A comparative syntax of the formal politeness markers in Japanese and Korean: -Mas/-Des and -(Su)pni

1 Introduction

Several researchers discuss phenomena where the form of agreement encodes some aspects of the relation between speaker and addressee such as levels of politeness and/or formality. In the literature, this phenomenon is called allocutive agreement (Alok 2020; Antonov 2015; Ceong and Saxon 2020; Kaur and Yamada 2022; Miyagawa 2012; Miyagawa 2017; Miyagawa 2022; Oyharçabal 1993; Portner et al. 2019; Portner et al. 2022; Yamada 2019 among others). In his recent work, Miyagawa (2022) examines the properties of the formal politeness markers -mas and -des in Japanese, as illustrated in (1) and (2), arguing that these elements instantiate allocutive agreement.

Miyagawa argues that the allocutive φ-feature of -mas and -des has to undergo head-to-head feature movement to a special functional projection at the top of the clause, namely, the SAP,^[2] to have the entire utterance in its scope and mark it as being in the formal registry. More specifically, the allocutive φ-feature of -mas and -des is assumed to move up to the sa head, where it c-commands the addressee element and have its φ-feature valued.

(3)

As evidence for the head-to-head feature movement, Miyagawa points out that certain heads are only licensed in the presence of -mas or -des. For instance, the occurrence of the interrogative particle ka depends on the occurrence of -mas and -des, as shown in (4) and (5).

Similarly, in (6a), negation takes the form -en unlike the regular form in (6b). In fact, this is the only environment in which negation is realized as -en, i.e., the form -en only shows up in the presence of the formal politeness marker -mas.

Furthermore, once -mas is inserted, all relevant heads must undergo allomorphy for politeness.

Miyagawa argues that the properties in (4)-(7) are all due to the fact that the allocutive φ-feature of -mas and -des moves up through all head positions on its way to the SAP, licensing the relevant elements.

Against this background, I examine in this paper the Korean counterparts of -mas and -des – namely, -supnita and -ipnita, shown in (8).

Similarly to -mas and -des, -supnita and -ipnita are markers of formal politeness. They are normally inappropriate when speaking to one’s own child or close friend. I show below that there are several aspects where the behavior of -supnita and -ipnita can be better understood when they are analyzed in comparison with -mas and -des than when they are examined in isolation. Of course, the same goes for -mas and -des, i.e., some aspects of Miyagawa’s analysis of -mas and -des receive further support from the behavior of -supnita and -ipnita. I also point out a potentially general parametric difference in grammaticalization in Japanese and Korean that can only be revealed when these languages are examined together. The discussion in this paper clearly illustrates the merits of conducting comparative research on Japanese and Korean, languages that often show systematic similarities and differences, providing an ideal ground for uncovering deeper properties of human language.

2 Basic properties of the formal politeness markers in Korean

As is well-known, Korean is an agglutinative language. Many suffixes, usually assumed to be independent functional heads, attach to the verb stem. In Korean grammar, these verbal suffixes are referred to as emi ‘word ending’, often translated simply as “ending”. The verb stem (or root) in Korean is always morphologically dependent and can never be used in isolation without an appropriate ending. In particular, Kang (1988) introduces the notion of morphological closure, meaning that a verb stem has to combine with an appropriate ending to close it off to be a well-formed independent word. For instance, in (9), several endings are attached to the verb stem o- ‘come’. Among the endings, the last element, i.e., the declarative clause type marker -ta, is classified as a “word-final ending” in Korean grammar literally because of its position within the word. Here, -ta also functions as a “morphological closer”. Basically, any element that is used as a word-final ending on a verb also functions as a morphological closer. Other elements that come between the verb stem and the word-final ending, such as the honorification marker -si and the past tense marker -ess in (9), are classified as “pre-final endings”. From a morphological point of view, the word-final ending is the only obligatory element among all the endings that can attach to a verb stem.

With this basic knowledge of Korean verbal morphology and the terminology for it, let us look into the basic properties of the formal politeness endings -supnita and -ipnita below.

3 The interrogative clause type marker -kka

Recall that according to Miyagawa (2022), all the relevant heads on the path of the φ-features of -mas and -des undergo allomorphy to be in polite form. In particular, Miyagawa argues that the φ-features of -mas and -des can license the interrogative particle -ka. He argues these phenomena to provide support for the head-to-head feature movement analysis. In Korean, there is no comparable allomorphy phenomena. Thus, even if we follow Miyagawa and assume that the formal politeness marker should have the entire utterance in its domain by undergoing head-to-head feature movement to the SAP at the top of the structure, it is difficult to show it directly. Still, as shown in (12), repeated here as (20), Korean has an interrogative clause type marker, namely, -kka, that is used in sentences involving -(su)pni, similarly to -ka in Japanese.

A formal question involving the copula also licenses -kka arguably due to the presence of -(su)pni, as shown in (16), repeated here as (21).

Furthermore, an interrogative sentence involving -kka becomes ungrammatical if -(su)pni is omitted, confirming that the former depends on the latter for its licensing.

Of course, in such contexts, i.e., without -(su)pni, other interrogative endings than -kka can be used, as shown below.^[7]

However, in the presence of -(su)pni, the interrogative ending must be -kka.

Given this, it is clear that there is a dependency between the interrogative ending -kka and the formal politeness marker -(su)pni.^[8] Recall that Miyagawa argues that ka can be licensed by -mas/-des in the course of their feature movement to the SAP (or via selection by some particular type of verb in the main clause). This may be extended to Korean to the effect that -kka is licensed by -(su)pni in the course of its feature movement. Note incidentally that -kka can also be licensed in the presence of a modal, as shown in note 8. Given that the modal precedes -kka, just like -(su)pni, it may be necessary to assume that the modal undergoes feature movement, though where and why it moves require further investigation. Alternatively, it may be that -kka selects for the projection of -(su)pni and that of a modal. If this is the case, there may be a slight difference between Korean and Japanese regarding the licensing of the interrogative endings. For the moment, I remain open between these possibilities. Note however that whichever option is taken, head-to-head feature movement of -(su)pni to the SAP is independently motivated.

4 The positions of -mas, -Des, and -(Su)pni

Recall that one important assumption that Miyagawa adopts regarding the position of -mas and -des is that these elements should have the entire utterance in their scope at some point in the derivation. With that in mind, I turn in this section to the structural positions of the formal politeness markers in Japanese and Korean and show that there are interesting similarities and differences among them.

5 The distribution of the formal politeness markers and their root sensitivity

Several researchers note that the distribution of -(su)pni (as well as several other speech style particles in Korean) is limited to root clauses, i.e., they cannot be embedded (Hong 2018, 2022; Park 2019; Portner et al. 2019, 2022). Their analyses converge on the idea that the unembeddability of the speech style particles stems from the fact that they have to be in a local relation with the SAP, which is assumed to occur at the top of the utterance, i.e., at the root only.

In this respect, -mas is not different, i.e., it cannot occur in a complement clause.

However, Miyagawa (2022) points out that the distribution of -mas is not entirely limited to main clauses either. This is illustrated by the availability of sentences like (38b) and (38c).

Miyagawa points out that the contexts in (38) correspond to what Emonds (1970) defines as the root in (39).

Interestingly, Park (2019) also notes that the distribution of -(su)pni fits into Emonds’ definition of the root, as shown in (40).^[15] Note in particular that (40b) and (40c) show that -(su)pni can in fact occur in some non-root environments.

Finally, recall that I pointed out in Section 4.1 that the formal politeness markers -mas and -des1 in Japanese and the subject honorification marker -si in Korean are similar in that they are base-generated in an agreement projection between vP and TP and undergo φ-feature agreement with the relevant elements. But, one crucial difference is that the former elements need to establish a local relation with the addressee representation in the SAP in the root, which leads to their restricted distribution, while the latter only needs a local relation with the subject of the clause. As expected, -si is not confined to root environments, i.e., there is no problem for -si to be used in a complement clause, as shown in (41).

To summarize, the discussion in this section shows that the formal politeness markers in Japanese and Korean, namely, -mas, -des1, -des2, and -(su)pni, all behave in the same way in that they are restricted to Emonds’ root environments. According to Miyagawa (2022), Emonds’ root contexts are those environments that allow the occurrence of the SAP. Given that the formal politeness markers are required to establish a local relation with the addressee element in the SAP, their root sensitivity follows. This is where Japanese and Korean facts align perfectly. One point of difference though is that while -mas, -des1, and -si are all inserted in an agreement projection between vP and TP, -si is free to be embedded, as it does not depend on the SAP.^[16]

6 Differences in grammaticalization

At the end of Section 4, I briefly mentioned An’s (2022) work where it is argued that the evidentiality marker po- in Korean is a TP-external functional category. Evidential po- basically means something like ‘seem’ in English. It indicates that the speaker has some reasons, if not direct evidence, to believe that the proposition of the clause is true. Usually, the speaker makes an observation about the situation and makes inferences based on it. For instance, (42) can be uttered when the speaker sees Hana’s car parked in front of her house, though the speaker did not actually see her arrive.

Japanese has a similar expression—namely, sooda, as discussed by An and Maeda (2023). Sooda has more than one uses, but in one use, it means something like ‘seem’ and indicates that the speaker has some reasons to believe the proposition to be true. That is, just like evidential po-, the speaker makes an inference based on his/her observation of the situation.

Evidential po- and sooda are also similar in that they are functional categories, as indicated by the fact that they are unable to introduce their own arguments.

Interestingly, however, there is a clear difference in their structural positions. That is, evidential po- is above TP, as can be seen from the fact that the tense marker is attached to the main verb and shows up before po- in (42). Attaching the tense marker to evidential po- leads to ungrammaticality, regardless of whether the main verb tensed or not, as shown in (46). Note further that the impossibility of tense-marking po- also indicates that it is not the matrix verb in a biclausal structure unlike seem.

Being TP-external, it is also impossible for evidential po- to occur below negation.

In contrast, sooda can be tense-marked, based on which An and Maeda (2023) argue that it is below T and is thus TP-internal.

Furthermore, sooda can occur below negation, which is consistent with the proposal that it is TP-internal.

The interesting observation here is that evidential po- and sooda have equivalent functions and meanings, while their structural realizations are different. That is, evidential po- is TP-external, while sooda is TP-internal. This looks quite similar to the situation of -(su)pni and -mas. That is, while -(su)pni and -mas have equivalent functions and meanings, the former is TP-external and the latter is TP-internal. In other words, in each pair, the Korean counterpart is realized in a higher position in the TP-external domain, while the Japanese counterpart is realized in a lower position in the TP-internal domain. Of course, this may simply be a coincidence and does not reveal anything significant. An alternative possibility is that there exists a systematic difference in the way certain linguistic functions are grammaticalized in Japanese and Korean. That is, while the relevant elements are realized as high functional categories in Korean, their Japanese counterparts are realized in lower positions. If that is the case, it becomes necessary and also intriguing to see how pervasive and how systematic this parametric difference is between Japanese and Korean and perhaps, cross-linguistically and why such a difference exists, though that task has to await future research.

7 Conclusions

Taking as a point of departure Miyagawa’s (2022) recent discussion on the properties of the formal politeness marker (or the allocutive agreement marker) -mas and -des in Japanese, I have examined in this paper the properties of their Korean counterpart -(su)pni. I have shown that there are systematic differences and similarities between -mas/-des and -(su)pni. First, given that the their allocutive φ-features require valuation by the addressee element in the SAP, the distributions of -mas/-des and -(su)pni are perfectly identical. More specifically, both -mas/-des and -(su)pni are restricted to Emonds’ (1970) root contexts, which according to Miyagawa are environments that involve the SAP. Second, -mas/-des and -(su)pni are also similar in that certain elements, such as the special interrogative particles -ka and -kka, require their presence to be licensed, which according to Miyagawa (2022) is due to the head-to-head feature movement of -mas/-des (and -(su)pni). On the other hand, I have also shown that there is a crucial difference in the structural positions of -mas/-des and -(su)pni to the effect that -mas and -des1 are TP-internal and occupy the head position of an agreement projection that I argue also hosts the subject honorification marker -si in Korean, while -(su)pni occupies a position between TP and CP. Furthermore, I have pointed out that the difference in the structural height of -mas/-des and -(su)pni appears to be parallel to the difference in the structural height of evidential sooda and evidential po-, which may potentially be revealing the tendency where Japanese elements are realized in lower structural positions than their Korean counterparts. How general or pervasive this parametric difference is between the languages in question and cross-linguistically and also why such a pattern exists require further research. Finally, I should note that honorification and politeness markers in Japanese and Korean have been studied quite extensively by numerous researchers, which I cannot begin to do justice to in this short paper. Here, I have almost exclusively focused on Miyagawa’s analysis, but interested readers are referred to Portner et al. (2019, 2022), Yamada (2019), Hong (2022) for relevant discussion and references.