Japanese/Korean Linguistics

Japanese/Korean Linguistics

William Giang, Lucien Brown, Shimako Iwasaki, Satoshi Nambu, and Daniel Pieper

This paper examines the temporal structure of intonation units (IUs) in English and Japanese using semi-controlled Covid-19 conversations recorded on Zoom. Intonation units, speech chunks demarcated by prosodic cues, were manually transcribed and analyzed for duration and preceding pause length. The study found Japanese IUs to be significantly longer (median 1.30 s) than English IUs (1.00 s), and Japanese pre-IU pauses longer (0.30 s vs. 0.10 s). Morphological factors, notably Japanese’s reliance on multisyllabic words versus English’s monosyllables, and pragmatic factors, greater encoding of discourse information and sociolinguistic markers in Japanese, explain these differences. Additionally, prolonged pauses in Japanese often trigger speaker shifts, whereas in English they occur at nongrammatical boundaries to support cognitive planning. The paper suggests extending this comparative approach to other languages and genres to uncover universal and language-specific IU patterns.

This paper investigates the morphosyntactic status of Korean 'particles', which should be reanalyzed as affixes, clitics, or words. Despite their central role in Korean grammar, few studies have acknowledged the existence of clitics in the language. Building on the foundational analysis in Chae (2020), this study revises and extends previous claims in two key directions. First, following Chae (2023a), we argue that the particles -i/ka and -ul/lul, conventionally treated as nominative and accusative case markers, are more accurately analyzed as delimiters with identical meanings. This reanalysis has broad implications for constructions traditionally explained in terms of case marking. Second, we propose a new lexical category of 'verbal postpositions' (vPs) to account for sentence endings, thereby eliminating the need to posit a complementizer phrase (CP) at the sentence level. We argue that the CP analysis is both conceptually and empirically problematic, as CPs are typically selected by elements within a clause. 

This paper aims to syntactically explore subject-object asymmetries of (thematic) topics in Japanese. In particular, I observe that in Japanese, a subject topic and an object topic can co-occur in the same clause, but that the latter must precede the former. Based on this observation, I propose that there are two topic projections in Japanese, and that the higher projection hosts an object topic, while the lower one hosts a subject topic. I then investigate each of these two projections in more detail: Regarding the higher topic projection, I argue that it can host not only an object topic but also a gapless topic. Regarding the lower topic projection, I claim that it is A/A'P, a projection proposed by Bošković (2024a,b) which has mixed A- and A'-properties. This claim gives an account of the word order asymmetry between a subject topic and an object topic.

Japanese and Korean use an existential predicate not only to express existence but also to convey possession and location (Lee 2008, Park 2009, Kim 2016, Tomioka 2007, Kishimoto 2016, among others). Korean uses the predicate iss- ‘be’, and Japanese uses aru/iru ‘be’. The fact that these patterns often coincide crosslinguistically has prompted earlier studies to propose a unified analysis for both constructions (e.g. Freeze 1992). Upon closer examination, however, it becomes clear that the two constructions do not exhibit the same patterns. In this paper, I propose an analysis that encompasses Japanese and Korean, arguing that locative and possessive constructions have distinct argument structures based on their distinct syntactic behaviors. The present study aims to provide a more systematic explanation of the complexities and differences inherent in each construction.

Cleft constructions have been one of the central issues in the field of child language acquisition. In order to deepen our understanding of children’s knowledge about these constructions, this study conducted a new experiment to determine whether Japanese-speaking preschool children are sensitive to the island effects in cleft constructions. The results of our experiment, which were obtained from 43 children between the ages of four and six, suggest that these children conform to the island constraints in clefts, which in turn gives prominence to the view that child language acquisition is supported by an innate faculty of language.

This study examines how Japanese and Korean differ in encoding events, using a basic classification of four information types: ENTITIES, HAPPENINGS, RELATIONS, and INCIDENTALS. Focusing on verb predicate sentences in translated American movie dialogues (2015–2023), it identifies three key patterns in Japanese: (1) predicates combining nominals and function verbs; (2) morphosyntactically incomplete predicates where function verbs are replaced or omitted, reducing explicit RELATIONS encoding; and (3) predicates lacking verbs entirely, omitting HAPPENINGS and relying on inference. Japanese translations tend to encode fewer HAPPENINGS and RELATIONS than Korean, favoring noun-based structures that support selective encoding and linguistic economy. In contrast, Korean encodes events more explicitly using fewer nominal constructions and preserving verbal forms. Japanese’s noun-oriented strategy, while less direct, enables prioritization of high-value content. These differences reflect broader tendencies in balancing efficiency and expressiveness and highlight the complexity of event encoding across languages, suggesting the need for further cross-linguistic analysis.

This conversation-analytic study presents an empirical analysis of the use of a Japanese interjection, hora. It focuses exclusively on the use of an X-hora-Y format and attempts to reveal how participants use this format to achieve an interactional goal. The analysis demonstrates that the X-hora-Y format is used when the speaker has exclusive access to an initial action (=X), and the subsequent use of hora then serves as a signal to initiate an action (=Y) that re-presents or reformulates the previous action in a shareable way. This format is thus used as an international practice to redress the epistemic imbalance between the speaker and the hearer and to establish a shared understanding. In his seminal paper, Heritage (2012) exemplified how the epistemic imbalance between the speaker and the hearer motivates a sequence of interactions. Even within the X-hora-Y format, a similar motivation is at work.

This study examines how motion events in the English original version of Harry Potter are translated into Japanese and Korean, with a focus on the encoding of Manner, the use of Deictic verbs, and ideophones. Following Talmy’s typology, the analysis compares 180 motion event descriptions in the English original with their translations in Japanese and Korean. The results show that while both languages generally preserve motion events, they employ different strategies to encode Manner: Japanese frequently uses ideophones and visual motion expressions, while Korean relies more on complex verbs incorporating Deictic elements. Although ideophones appear in both translations, their semantic roles differ; Japanese ideophones tend to represent detailed visual imagery, while Korean ideophones represent the overall intensity or emotional quality of the Motion. The findings highlight significant differences between two typologically similar V-languages, demonstrating how language-specific resources shape motion event descriptions in translation.

Southern Ryukyuan is one of the two main subgroups of the Ryukyuan language family, consisting of Miyako, Yaeyama, and Yonaguni. One of the notable features of the subgroup is word-initial fortition of Proto-Ryukyuan *w into b, exhibited by all three members; in contrast, only Yonaguni underwent *j > d, while the other two languages have retained the glide value for *j.

Due to the asymmetry, it is not entirely clear what value should be reconstructed for the Proto-Ryukyuan glides at the Proto-Southern Ryukyuan level. To answer this question, this paper investigates late fifteenth- and early sixteenth-century Korean transcriptions of Miyako in the Veritable Records of the Chosŏn Dynasty. In particular, it is argued that the initial [(ⁿ)ʑ] found in transcriptions of the name of Irabu Island faithfully reflects Proto-Miyako *[ʑ]. This evidence supports the reconstruction of Proto-Southern Ryukyuan *β and *ʑ as the reflexes of Proto-Ryukyuan *w and *j.

This paper discusses the issue of reconstructing Proto-Japonic demonstratives. Since demonstratives constitute a small and closed set, they are subject to internal pressures from other items within the paradigm. Therefore, a theoretical approach that emphasizes analogy, compounding, and reduction—processes operating within the system—is required.
Building on this framework, the paper explains the differences between Old Japanese and Proto-Ryukyuan as resulting from two reductive processes from Proto-Japonic: *sə- > ə- (> o-) in Proto-Ryukyuan, and ka- > a- in Proto-Northern Ryukyuan.
These developments support identifying the Proto-Japonic demonstrative system as comprising kə- and ka- for proximate and distal deictic nouns, respectively, and sə- for anaphoric nouns, with adverbial ka- used to refer to proximate manner.
Although the status of the anaphoric pronoun si in Old Japanese remains unclear, further investigation of the Ryukyuan languages may help clarify its relationship to sə- in Proto-Japonic.

This study examines the distribution of the accusative marker =NcjaN in Yoronese Ryukyuan through corpus analysis of 314 pages of folktales. Previous accounts claimed that =NcjaN is optional when objects are adjacent to verbs but obligatory elsewhere. However, analysis of transitive clauses reveals that while adjacency correlates with marking frequency (67% adjacent vs. 86% nonadjacent objects marked), some nonadjacent objects remain unmarked. The data show differential object marking patterns: animate and definite objects are more frequently marked than inanimate and indefinite ones, regardless of adjacency. Cross-dialectal comparison reveals cognate forms functioning as limitative particles (‘only’) in neighboring varieties of the Northern Ryukyuan languages, suggesting a grammaticalization path from limitative to accusative. I propose that contact-induced borrowing of Japanese =dake ‘only’ (=daki in Yoronese) triggered functional redistribution, with the indigenous =NcjaN shifting from limitative to accusative functions. These findings demonstrate how small endangered languages can develop unique case-marking strategies through internal reorganization catalyzed by language contact.

This study investigates the pivotal role of Japanese first-person singular (1SG) forms in self-praise episodes, utilizing an extensive analysis of over 50 hours of conversational data. The findings reveal a strikingly higher frequency of 1SG forms—44%, or 66 out of 150 self-praise episodes—contradicting the prevailing belief that these forms rarely occur in Japanese conversation. A detailed examination of these instances reveals that most do not semantically necessitate 1SG forms; they are divided equally into "emotive" and "frame-setting" categories. Emotive expressions often carry humor and are typically accompanied by laughter, while frame-setting functions as a crucial floor-holding device, signaling a transition to a more formal speech level. This study underscores the vital significance of 1SG forms in the construction of identity and social performance, illustrating their essential role in managing self-presentation and evaluation during interpersonal interactions.

A Korean Sign Language (KSL) toponym can emerge as a monomorphemic sign output indexed to the initial morpheme from a source Korean toponym; however, in some instances the KSL toponym creates an output mapped to the second source morpheme (Ji 2022). This shift flouts positional salience since KSL toponyms typically represent the initial source morpheme semantics. A comparison between the related sign languages Japanese Sign Language (JSL) and KSL offers a possible explanation. Ji (2022) observations about a collection of roughly 300 KSL toponyms were examined via a sign toponym typology developed in George (2022). An etiological examination of JSL shows that JSL toponyms generally index to semantically transparent Japanese source toponym morphemes (George 2022); KSL may share a similar process. The apparent crosslinguistic tendency to index semantically transparent morphemes may explain the non-indexation of an initial morpheme in KSL.

This article explores local configurations of features, resources, and symbolic relations which produce a distinctive urban semiotic style of 'Koreanness'. Situated in the context of Perth, Western Australia, the article draws on a corpus of 341 street signs recorded across different neighborhoods in the city, to illustrate how popular and traditional Korea cultural emblems, as well as a variety of culturally coded indexes, are used to produce intricate and highly layered local landscapes. Given the relatively small Korean population in Perth, Koreanness is explored as an emergent phenomenon shaped by the broader frame of ‘regionalization’— a process theorized as a stage in the making of nationally circumscribed landscapes. The implications point to complex patterns of reification of national and regional identities and relations in urban contact spaces, bringing about more nuanced understandings of how patterns of ‘nationness’, and in this case Koreanness, are produced and maintained away from the nation-state.

Some phonemic analyses of Old Japanese (OJ) include contrasts between consonant+vowel (CV) syllables and consonant+glide+vowel (CGV) syllables, where G is OJ/y/ or OJ/w/. For example, some scholars transcribe OJ with a contrast between OJ/Co/ and OJ/Cwo/. Arisaka’s Law is a set of cooccurrence restrictions on OJ syllables within a root, and in an analysis that incorporates this contrast between OJ/Co/ and OJ/Cwo/, Arisaka’s Law prohibits roots containing both OJ/Co/ and OJ/Cwo/. OJ also allowed glide+vowel (GV) syllables, including OJ/wo/, and it is natural to assume that OJ/wo/ in a GV syllable is phonemically the same as OJ/wo/ in a CGV syllable. If so, Arisaka’s Law predicts that the GV syllable OJ/wo/ should not occur in a root that also contains a CV syllable of the form OJ/Co/, but this prediction is false. This contradiction prompts consideration of alternative phonemic analyses that do not include OJ/Cwo/ syllables.

This paper investigates the Japanese comparison class marker -ni-shite-wa. Mizutani (2022) identifies two key observations: (i) the comparison class marker requires that its argument should have a low degree on a relevant scale, and (ii) the available interpretations of subjects vary depending on whether the predicate is stage-level or individual-level. This paper proposes a formal semantic analysis that accounts for these observations by incorporating a scalar presupposition introduced by the contrastive topic marker -wa (Sawada 2007), and by distinguishing between comparison classes formed from sets of distinct individuals and those formed from sets of temporal stages of single individuals (Toledo and Sassoon 2011). The proposed analysis contributes to ongoing debates on absolute and relative gradable adjectives, indirectly supporting the view that these two types of comparison classes play a crucial role in interpreting these adjectives.

We explore multiple Right Dislocation (RD) in Japanese, which has previously received very little attention. We argue that multiple RD is derived not by syntactic movement but by Prosodic Right Dislocation, evidenced by its insensitivity to syntactic constraints, its lack of LF interpretive effects, and its prosodic properties. We show that single RD obeys syntactic island constraints, whereas multiple RD is insensitive to syntactic islands. Furthermore, argument/adjunct asymmetries for binding which are exhibited under single RD disappear under multiple RD. In addition, phonetic evidence suggests that multiple RD involves the dislocation of a prosodic constituent – a Major Phrase – to which tonal downstep applies. The evidence suggests that though single RD is derived by syntactic movement of a syntactic constituent, multiple RD is not; rather, multiple RD is the product of right dislocation of a prosodic constituent within the phonology.

In this paper, we present a new analysis of the i-ochi construction in colloquial/informal Japanese. We demonstrate that our analysis solves several empirical puzzles with Konno's (2012, 2017) root small clause analysis. We also show that this analysis brings about several theoretical implications for other areas of Japanese grammar, including the nature and mode of structural case marking in Standard Japanese. Finally, we present novel data showing that the so-called saa-exclamative construction in Hichiku Dialect exhibits structural and functional similarities with the i-ochi construction and argue that our analysis developed for the latter can be directly transported to the latter. 

Boskovic (2005) lays out five prerequisites for Left Branch Extraction (LBE): (i) scrambling; (ii) the absence of DP; (iii) NP-over-AP structure; (iv) the left-edge condition; and (v) agreement (see also Boskovic 2008, 2012, 2013). Japanese seems to satisfy the prerequisites (i)-(iv), but not (v) agreement. If agreement is a strict requirement for LBE, languages that lack φ-agreement, such as Japanese, are expected to disallow LBE. However, if agreement is merely a preferred or default option for LBE, it is plausible that languages without agreement may employ an alternative strategy to make LBE possible. We argue that Japanese, which lacks φ-agreement, actually employs an alternative strategy based on prosody to facilitate LBE. Specifically, we argue that the low acceptability of LBE in Japanese results from a phonological restriction on the relevant movement, showing that the phonological restriction can be alleviated or overridden by prosodic factors related to focus and topic.

This paper argues that Japanese owns VP ellipsis observed in British English (do complement ellipsis, Light verb complement deletion) in terms of apparent VP ellipsis (VP ellipsis with a particle) and Sino-Japanese VP ellipsis. Apparent VP ellipsis cannnot be captured by vP ellipsis in English (Fujii 2016, Funakoshi 2020) and argument ellipsis (Sakamoto 2021), considering extraction, nominal property of VP with a particle, and selection. Light verb complement deletion approach is proposed to reconcile these issues. In addition, the current analisis is shown to be extended to Sino-Japanese VP ellipsis (Hayashi 2015). This extension explains its extraction possibilities, which is mysterious in vP ellipsis approach in Hayashi (2015). Finally, Phase Complement Condition, which stipulates only the phase head complement undergoes LF copy, while the others are deleted by PF-Deletion, is pointed out to correctly describe patterns of extraction from ellipsis such as vP ellipsis, predicate ellipsis, and (nominal) argument ellipsis.  

This paper reexamines the Korean aspectual marker -ko iss- by addressing its long-standing ambiguity between progressive and resultative readings. While earlier studies have linked resultative interpretations to lexical reflexivity, this study proposes a biclausal syntactic structure as the source of the resultative reading, contrasting with the monoclausal structure underlying the progressive meaning. Through syntactic diagnostics—such as adverbial placement and negation—as well as pragmatic factors including spatial, possessive, force-dynamic, and episodic involvement, this paper demonstrates that resultative readings are not inherent in the verb alone but arise through pragmatic inference. It argues for a revised understanding of the Korean aspect system, recognizing -ko iss- as an imperfective marker expressing progressive and habitual readings, while perfective aspect is marked by -e iss-. This syntactic-pragmatic perspective not only clarifies the function of -ko iss- but also contributes to the broader typological understanding of aspect in Korean.

This article presents new evidence that particle-stranding ellipsis in Japanese is best explained by PF-deletion rather than LF-copying. I first examine whether overt extraction out of an ellipsis site is possible, and conclude that such extraction is indeed possible, contrary to the prediction made by the LF-copying analysis. Our novel finding comes from hitherto unobserved prosodic evidence that particle-stranding ellipsis involving the negative polarity item -sika can bear focus intonation. I argue that this prosody-based evidence is unexpected under the LF-copying theory, as this analysis recovers syntactic structure only covertly and does not account for the observed intonational pattern. If the present analysis is on the right track, it not only contributes to a better understanding of elliptical phenomena in Japanese, but also offers an important perspective on ellipsis–prosody interactions, an area that remains underexplored in the literature.

In the literature, properties of scrambling out of control clauses in Japanese are not completely clear.  For example, although some argue that scrambling out of control clauses provoke weak-crossover effects, others claim that it does not (e.g. see Nemoto 1993, 1999, Takano 2009, 2010, Yoshimoto 2012, Goto 2017, and references therein).  Based on the previous studies (e.g., Skumoto 2024a), this paper endeavors to clarify its properties by providing more data and shows that weak-crossover effects are not provoked in scrambling out of control clauses.  Furthermore, I consider an analysis of scrambling out of control clauses from the perspectives of phases (cf. Chomsky 2000, Saito 2017a, Goto 2017, Sakumoto 2024a, among others).

This study offers a typological analysis of the Korean case particles poko and tele, traditionally treated as dative markers. Drawing on corpus data and construction grammar, it identifies four major constructions where these particles appear: speech quotation, naming, evaluative, and causative. Rather than marking indirect objects or recipients like typical datives (eykey, hanthey), poko and tele function primarily as addressee or goal markers. The study introduces the 'on-the-spot' constraint, requiring that the referent be directly involved in the speech context. Crosslinguistic evidence supports the grammaticalization paths SEE > ALLATIVE (for poko) and ACCOMPANY > ALLATIVE (for tele), aligning with well-attested typological trajectories. By mapping their semantic and syntactic behaviors, the paper argues for reclassifying poko and tele as specialized allative markers with pragmatic constraints. The findings refine our understanding of Korean case marking and contribute to broader typological models of grammaticalization across languages.

This study examines the acquisition order of Korean verb conjugational morphemes in first language (L1) development using Mean Length of Utterance (MLU) as a key indicator. Based on longitudinal spontaneous speech data from the Ryu Corpus (CHILDES), the study analyzes the emergence of grammatical forms such as tense, aspect, negation, and politeness. Results show that verb conjugational morphemes begin to appear at MLU 2.4–2.6, with significant grammatical expansion occurring at MLU 2.9–3.2. The acquisition follows a general sequence: present → past → imperfective → future. Negative forms, especially the pre-verbal an, are acquired early, while polite forms emerge later. The findings suggest that MLU is a more reliable measure of grammatical development than age, offering a more consistent framework for understanding Korean L1 acquisition.

Previous literature agrees (J. M. Yoon 1989, Kang 1998) that Korean native speakers differentiate three reflexive forms (caki, casin, caki-casin) via their binding distance preference. However, most previous empirical research has been conducted via comprehension/corpus-based methods; there is a lack of elicited oral production controlling for speaker consistency, making it difficult to fully understand the treatment of Korean reflexives by native speakers.  
 The current study investigated the treatment of Korean reflexives in comprehension and production. For local binding, caki-casin was uniformly preferred. For long-distance binding, however, comprehension and production diverged. In comprehension, speakers preferred caki, but caki-casin was produced most, despite being dispreferred in comprehension (p < 0.001). This misalignment can be considered evidence of caki-casin becoming a preferentially local reflexive licensed as a logophor in long-distance contexts, consistent with previous research (J.H. Kim & J. Yoon 2009, E.H. Kim & J. Yoon 2020, J.H. Kim & Y.H. Lee 2022).

This paper advances a theory of ‘Predicativization’ whereby a negatively intensifying adverb such as sappari ‘(not) at all’ in SpecNegP becomes reanalyzed as the Neg head, which then undergoes a categorial shift from Neg to A. I argue that the derived, X⁰ status of the adjectival predicate in question finds support in the unavailability of the predicative use of zenzen ‘(not) at all’ in the positively intensifying sense and that of sukosimo ‘(not) a bit’. The proposed Spec-to-Head reanalysis naturally extends itself to aspectual adverbs. In fact, the presence of predicative mada ‘(not) yet’ on the one hand and the absence of predicative moo ‘(not) any longer’ on the other lends empirical support to the syntactic nature of Predicativization. Predicativization, as understood here, aligns in spirit with syntactic analyses of lexicalization (Hale & Keyser 1993, 2002, Marantz 1997, among others) and is considered a case of ‘antigrammaticalization’ (cf. Haspelmath 2004).

This study investigates specificational afterthought right dislocation (ARD) in Japanese, where a sentence-final constituent (appendix) clarifies a pronoun-like element (correlate) within the main clause. Unlike prior biclausal or antisymmetric accounts, this paper proposes a mono-clausal, head-final structure incorporating topic and focus movement within a cartographic framework. The analysis explains the observed properties of ARD—such as island sensitivity, anaphor binding, and case/particle matching—by positing that the correlate and appendix form a base DP, with the appendix undergoing movement to the right periphery. Crucially, the study accounts for the fixed ordering of multiple sentence-final discourse particles (e.g., wa > yo > ne) by aligning them with distinct syntactic projections. This approach resolves issues faced by previous models and provides a unified derivation for both simple and particle-rich ARD constructions, reinforcing a head-final structure while maintaining the explanatory benefits of cartographic syntax.

Japanese noun phrases are not morphologically inflected for singular/plural contrast. However, the language employs several optional plural markers. This paper argues that while Japanese lacks morphological singular/plural distinctions, the distinction is reflected semantically, as plural markers can only attach to count nouns. Additionally, the most productive plural marker, -tati, is found to be exclusively associative. This finding supports the nominal mapping hypothesis proposed by Chierchia (1998a, 1998b, 2010), indicating that Japanese does not serve as a counterexample to the framework.

This study tested predictions of two hypotheses for scrambling. Cyclic Linearization (CL) argues that the word order established in a syntactic cycle cannot be overridden in a later cycle. Prosodic Scrambling (PS) claims that a string of words must be scrambled in syntax if it represents a syntactic constituent but scrambled at PF if it does not. Two experiments were designed with Japanese transitive sentences containing an adverb. Experiment 1, a production task, found that the A(dverb)S(ubject)O(bject)V(erb) order is overwhelmingly preferred with a modal adverb, whereas ASOV and SAOV are equally preferred with a temporal adverb. Experiment 2 examined acceptability of transitive sentences with the two types of adverbs in four orders: ASOV, AOSV, OSAV, and SOAV. CL predicts SOAV is underivable, while PS predicts all four orders are derivable. The results show that SOAV received the second highest mean rating, contradicting CL but consistent with PS. 

This paper examines the nature of passive sensitivity, focusing primarily on Japanese MA adverbs. We have aimed to test Kubota’s (2015) claim that MA adverbs in Japanese can adjoin at either a high or low position in the structure, yielding both SS and DS interpretations in passives. Using pseudoclefts, we examined whether each reading can be obtained while controlling for the adverb’s adjunction site. Our results provide empirical support for Kubota’s (2015) proposal. 

This paper focus on the syntactic properties of psychological adverbs (PAs) constructions in Japanese. Matsuoka (2021, 2023) argues that PAs in Japanese have a CP structure, in which the experiencer PPs occupy Spec, TP of the PAs, and a null operator movement to Spec, CP is involved. However, I argue that the evidence in Matsuoka (2021, 2023) is faced with serious problems in both empirical and theoretical respects; therefore a new  analysis is called for. I propose that PAs in Japanese have a simpler structure, in which neither TP nor CP is projected, and that the experiencer PPs are not syntactically included in PAs; they are adjoined to VP. It is also shown that the structure proposed can account for the various syntactic properties of PAs in Japanese, demonstrating the validity of my proposal.

Terms of address in Japanese have been studied abundantly, with particular attention paid to their referential use in daily life, for instance, kinship terms and personal pronouns. In order to broaden our understanding on terms of address, however, it is essential to describe and analyze less-frequently-used and 'marked' terms of address. This study paid particular attention to the addressee–reference usage of first-person pronouns watasi and watasitati. Specifically, these pronouns can be employed typically by male adult speakers to address younger adult female addressee(s). These pronouns are strategically chosen to satisfy conflicting sociolinguistic/communicative needs that cannot be satisfied by genuine second-person pronouns. The pronouns (watasi and watasitati), which copy the forms originally used by addressees to refer to themselves, neither sound too close nor too formal, striking a delicate balance in terms of interpersonal distance.

This study investigates the ‘to omou’ construction in Japanese, [a complement clause + a quotation marker + omou ‘think’], with the aim of identifying recurrent forms associated with the construction and analyzing their functions in natural conversation. The data come from the Corpus of Everyday Japanese Conversation. Our quantitative analyses identify various distributional patterns surrounding the construction, including the ending forms of the complement clause, combinations of quotation markers and conjugated forms of omou, and the construction’s position within an utterance. We also conducted qualitative analyses of the discourse functions of toka omotte and found that it serves multiple functions, such as explaining the reason for an action or quoting one’s thought in a complaint. From a constructional perspective, our findings suggest that various recurrent patterns should be taken into consideration to adequately capture the multiple functions of the complement-taking predicate omou in Japanese conversation.

This study explores morphosyntactic variations in monologic speech across Japanese, Korean, and English, with particular focus on shifts from monologic to other-directed utterances. From typological and grammatical perspectives, Japanese and Korean are generally considered to be more analogous to each other than to English. Nevertheless, Japanese exhibits a notable contrast with both Korean and English in employing prepatterned or prefabricated expressions in monologic speech, which are distinct from those employed in corresponding interactional contexts. This divergence aligns with another cross-linguistic pattern: in utterances expressing self-blame or self-encouragement, Japanese generally disallows second-person pronouns to refer to the monologic speaker, whereas both Korean and English readily permit such self-reference. These findings suggest that Japanese can be best characterized as inherently distinguishing between monologic speech (i.e., undirected and self-directed) and other-directed speech (i.e., dialogic or interactive) and requiring speakers to adopt distinct morphosyntactic prepatterned or prefabricated expressions accordingly.

This study investigates referential density (RD) in the history of Japanese through manually annotated data and corpus analysis. Following Bickel’s (2003) RD1 (the ratio of overt argument NPs to available argument positions) and Noonan’s (2003) RD2 (the ratio of overt arguments to verbs), I introduce new indices to quantify the referential contribution of exaltation morphology (keigo). Drawing on prose texts from Old Japanese (OJ) and their contemporary Japanese (cNJ) translations, I demonstrate a diachronic increase in overt arguments (RD2) alongside a decline in honorific marking. While these changes might appear causally related, statistical testing reveals no direct correlation. As a preliminary conclusion, I hypothesise that the observed trends are discourse-driven and not structurally conditioned by morphosyntactic change.

This study analyzes remnant Japanese spoken by an elderly Korean woman educated during Japan’s colonial rule. It examines phonetic, grammatical, and lexical features influenced by Korean, revealing language attrition and cross-linguistic interference.

This study explores obstruent characteristics in Korean dialects, particularly distinctions between lax, tense, and aspirated sounds, using phone-level speech recognition on a 2,200-hour corpus. Using the AI-Hub senior dialect speech dataset, we analyzed Gyeongsang (1,203 hours) and Jeolla (1,015 hours) dialects. A G2P model and wav2vec 2.0 XLS-R speech recognition model compared canonical with actual phone sequences.

Gyeongsang dialect showed lax-tense confusion and lax-to-aspirated changes, while Jeolla dialect exhibited aspirated-to-lax changes. The lax-tense merger in Gyeongsang extends beyond previous /s/-/s*/ merger findings to general obstruents. Obstruent deletion in onset and coda positions in Gyeongsang represents a previously unreported phenomenon. The study demonstrates the effectiveness of large speech corpora for dialectal analysis, confirming known patterns while revealing new obstruent variations. Future research should examine how deletions vary by word position.

Compound tensification (CT) in Korean refers to the process where an underlyingly plain obstruent surfaces as tense at the onset of the second noun in a noun + noun compound, traditionally attributed to a historical genitive marker. Thus, CT is known to only affect subordinate compounds, where the first noun modifies the second noun. However, this study tests whether CT generalizes to coordinate compounds, where the two nouns form a conjunctive relation. In a production experiment with 33 native Seoul Korean speakers, tokens containing plain stops at compound boundaries were elicited and perceptually categorized. Results show that while CT occurred more frequently in subordinate compounds, it also emerged in coordinate compounds, especially in shorter, high-frequency words like pom-kaul (‘spring and fall’). Acoustic analysis confirmed that stops perceived as tense exhibited significantly longer closure durations across both compound types. These findings suggest an expansion of CT beyond its historically constrained environment.

In spoken Japanese, adjectives are used in clipped (or shortened) forms like ita, itta, ite ‘painful.’ A similar phenomenon can also be observed in colloquial Korean. Focusing on bieub (ir)regular adjectives (e.g., mugeobda ‘heavy’), the present study identifies two types of clipped forms: bieub-absent form (e.g., mugeo) and bieub-present form (e.g., mugeob). A questionnaire survey was conducted on Korean native speakers to investigate the use of thirteen bieub (ir)regular adjectives in the two clipped forms. The results revealed that the bieub-absent form is characterized by four factors (semantic, morphological, phonological, and sociopragmatic factors), while the bieub-absent form mainly involves two factors (morphological and sociopragmatic factors). It is suggested that adjectives with the stem vowel -eo are preferred in the bieub-absent form due to the influence of phonological contraction and the increasing use of the bieub-present form is motivated by the widespread adoption of computer-mediated communication (CMC).

This paper investigates verb reduplication (VR) in Japanese, focusing on three key properties: (i) the obligatory insertion of the light verb su, (ii) iterative aspectual meaning, and (iii) total reduplication of the verbal stem. Drawing on the framework of Distributed Morphology, I argue that VR is mediated by an aspectual head (Asp) that projects between vP and VoiceP. This head simultaneously blocks head movement (triggering su insertion), introduces the [iterative] aspectual feature, and undergoes morphological reduplication via a RED morpheme. The analysis accounts for the syntactic, semantic, and morphophonological properties of VR and extends to related constructions like hito-V compounds. The findings contribute to our understanding of aspectual morphology and the interaction between syntax and morphology in Japanese.

Relative tense is a phenomenon where the reference time of a clause is determined based on some context-dependent time, not necessarily the speech time. While relative tense can be observed with Japanese subordinate clauses, most previous studies have focused only on a single level of embedding, leaving cases of multiple embedding underexplored. Against this backdrop, this study investigates the temporal interpretations of multiply embedded relative clauses in Japanese. We report that a multiply embedded relative clause can receive a relative tense interpretation that depends on its nonimmediately higher clauses. We also show that this 'clause skipping' interpretation is constrained by the tense forms of the clauses. Based on these observations, we then develop a formal account of the temporal interpretations of Japanese relative clauses. Our account employs the notion of a time interval list, which enables us to derive the attested interpretations compositionally.

In formal Japanese, an argument nominal must be accompanied by a particle (ga, wa, etc.). In colloquial Japanese, on the other hand, an argument nominal may be bare, or ZERO-marked. While in many cases ZERO may alternate with (i) nominative-marking ga or accusative marking o, or (ii) the information-structural particle wa, under certain configurations ZERO is preferred to ga/o or wa, and vice versa. This work discusses information-structural conditions licensing and motivating wa- and ZERO-marking on argument nominals in colloquial Japanese. Additionally, it makes a proposal as to how wa-marking on adjuncts is regulated in both colloquial and formal registers. 

This study explores one facet of idiomatic creativity: lexical substitution, focusing on substitution for nouns in Korean, Japanese, and English idioms. Analyzing 15 idioms per language, we identify patterns in how speakers creatively replace nouns in the canonical idioms, revealing cognitive and linguistic strategies for idiom manipulation. These strategies serve communicative goals such as frame role assignment, reducing the burden of referentiality, and contextual relevance. We highlight two key findings: (1) three distinct patterns of synonymy based on the replaced noun’s literal, figurative, or extended meaning, and (2) the emergence of rare collocations which are nevertheless immediately interpretable - a phenomenon which we argue stems from the complex referential demands of idiomatic expressions.

This study examines the semantic evolution of the Korean adverb iltan, which originally denoted a short duration and later developed into a conditional marker. Initially expressing brief temporal span, iltan came to indicate temporal priority among alternative actions and eventually grammaticalized into a marker of sufficient condition—signaling that once event A occurs, event B follows. The semantic shift is argued to result from the interplay of speaker intention, perceived temporal proximity, and event prioritization. A comparison with the Chinese cognate yidan highlights how these forms diverged cross-linguistically. This synchronic study provides a basis for future diachronic and typological research on grammaticalization in East Asian languages.