook of Japanese linguistics (pp. 165–191). Oxford: Oxford University Press.Find this resource:

Kubozono, H. (2015). Loanword phonology. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 313–362). Berlin: De Gruyter Mouton.Find this resource:

Kuno, S. (1973). The structure of the Japanese language. Cambridge, MA: MIT Press.Find this resource:

Kuroda. S.-Y. (1965). Generative grammatical studies in the Japanese language. Doctoral diss., MIT, Cambridge, MA.Find this resource:

Kuroda, S.-Y. (1978). Case-marking, canonical sentence patterns, and counter equi in Japanese. In J. Hinds & I. Howard (Eds.), Problems in Japanese syntax and semantics (pp. 30–51). Tokyo: Kaitakusha.Find this resource:

Labrune, L. (2012). The phonology of Japanese. Oxford: Oxford University Press.Find this resource:

Larson, R. (1988). On the double object construction. Linguistic Inquiry, 19, 335–391.Find this resource:

Lovins, J.B. (1975). Loanwords and the phonological structure of Japanese. Bloomington, IN: Indiana University Linguistics Club.Find this resource:

Manning, C., Sag, I., & Iida, M. (1999). The lexical integrity of Japanese causatives. In R. D. Levine & G. Green (Eds.), Studies in contemporary phrase structure grammar (pp. 39–79). Cambridge, U.K.: Cambridge University Press.Find this resource:

Martin, S. (1987). A reference grammar of Japanese. Rutland, VT: Charles E. Tuttle.Find this resource:

Matsumoto, Y. (1996). Complex predicates in Japanese: A syntactic and semantic study of the notion ‘word’. Stanford, CA: CSLI Publications.Find this resource:

Matsumoto, Y. (1998). Nihongo no goiteki hukugō dōshi ni okeru dōshi no kumiawase [The combination of verbs of Japanese lexical compound verbs]. Gengo Kenkyū, 114, 37–83.Find this resource:

Matsuoka, M. (2003). Two types of ditransitive constructions in Japanese. Journal of East Asian Linguistics, 12, 171–203.Find this resource:

McCawley, J. (1968). The phonological component of a grammar of Japanese. The Hague, Netherlands: Mouton.Find this resource:

Miyagawa, S. (1986). Restructuring in Japanese. In T. Imai, & M. Saito (Eds.), Issues in Japanese linguistics (pp. 273–300). Dordrecht, Netherlands: Foris.Find this resource:

Miyagawa, S. (1989). Structure and case marking in Japanese. San Diego, CA: Academic Press.Find this resource:

Miyagawa, S. (1997). Against optional scrambling. Linguistic Inquiry, 28, 1–26.Find this resource:

Miyagawa, S. (2005). EPP and semantically vacuous scrambling. In J. Sabel & M. Saito (Eds.), The free word order phenomenon (pp. 181–220). Berlin: Mouton de Gruyter.Find this resource:

Miyagawa, S. (2006). On the “undoing” property of scrambling: A response to Bošković. Linguistic Inquiry, 37, 607–624.Find this resource:

Miyagawa, S., & Saito, M. (Eds.) (2008). The Oxford handbook of Japanese linguistics. Oxford: Oxford University Press.Find this resource:

Miyagawa, S., & Tsujioka, T. (2004). Argument structure and ditransitive verbs in Japanese. Journal of East Asian Linguistics, 13, 1–38.Find this resource:

Miyamoto, E., & Nakamura, M. (2003). Subject/object asymmetries in the processing of relative clauses in Japanese. In G. Garding, & M. Tsujimura (Eds.), Proceedings of the 22nd west coast conference on formal linguistics (pp. 342–355). Summerville, MA: Cascadilla Press.Find this resource:

Muraki, M. (1978). The sika nai construction and predicate restructuring. In J. Hinds, & I. Howard (Eds.), Problems in Japanese syntax and semantics (pp. 155–177). Tokyo: Kaitakusha.Find this resource:

Nakayama, M., Mazuka, R., & Shirai, Y. (Eds.) (2006). The handbook of East Asian psycholinguistics, Vol. 2. Japanese. Cambridge, U.K.: Cambridge University Press.Find this resource:

Nasu, A. (2015). The phonological lexicon and mimetic phonology. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 253–288). Berlin: De Gruyter Mouton.Find this resource:

Nishimura, K. (2003). Lyman’s law in loanwords. Master’s thesis, Nagoya University, Japan.Find this resource:

Nishiyama, K. (1998). V-V compounds as serialization. Journal of East Asian Linguistics, 7, 175–217.Find this resource:

Oehrle, R. (1976). The grammatical status of the English dative alternation. Doctoral diss., MIT, Cambridge, MA.Find this resource:

O’Grady, W. (1997). Syntactic development. Chicago: University of Chicago Press.Find this resource:

Poser, W. (1981). The “double-o constraint”: Evidence for a direct object relation in Japanese.Find this resource:

Richards, N. (2001). An idiomatic argument for lexical decomposition. Linguistic Inquiry, 32, 183–192.Find this resource:

Rosen, E. (2001). Phonological processes interacting with the lexicon: Variable and non-regular effects in Japanese phonology. Doctoral diss., University of British Columbia, Vancouver, BC.Find this resource:

Rosen, E. (2003). Systematic irregularity in Japanese rendaku: How the grammar mediates patterned lexical exceptions. Canadian Journal of Linguistics, 48, 1–37.Find this resource:

Saito, M. (1985). Some asymmetries in Japanese and their theoretical implications. Doctoral diss., MIT, Cambridge, MA.Find this resource:

Saito, M. (2005). Further notes on the interpretation of scrambling chains. In J. Sabel, & M. Saito (Eds.), The free word order phenomenon (pp. 335–376). Berlin: Mouton de Gruyter.Find this resource:

Saito, M., & Hoji, H. (1983). Weak crossover and move-α‎ in Japanese. Natural Language and Linguistic Theory, 1, 245–259.Find this resource:

Sells, P. (2008). Oblique case marking on core arguments in Korean and Japanese. SOAS, University of London.Find this resource:

Shibatani, M. (1990). The languages of Japan. Cambridge, U.K.: Cambridge University Press.Find this resource:

Shibatani, M., & Kageyama, T. (Eds.). (2015–). Handbooks of Japanese language and linguistics (12 vols). Berlin: De Gruyter Mouton.Find this resource:

Sugioka, Y. (1986). Interaction of derivational morphology and syntax in Japanese and English. New York: Garland.Find this resource:

Sugioka, Y., Ito, T., & Hagiwara, H. (2001). Computation vs. memory in Japanese causative formation: Evidence from agrammatic aphasics. Cognitive Studies, 8, 37–62.Find this resource:

Takano, Y. (1999). Object shift and scrambling. Natural Language and Linguistic Theory, 16, 817–889.Find this resource:

Tamori, I., & Schourup, L. (1999). Onomatope—keitai to imi [Onomatopoeia—form and meaning]. Tokyo: Kurosio.Find this resource:

Tsujimura, N. (Ed.) (1999). The handbook of Japanese linguistics. Oxford: Blackwell.Find this resource:

Tsujimura, N. (Ed.) (2005). Japanese linguistics (Vols. 1–3). London: Routledge.Find this resource:

Tsujimura, N. (2014a). An introduction to Japanese linguistics (3d ed.). Oxford: Wiley Blackwell.Find this resource:

Tsujimura, N. (2014b). Mimetic verbs and meaning. In F. Rainer, F. Gardani, H. Christian Luschützky, & W. U. Dressler (Eds.), Morphology and meaning (pp. 303–314). Amsterdam: John Benjamins.Find this resource:

Tsujioka, T. (2011). Idioms, mixed marking in nominalization, and the base-generation hypothesis for ditransitives in Japanese. Journal of East Asian Linguistics, 20, 117–143.Find this resource:

Ueno, M., & Garnsey, S. M. (2008). An ERP study of the processing of subject and object relative clauses in Japanese. Language and Cognitive Processes, 23, 646–688.Find this resource:

Vance, T. (1987). An introduction to Japanese phonology. New York: SUNY Press.Find this resource:

Vance. T. (2015). Rendaku. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 397–441). Berlin: De Gruyter Mouton.Find this resource:

Yatsushiro, K. (2003). VP internal scrambling. Journal of East Asian Linguistics, 12, 141–170.Find this resource:

Yumoto, Y. (2005). Hukugō dōshi, hasei dōshi no imi to tōgo—mojūru keitairon kara mita nichieigo no dōshi keisei [Syntax and semantics of compound verbs and derived verbs—verb formation in English and Japanese from the perspectives of modular morphology]. Tokyo: Hitsuji Shobō.Find this resource:


(1.) I use the term “typology” to mean the various categorization of specific grammatical phenomena, such as the prosodic classification based on stress, tone, and pitch/accent, and the structural classification of relative clauses into head-initial and head-final. That is, I am not referring to the Greebergian typology (Greenberg, 1963) in terms of implicational universals. This terminological use will be applied throughout this article.

(2.) Iwasaki (2013) and Tsujimura (2014a) follow previous editions.

(3.) Geminates of voiced stops such as [bb, dd, gg] as well as geminates of fricatives like [hh, ΦΦ‎] tend to occur mainly in loanwords.

(4.) See Kawahara (2015) for geminates that appear word-initially in Japanese. His article and Kawagoe (2015) together provide a comprehensive overview of the phonetics and phonology of geminates internal to Japanese as well as the comparison with other languages with geminates.

(5.) The precise mechanism leading to the surface realization of the pitch contour requires reference to the Initial Lowering Rule of Haraguchi (1977) applicable to the Tokyo dialect.

(6.) A thorough examination of Japanese loanwords is found in Irwin (2011), where not only phonology, but also morphology, semantics, orthography, and historical background of Japanese loanwords are presented along with ample examples.

(7.) Irwin (2011) gives [š] as in [kyaššu] “cash” and [sutaiɾiššu] “stylish” as common exceptions. He further notes that the stop-fricative sequence in English like [ks] and [ps] are perceived as affricates, triggering gemination. Geminates in examples like [bokkusu] “box” and [čippusu] “chips” result from such perceived affricates.

(8.) Ito and Mester (2008) discuss the voicing alternation of geminates in loanwords as well as the borrowing of the English plural –s suffix. Some pluralized English nouns show both voiced and voiceless variants of the plural suffix, as in [ɾedi:zu]~[ɾedi:su] “ladies,’” while others are only possible with the voiced variant, as in [menzu]~*[mensu] “men’s.” In their analysis based on Optimality Theory, they note that, when a noun ends with a nasal sound, the variation is not available, and that this is consistent with the constraint that disallows voiceless consonants after a nasal sound in native words.

(9.) See endnote 7 for the same point.

(10.) McCawley (1968) uses “onomatopoeia” but notes that Japanese words that belong to this stratum are different from English onomatopoeia in that they “may refer to any aspect (visual, emotional, etc.) of the activity involved, rather than just its sound” (p.64). To make this point clear, I will use the term “mimetic,” following more recent literature on the lexical stratification.

(11.) See Matsumoto (1996) for a more detailed classification of the semantic relations that lexical V-V compounds exhibit.

(12.) A comprehensive discussion of Rendaku is found in Vance (2015).

(13.) Rosen (2003) reports that 75% of the native N-N compounds that meet the environments of Rendaku undergo voicing. See below for the regularity of non-applicability of Rendaku.

(14.) A refined version of Rosen’s Rule is found in Irwin (2009).

(15.) While there seems to be ample evidence to show that lexical causatives are mono-clausal, productive causatives do not exhibit properties exclusive to the bi-clausal structure. For example, productive causative sentences pattern with the mono-clausal structure in complying with the Double-o Constraint (Miyagawa, 1986) and the clausemate condition of negative polarity items (Muraki, 1978).