DM 1 Korean Honorification and Distributed Morphology Professor: Dr. Henderson Course: Issues in Morphology Date: December 17, 2008 Name: Sangyeon Park DM 2 I. Introduction The honorific system in the Korean language is notorious for its degree of complexity; there are different, variegated degrees of honorifics, which depend on who a speaker is talking to as well as the agreement of honorifics with the subject or the object in a sentence. Furthermore, while in Korean most of the verbs follow a default form by inserting –si‐ as a suffix to a verb, a restricted number of verbs have the allomorphy of their verb roots, i.e. the instantiation of the suppletive forms. It has been taken for granted in that the pairs of the same meaning are the net result of being stored in our lexicon, therefore having been introduced into our lexicon as separate lexical items without question. Few works, as far as I know, have forwarded serious questions about the basis and specifics as they pertain to the Korean honorific system specifically, in particular, the suppletive verb roots. In this paper, I shall illustrate the honorific system and the honorific suppletive roots’ forms in the Korean language and argue that they are better explained within the theoretical framework of Distributed Morphology (henceforth, DM) rather than in the confines of a lexicalist analysis. II. Syntax vs. Pragmatics In the wake of the controversial discussions debating the basis of whether Japanese Honorification (Boeckx & Niinuma, 2004) is syntax (obligatory) or pragmatics (optional), in the Korean language one also encounters the same linguistic quandary as the honorific system works almost within the same framework except for the fact that Korean does not use a the prefixed object agreement, as in the case of o‐ in Japanese. As for the system of Korean Honorification, it appears for linguists to be fuzzy on a certain level, making it difficult to draw the line between pragmatics and that of syntax. So, as to the question at hand will be “Is Honorification near enough to be adjacent or familiar to that of !"#$%#&’()*+,-",./0-123,4-5-"’6’(#&’-5,78,(-5)’38"83,&-,-58,-6,&08,9,(Phi) features such as number, person, and gender which are the essential features in syntactic agreement?” DM 3 In fact, there are a number of words which have [+honorific] feature present within their words in a lexical form to make sentences that sound pragmatically correct without any regard to syntax; the pairs ((pro)Noun‐(pro) Noun ([+HON]) such as bap‐jinji (or siksa) ‘rice,’ mal‐malssum ‘word,’ gu‐dangsin ‘he,’ byeng‐byenghwan ‘sickness’ as well as others. However, in the usage of lexical honorific words, the inappropriate usage of honorific words will make sentences appear to be awkward or make a speaker appear uneducated or impolite, but not necessarily ungrammatical. For example, ?? apessi‐ga bap‐ul du‐si‐n‐da father.HON‐NOM rice‐ACC eat.HON‐HON‐PRES‐DECL ‘Father eats rice’ Furthermore, the important thing present in the system of Korean Honorification is that of the relationship between a speaker who utters a sentence and a listener can be of paramount importance pragmatically (Strauss & Eun, 2005). It is thus that means that in the above sentence, apeci ‘father’ inherently can carry the [+HON] feature when the speaker is a person who ranked in a lower position regarding age or social rank. If I were to say to my grandfather or grandmother a sentence which contained the same meaning as referenced above, I would say the following sentence in its stead in this instance, which makes the sentence pragmatically and syntactically perfect, apessi‐ga bap‐ul du‐si‐m‐ni‐da father‐ rice‐ eat. ‐ .PRES‐DECL NOM ACC HON HON ‘Father eats rice’ Even to the listener who is ranked higher in age or in social status than a person which a speaker wants to honor, the same basic syntactic agreement occurs. With conforming to the honorific agreement obligatorily, as it regards the addressee levels of the Honorifics, the verb endings are diverse; for example, ‐ta (plain declarative), ‐e (intimate declarative), (e)yo (polite declarative), (su)pnida (deferential declarative) (Sohn, 1999, p. 413). The corresponding verb endings that relate to the addressee’s honorific levels convey some important nuances as well as a certain level of style and linguistic refinement. The different styles vary also not only in declarative sentences, but also in other types of sentences such as interrogative, imperatives, and propositive sentences, respectively. DM 4 ameni‐ga yoli‐lul ha‐si‐n‐da (plain declarative) mother‐NOM food‐ACC do‐HON‐PRE‐DEC ‘My mother does (cook) food’ ameni‐ga yoli‐lul ha‐si‐e (intimate declarative) ameni‐ga yoli‐lul ha‐si‐eyo (polite declarative) ameni‐ga yoli‐lul ha‐si‐pnida (deferential declarative) The verb endings that show another honorific agreement for the speaker‐addressee interrelationship shall be a topic for further discussion later on in the paper. While I do have to admit that the grammaticality judgment on the Honorifics can be slightly different from person to person and issue of the Korean (and Japanese) Honorification remains an unsolved issue about whether it is conditioned obligatorily or optionally, I draw a conclusion that the Korean honorific system like the Japanese one is processed by syntax by following the previous works (Kim, 1999; Sohn, 1999). Based on such findings, there is obvious agreement between the NP (noun phrase) and the predicate in the same way as the person, gender and number features. As it will be shown in the next section, the honorific is ungrammatical if the mismatch between the subject (or object in limited cases) and predicate in the agreement takes place. The role of agreement in the explanation of Korean verbal honorification can be one of the best options so as to explain the subject. Given this potentially contentious conclusion as it is in this paper, I will focus on explaining the limited number of suppletion of verb roots. III. Two different honorific agreements in Korean: Subject‐honorific vs. Object‐honorific 1. The Honorific System in Korean In this section, I will aim to illustrate in more detail the honorific system of syntactic agreement in Korean. Although it is often posited that the Korean language lacks a system of agreement, e.g., agreement between number, gender, person, etc. which is a fixture of a number of languages, Korean does show a strict honorific agreement system; subject‐honorific and object‐honorific system. Along DM 5 with the Minimalist approach, the division by the agreement clearly shows that the Honorific system is in point of fact a syntactic construct. 2. Subject‐honorific agreement 2.1. Default Form: It is the suffixation of –si‐ having [+human, +honorific] feature in verbs; [+HON] apeci‐ga ga‐si‐n‐da *apeci‐ga ga‐n‐da father‐ go‐ ‐ ‐ father‐ go‐ ‐ NOM.HON HON PRES DEC NOM.HON PRES DEC ‘Father goes’ [‐HON] dongsayng‐i ga‐n‐da *dongsaying‐i ga‐si‐n‐da brother‐ go‐ ‐ brother‐ go‐ ‐ ‐ NOM PRES DEC NOM HON PRES DEC ‘Brother goes’ [+HON] apeci‐ga chay‐ul ik‐u‐si‐n‐da *apeci‐ga chay‐ul ik‐nun‐da father‐NOM.HON book‐ACC read‐u‐HON‐PRES‐DECL ‘Father reads a book’ [‐HON]haksayng‐i chay‐ul ik‐nun‐da *haksaying‐i chay‐ul ik‐u‐si‐n‐da student‐ book‐ read‐ ‐ NOM ACC PRES DECL ‘The student reads a book’ As one can see in the two examples above, if the subject extant in a sentence should out of the societal norms be an object deserving in respect by a speaker, the agreement occurs obligatorily by adding the suffix –si‐ immediately after the verb roots such as ik‐‘read’ and ga‐‘go’; otherwise, the sentences are ungrammatical. 2.2. Suppletive forms (Root ‐Root with the same meaning) with the suffixation of –si‐ 1 2 [+HON] apeci‐ga cam‐ul cumu‐si‐n‐da Father‐ sleep‐ sleep. ‐ ‐ ‐ NOM.HON ACC HON HON PRES DECL ‘Father sleeps a sleep’ [‐HON] haksayng‐i cam‐ul ca‐n‐da Student‐ sleep‐ sleep‐ ‐ NOM ACC PRES DECL ‘The student sleeps a sleep’ The very limited number of verbs or adjectives have their suppletive pairs for verbal honorification; for example, ca‐cumu ‘sleep,’ apu‐pyenchan ‘be sick,’ and mek‐capsu(or du) ‘eat.’ The usage of the verb roots for the Honorifics can be illustrated as shown in the table below; DM 6 [‐HON] [+HON] (1) ik‐nun‐da ‘read’ ik(u)‐si‐n‐da (2) ca‐n‐da ‘sleep’ cumu‐si‐n‐da (3) ca‐n‐da *ca‐si‐n‐da (4) *cumu‐n‐da cumu‐si‐n‐da <Table 1:Usage of Verb Roots> (1) is a honorific default form without any modification to the verb root. In contrast, (2), (3) and (4) show that the honorific suppletive form for the word ‘sleep’, cumu‐, should be used correctly to make the sentences grammatical. Cumu‐ cannot be used with the subject [‐HON], nor can ca‐be used in the subject [+HON]. If a verb applies to the second type, (2), the Korean language does not allow for what isthe incorrect alternation as shown in (3) and (4). 3. Object‐honorific agreement As the previous woks indicated (Kim, 1999; Sohn, 1999), the object‐honorific agreement in Korean unlike that of Japanese is the least significant and productive. There seems to be no default form of the object‐ honorific agreement. If the sentence has a transitive verb that has an object, the main agent of the verb will be usually the subject so that there should be no reason to respect the object as far as the agent is not respected. In this sense, it may be a considerable amount of differences from Korean and Japanese in object agreement; the Japanese language exhibits a considerable amount of evidence that the object agreement as discussed in Boecksx and Ninuma’s paper is valid and extant (2004). However, the Korean language also exhibits a few examples that show that some of the verbs trigger the object‐honorific agreement (Kim, 1999; Sohn, 1999). The significant and productive example is the honorific case marker, kkey of honorific indirect object. Also, some of the verbs are realized lexically. I will introduce the examples of Korean Object‐honorific agreement here, but the in‐depth discussion of that issue, which must be an extensive work, is not the concern for this paper. DM 7 3.1. Honorific dative marker –kkey‐ [+HON] vs. dative marker –eke‐ [‐HON] dongsaying‐eke chak‐ul cu‐n‐da *dongsaying‐kkey chak‐ul cu‐n‐da brother‐ book‐ give‐ ‐ DAT ACC PRES DECL ‘(someone) gives my brother a book’ apeci‐kkey chak‐ul duri‐n‐da * apeci‐eke chak‐ul duri‐n‐da father‐ book‐ give. ‐ ‐ DAT.HON ACC HON PRES DECL ‘(someone) gives my father a book’ 3.2. Lexical realization as a suppletive form dongsayng‐i halmeni‐lul moseywa‐ass‐da * dongsayng‐i halmeni‐lul deryewa‐ass‐da brother‐NOM grandmother.HON‐ACC. bring(back).HON‐‐PAST‐DECL ‘Brother brought Grandmother’ dongsaying‐i ayki‐lul deryewa‐ass‐da * dongsaying‐i ayki‐lul moseywa‐ass‐da brother‐ baby‐ bring(back)‐ ‐ NOM ACC PAST DECL ‘Brother brought a dog’ IV. The analysis of the Korean Honorifics If I am on the right track in terms of analyzing these honorifics, the subject‐agreement and the object‐ agreement can be explained by AgrS and AgrO, respectively. According to Understanding Minimalism(Hornstein, Nunes, & Grohmann, 2005), it introduces a fine‐grained variant of Chomsky’s proposal on two projections of AgrS and AgrO (p.119). The following syntactic trees follow the basic mechanism of the proposal. apeci‐ga ga‐si‐n‐da nae‐ga apeci‐lul mosyewa‐ass‐da father‐ go‐ ‐ ‐ ‘Father goes’ I‐ father‐ bring. ‐ NOM HON PRES DEC NOM ACC HON PAST‐DECL [+HON] [+HON] <Object‐honorific agreement> <Subject‐honorific agreement> If we assume the Subject‐Honorific agreement as AgrSP, the syntactic tree for the above sentence can be drawn as follows(Kim, 1999, p. 44); DM 8 CP C’ TP C ‐da T’ AgrSP T apeci‐ga AgrS’ [+HON] VP AgrS ‐n [+PRES] V si‐ ga <Spec‐Head agreement> V. Lexicalist Approach vs. Distributed Morphology As I mentioned before, by following the previous arguments, I posit that the Korean honorific system is one of syntax rather than one of semantics, i.e. the feature agreement such that is shown in the person, number and gender in other languages. This paper mainly is concerned with how to deal with the suppletive verb roots (irregular form) that seem inherently to carry [+HON] feature, which is also added by the –si‐ suffix in subject‐honorific agreement. 1. Lexicalist approach From the lexicalists’ point of view, the honorific verbal roots, cumu‐, capsu‐ as well as others are generatively or computationally created in the lexicon in the same way that the other verbal roots which do not have any corresponding suppletive forms (Marantz, 1997). While the other verb roots do not have their suppletive forms make Subject‐honorific agreement by inserting the suffix –si‐, only a couple of verbs, cam‐cumu ‘sleep’ and muk‐capsu ‘eat,’ change their forms irregularly for the purposes of Honorification. They also take the suffix –si‐ as the other regular verbs make honorific predicates. If DM 9 the honorific suppletive forms are stored in the lexicon, why is it such that they do need to insert –si‐ again in the syntactic processing? The basis of such is that, the suppletive forms themselves are not an honorific pair for its own root, as the pairs of which do not show phonological similarities at all. The Korean verb roots can be the smallest word‐sized verb by combining with –da (DECL) such as muk ’eat’‐ da, ha’do’‐da, ik’read’‐da, in addition to others. On the other hand, the same thing does not occur to the suppletive forms; *cumu’ sleep, [+HON]’‐da, *capsu ’eat, [+HON]’‐da. The suppletive forms are not generative in within the lexicon. The irregular correspondences exist as a root in the lexicon, yet they are absent from the presence of the –si‐, and they cannot be lexical entries; this can only be with –si‐ suffix, yet they can be words such as ‘cumu‐si‐da,’ ‘capsu‐si‐da.’ The fact that the lexical items such as cumu‐ and capsu‐ can be meaningful only in the syntactic processing (agreement) indicates that morphology does not come before syntax as the lexicalists argue. In other words, if syntax has no access to the roots of words, it does not explain why the lexical items are influenced by syntactic construction after their morphology. It may be that in the lexicon, the speakers do not have two lexical entries for one meaning in terms of the Korean irregular honorific verb roots. Also, for the irregular allomorphies like ca‐cumu ‘sleep,’ muk‐capsu‐‘eat,’ the experiment (Stockall & Marantz, 2006) supports the idea that cumu‐ would not been stored in the form of ‘ca‐in the Honorific.’ It argues as such against a full listing approach; the Korean allomorphy examples are the extreme cases in terms of the phonological similarities. 2. In the Framework of Distributed Morphology 2.1. How DW works On the assumption that Korean Honorification is a syntactic processing (agreement), the lexicalist approach evidences a problem in that the honorific suppletive forms of verb roots happen before the syntax. Once again, if we had held to the morpheme‐based morphology positing that Word Formation Rules are computationally processed in the Lexicon, the explanation regarding having cumu‐ DM 10 si‐da or capsu‐si‐da in the lexicon is neither elegant nor is parsimonious. Why do we need to have –si‐ as an affix for other regular verbs and cumu‐si and capsu‐si at the same time as morphemes? Likewise, if the split‐morphology hypothesis is that inherent morphemes are processed in lexicon, but the contextual morphemes like cumu‐ and capsu‐ work after syntax is adopted, the problem of the honorific allomorphy can be better explained. 2.2. Distributed Morphology The concept of Distributed Morphology (referenced in Embick & Noyer, 2004; Halle & Marantz, 1993; Harley & Noyer, 1999) posits that there is no Lexicon in the sense that lexicalism uses the term. It minimizes the explicit listing of special sound/meaning pairings, argument structure alternations and categorical specifications. While DM argues that there are no lexical items in Lexicon, rather that there are only morphosyntactic features, l‐morpheme (which can be called a ‘root’) and f‐ morpheme (which can be called ‘licencer’) are listed in the inventory for syntactic operation by which in a DM every word is created. Instead of having no specific category such as a noun or a verb, the f‐ morphemes license the l‐morpheme category. The l‐morphemes (ROOTS) do not as such carry grammatical features; on the other hand, the f‐morphemes (abstract morphemes) possess non‐phonetic universal features. As for the phonological components of a word, underspecified vocabulary items will be inserted in the post‐syntactic stage. Before discussing further, if it is schematized in a simple way, it will be as follows (revised from (Harley & Noyer, 1999, p. 2); morphosyntactic features (Abstract morphemes +ROOTs) Syntactic operations Spell‐out Morphological Operations Vocabulary Insertion PF(phonetic form) LF(logical form)