Changguk Yim

The unembeddability of imperatives in Korean: Two different types of imperative morphology

De Gruyter | 2021


This article presents additional data in support of the fact that imperatives cannot be embedded in Korean. It demonstrates that the language employs two different types of imperative morphology: one that occurs in main clauses and the other that occurs in embedded environments, and that their occurrence is mutually exclusive. That being the case, the main imperative morphology is a bona fide illocutionary force marker that is syntactically encoded in the main clauses only, whereas the embedded imperative morphology simply serves as a clause-type indicator with no illocutionary force.

1 Introduction

Han (1998: 113–4) argues that imperatives cannot be embedded in Korean (see also Sadock and Zwicky 1985, Rivero and Terzi 1995, among many others), as shown in (1) below (from Han 1998: 187).

(1) a. Ppalli o-ala.
quickly come-imp
“Come quickly.”
b. *Na-nun Mary-eykey ppalli o-ala-ko myenglyenghassta.
I-top Mary-to quickly come-imp-C ordered
“I ordered Mary to come quickly.”

Han claims that the sharp contrast in grammaticality between (1a) and (1b) attests to the fact that the imperative is unembeddable in the language.

However, the unembeddability of Korean imperatives is complicated by the fact that it is possible that another similar particle occurs in embedded clauses, as demonstrated in (2) (taken from Portner (2007: 13) with minor romanization modifications; the grammaticality judgment is attributed to Miok Pak).

(2) Inho-ka Sooni-eykey cip-ey ka-la-ko malhayssta.
Inho-nom Sooni-to home-to go-imp-C said
“Inho said to Sooni to go home.”

The imperative particles -ala in (1) and -la in (2) look alike but are not identical. More interestingly, the -la can appear in the main clause (as well as in the embedded clause), as shown in (3) below.

(3) a. Kukes-ul sa-la.
it-acc buy-imp
“Buy it.”
b. Ce-A-nun Celin-eykey kukes-ul sa-la-ko myenglyenghassta.
Ce-A-top Celin-to it-acc buy-imp-C ordered
“Ce-A ordered Celin to buy it.”

This leads some linguists to the conclusion that Korean allows embedded imperatives (see, e.g., Platzack 2007, Portner 2007, Pak 2008, Lee 2012, and Lee and Park 2014, among others). In this article, I argue that the language at stake employs two different types of imperative particles: specifically, -(a)la is a bona fide imperative particle that may only occur in main clauses, while -(u)la simply serves as a clause-type marker that may only appear in embedded clauses. This constitutes substantive support for the fact that imperatives are unembeddable in the language.

2 Two different types of imperative morphology

2.1 -(a)la vs -(u)la

Let us consider the sentences given in (4):

(4) a. Kukes-ul kam-{-ala/*-ula}.
it-acc wind-imp
“Wind it up.”
b. Ce-A-nun Celin-eykey kukes-ul kam-{-ula/*-ala}-ko myenglyenghassta.
Ce-A-top Celin-to it-acc wind-imp-C ordered
“Ce-A ordered Celin to wind it up.”
c. kukes-ul kam-{-ula/*-ala}-nun myenglyeng
it-acc wind-imp-C order
“the order that (you should) wind it up.”

The examples in (4) show that -ula and -ala have a mutually exclusive distribution: In (4a), only -ala can occur in roots.[1] In (4b), -ula is embeddable whereas -ala is not. And in (4c), only -ula is possible in complementation. This stark contrast in distribution demonstrates that the language under consideration draws a definite distinction between main and embedded contexts for imperatives. As summarized in (5), -(a)la only appears in main clauses, whereas -(u)la only appears in embedded contexts.[2]

(5) Embedded Main[3]
-(u)la -(a)la

2.2 Special imperative particles only for roots

Having attested that -(a)la is the root imperative marker, while -(u)la is the embedded imperative marker, let us now examine some further evidence in favor of the unembeddability of Korean imperatives.

Other than -(a)la, Korean employs special imperative morphology that only follows certain verbs; for example, the particles -nela and -kela. Consider the sentences in (6) below:

(6) a. Ili o-nela.
here come-imp
“Come over here.”
b. Celi ka-kela.
there go-imp
“Go over there.”

It is interesting to note that -nela combines with o- “come” only: that is, it does not follow any other verb than that verb. In the same vein, -kela has such a narrow distribution: it combines a limited set of verbs such as ka- “go.” Of particular importance is that neither of these imperative particles is embeddable, as illustrated in (7) below:

(7) a. *Ce-A-nun Celin-eykey ili o-nela-ko myenglyenghassta.
Ce-A-top Celin-to here come-imp-C ordered
“Ce-A ordered Celin to come over here.”
b. *Ce-A-nun Celin-eykey celi ka-kela-ko myenglyenghassta.
Ce-A-top Celin-to there go-imp-C ordered
“Ce-A ordered Celin to go over there.”
c. Ce-A-nun Celin-eykey {o-/ka-}-la-ko myenglyenghassta.

As demonstrated in (7a) and (7b), the particles under consideration cannot be embedded, and in order to embed them, they would need to be replaced by the embedded imperative particle -(u)la, as demonstrated in (7c). This strongly confirms the fact that there are two different types of imperative marker that are subject to embeddability in Korean.

The imperative particle -lyem is also unusual. As shown in (8) below, it is used to make a soft command or order to an addressee who is younger than the speaker.

(8) a. Cokum te cham-ko kitali-lyem.
a.bit more endure-and wait-imp
“Be a bit more patient and wait.”
b. Cal swiess-uni yelsimhi kongpwuha-lyem.
well rested-because hard study-imp
“Since you have had a good rest, work hard.”

What is interesting about -lyem is that the soft imperative cannot be embedded, as demonstrated in (9a) and (9b), and that in order to embed it, once again, the -(u)la particle should be used, as illustrated in (9c) and (9d).

(9) a. *Ce-A-nun Celin-eykey cokum te cham-ko kitali-lyem-ko myenglyenghassta.
Ce-A-top Celin-to a.bit more endure-and wait-imp-C ordered
“Ce-A ordered Celin to be a bit more patient and wait.”
b. *Ce-A-nun Celin-eykey cal swiess-uni yelsimhi kongpwuha-lyem-ko myenglyenghassta.
Ce-A-top Celin-to well rested-because hard study-imp-C ordered
“Ce-A ordered Celin to work hard since you had a good rest.”
c. Ce-A-nun Celin-eykey cokum te cham-ko kitali-la-ko myenglyenghassta.
Ce-A-top Celin-to a.bit more endure-and wait-imp-C ordered
“Ce-A ordered Celin to be a bit more patient and wait.”
d. Ce-A-nun Celin-eykey cal swiess-uni yelsimhi kongpwuha-la-ko myenglyenghassta.
Ce-A-top Celin-to well rested-because hard study-imp-C ordered
“Ce-A ordered Celin to work hard since you had a good rest.”

Note that special imperative particles such as -nela/-kela/-lyem only occur in main clauses, that is to say, they are not embeddable.[4]

It should be noted that apparently, as noted in fn. 1, -(u)la can also occur in a root context. It is controversial whether the so-called “indirect imperatives” which have no addressee are a true form of imperative. Im (1983) refers to “-(u)la root imperative” such as (10a) as “absolute sentence” (celtaymwun), and he argues that it behaves differently from a normal root imperative (see Im 1983 for detail). One of the most striking differences between the two types of imperatives is that, as illustrated in (10), in case of -(u)la indirect imperative (10a), the first-person reflexive is licit. By way of contrast, in case of -(a)la direct imperative (10b), the first person is illicit.

(10) a. Na casin-ul chac-ula.
I self-acc find-imp
“Find my own self.” (OK as a book title)
b. *Na casin-ul chac-ala.
I self-acc find-imp
Lit. “Find my own self.”

This fact can be explained straightforwardly by the fact that the -(u)la indirect imperative has no illocutionary force, since it has no addressee to perform the order or command in the working discourse.

The fact that an -(u)la indirect imperative has no addressee can be well understood under Se 1985. Se analyzes -ala as a complex: -a + -l + -a. According to his analysis, the (boldfaced) -a morpheme marks the presence of addressee. That being the case, the fact is straightforwardly explained that -(u)la indirect imperatives have no addressee. Due to the lack of -a addressee-marking, -(u)la indirect imperatives can only be used as “absolute sentence” as in a book title or strike slogan. In embedded contexts, they only function to “type” the embedded clause to which -(u)la attaches.

3 Theoretical implications

Thus far, we have attested that -(a)la/-nela/-kela/-lyem only occur in main clauses – that is, they are unembeddable – whereas -(u)la appears in embedded contexts, as summarized in (11) below.

(11) Embedded Main
-(u)la -(a)la/-nela/-kela/-lyem

This fact, illustrated in (11), can be explained straightforwardly if it is assumed that -(a)la/-nela/-kela/-lyem make up the bona fide imperative morphology that bears directive illocutionary force (Han 1998), while -(u)la is simply a clause-type marker with no illocutionary force. Suppose that the main imperative morphology is an illocutionary force indicator of imperatives, which is syntactically encoded in the discourse-related domain of the clause, Speech Act Phrase (SAP), as illustrated in (12) below (see Haegeman 2014, Haegeman and Hill 2010, Pak 2008, Speas 2004, Speas and Tenny 2003, and Tenny 2006, among others).


SAP can best be understood as a syntactic projection that is involved with the performative acts and that encodes the directive illocutionary force of imperatives. In this view, the -(a)la/-nela/-kela/-lyem particles are exponents of the SA head. It is reasonable to assume that the -a morpheme in -(a)la that indicates the presence of addressee is associated with SAP.

By way of contrast, the embedded imperative morphology only marks the clause-type of the embedded clause in which it appears: -(u)la indicates that the event of the embedded sentence is not realized at the referent time (i.e., the event time of its matrix clause), and that it can occur in the future with respect to the referent time. In this view, -(u)la simply refers to “the modality of unrealized interpretation” (Han 1998) and, that being the case, it is reasonable to conclude that it is not associated with the discourse pragmatic domain of clausal structures like SAP: hence, it can be assumed to be an exponent of the Mood head below SAP (Lee 2015, Whitman 1990).

One of the most important advantages of the present analysis is that it accounts for the unembeddability of -(a)la/-nela/-kela/-lyem that reside on SAP. Such unembeddability follows in a straightforward manner from Ross’s (1970) analysis of “implicit performatives,” which states:

(13) Every declarative sentence has one and only one performative sentence as its highest clause (Ross 1970:252, emphasis mine).

Of particular interest is that the implicit performative structure, equivalent to SAP under the present analysis, only occurs in the “highest” or root clause; that is, it is unembeddable. That being the case, the Speech Act projection is the only possible locus for the main imperative morphology. Note also that from the fact that SAP is only available in root clauses, it follows that direct exponents of that head, such as -(a)la/-nela/-kela/-lyem, cannot appear in embedded clauses.

To sum up, Korean employs two different types of imperative morphology: -(a)la/-nela/-kela/-lyem for root imperatives and -(u)la for embedded imperatives. Only the main imperative particles have directive illocutionary force, and consequently, they comprise a direct morphological realization of the Speech Act head that is only present in a root clause. The embedded imperative particle has no illocutionary force, and only labels the clause-type of the clause in which it appears; hence, it resides below performative SAP.

An anonymous reviewer asks about the propositive marker -ca. A full discussion of it is beyond the scope of the article. I conjecture a possible extension of the current view to propositives. Just like imperatives, there are two different types of propositives in the language. First of all, -psita “let’s[formal],” -sey “let’s[familiar],” and -cakkwuna “let’s[intimate]” never occur in embedded clauses. That being the case, they are associated with SAP. However, -ca seems to be problematic since it can occur both in main and in embedded clauses. My assumption is that the propositive particle in question accidentally has the same form for both cases. That is, there are two different -ca’s: One is an SA head and the other is Mood only. I leave this for future research.


Han, Chung-hye. 1998. The structure and interpretation of imperatives: mood and force in universal grammar. Philadelphia, PA: University of Pennsylvania Dissertation. Search in Google Scholar

Haegeman, Liliane. 2014. “West Flemish verb-based discourse markers and the articulation of the speech act layer.” Studia Linguistica 68: 116–39. Search in Google Scholar

Haegeman, Liliane and Virginia Hill. 2010. The syntacticization of discourse. Ms., Ghent University and University of New Brunswick-SJ. (Online access: October 8, 2013). Search in Google Scholar

Han, Chung-hye and Chungmin Lee. 2007. “On negative imperatives in Korean.” Linguistic Inquiry 38: 373–95. Search in Google Scholar

Im, Hong-bin. 1983. “Kwukeuy ‘celtaymwun’ey tayhaye [On ‘absolute sentences’ in Korean].” The Chin-Tan Hakpo 56: 97–136. Search in Google Scholar

Lee, Hyeran. 2012. “Embedded Imperatives in Korean and their implication to parameterization.” Studies in Generative Grammar 22: 595–624. Search in Google Scholar

Lee, Hyo Sang. 2015. “Modality.” In The handbook of Korean linguistics, ed. Lucien Brown and Jaehoon Yeon, 249–68. Malden, MA: John Wiley & Sons. Search in Google Scholar

Lee, Hyeran and Jong Un Park. 2014. “On the syntactic condition for embedded imperatives.” Studies in Generative Grammar 24: 429–53. Search in Google Scholar

Pak, Miok. 2008. “Types of clause and sentence end particles in Korean.” Korean Linguistics 14: 113–56. Search in Google Scholar

Platzack, Christer. 2007. “Embedded imperatives.” In Imperative Clauses in Generative Grammar, ed. Wim van der Wurff, 181–203. Amsterdam and Philadelphia: John Benjamins. Search in Google Scholar

Portner, Paul. 2007. “Imperatives and modals.” Natural Language Semantics 15: 351–83. Search in Google Scholar

Portner, Paul, Miok Pak, and Raffaella Zanuttini. 2018. The speaker-addressee relation at the syntax-semantics interface. Ms. To appear in Language. Search in Google Scholar

Rivero, María L. and Arhonto Terzi. 1995. “Imperatives, V-movement and logical mood.” Journal of Linguistics 31: 301–22. Search in Google Scholar

Rizzi, Luigi. 1997. “The fine structure of the Left Periphery.” In Liliane Haegeman, Elements of Grammar, 281–337. Dordrecht: Kluwer. Search in Google Scholar

Ross, John. 1970. “On declarative sentences.” In Readings in English transformational grammar, ed. Roderick Jacobs and Peter Rosenbaum, 222–77. Boston: Ginn. Search in Google Scholar

Sadock, Jerald M. and Arnold Zwicky. 1985. “Speech act distinctions in syntax.” In Language Typology and Syntactic Description, ed. Tim Shopen, 155–96. Cambridge: Cambridge University Press. Search in Google Scholar

Se, Taylyong. 1985. “Kwukeuy myenglyengmwuney tayhaye [On Korean imperatives].” Journal of Korean Linguistics 14: 437–61. Search in Google Scholar

Speas, Margaret. 2004. “Evidentiality, logophoricity and the syntactic representation of pragmatic features.” Lingua 114: 255–77. Search in Google Scholar

Speas, Peggy and Carol Tenny. 2003. “Configurational properties of point of view roles.” In Asymmetry in grammar, ed. Anne-Marie Di Sciullo, 315–45. Amsterdam: John Benjamins Publishing. Search in Google Scholar

Tenny, Carol. 2006. “Evidentiality, experiencers, and the syntax of sentience in Japanese.” Journal of East Asian Linguistics 15: 245–88. Search in Google Scholar

Whitman, John. 1990. “Topic, modality, and IP structure.” In Harvard Studies in Korean Linguistics III, ed. Susumu Kuno, Ik-Hwan Lee, John Whitman, Sung-Yun Bak, Young-Se Kang, and Young-joo, Kim, 341–56. Seoul: Hanshin Publishing Company. Search in Google Scholar