Sentence fragments or stripped expressions such as fragment answers (FAs), illustrated in (1a), have evoked several questions. Among these questions, two crucial ones are (a) whether or not stripped expressions involve underlying full sentences; and (b) if stripped expressions are derived from full-fledged sentences followed by deletion, whether or not the elided part forms a syntactic constituent (Morgan 1973, 1989, Yanofsky 1978, Chung, Ladusaw, and McCloskey 1995, Merchant 2004, and references therein). Assimilating FAs to sluicing, Merchant (2004) provides an analysis that addresses both of these issues. Merchant posits that FAs involve deletion of a syntactic constituent from underlying full sentences. Specifically, the fragment moves to a clause-peripheral position followed by TP-ellipsis, as schematized in (1b).
In this squib, examining FAs containing the particle -yo in Korean, illustrated in (2), I provide new evidence in support of Merchant’s PF deletion theory of ellipsis. The analysis demonstrates that there is a remnant particle on the left periphery of the clause that has survived the deletion process, such that it becomes adjacent to the fronted fragment, yielding the surface “fragment-yo.”1
Kim-i nwukwu-rul mannass-ni?
Kim-NOM who-ACC met-Q2
‘Who did Kim meet?’
As a starting point for the discussion, let us consider some characteristics of -yo. This particle usually follows the verb at the end of a sentence, conveying politeness toward the addressee.
Kim-i ecey kongwen-eyse Lee-rul mannasse-yo.
Kim-NOM yesterday park-at Lee-ACC met.C-YO
‘Kim met Lee in the park yesterday.’
And the particle can potentially be ubiquitous in a sentence.
Kim-i-yo ecey-yo kongwen-eyse-yo Lee-rul-yo
Kim-NOM-YO yesterday-YO park-at-YO Lee-ACC-YO
‘Kim met Lee in the park yesterday.’
These ubiquitous/sentence-medial instances of -yo, however, cannot appear without the sentence-final -yo.
(5) *Kim-i-yo ecey-yo kongwen-eyse-yo Lee-rul-yo mannasse-Ø.
Interestingly enough, there are certain categories and constructions that resist sentence-medial -yo. As Lee and Park (1991:371) observe, this is the case with some adverbs of manner and locative adverbs.
Ikes-ul tangcang(*-yo) chelihase-yo.
this-ACC immediately(*-YO) handle-YO
‘Have this done immediately.’
Kapang-i yeki(*-yo) isse-yo.
bag-NOM here(*-YO) exist.C-YO
‘Here is the/a bag.’
It is impossible for -yo to follow tangcang ‘immediately’ and yeki ‘here’ in regular (i.e., nonelliptical) sentences.3 However, -yo-attachment becomes available for the manner and locative adverbs in FAs.
Ence-kkaci ikes-ul chelihayyahacyo?
when-by this-ACC handle.must.Q
‘By when must I have this done?’
Kapang-i eti iss-e?
bag-NOM where exist-Q
‘Where is the bag?’
Yay-ka [ku namca-ka kitali-ten(*-yo)] sonye-yey-yo.
this-NOM that man-NOM wait-C(*-YO) girl-be.C-YO
‘This is the girl who the man was waiting for.’
Etten salam-i wasstako? (“reclamatory”)
which person-NOM came.C
‘Which person (did you say) has come?’
Ku namca-ka kitali-ten-yo!
that man-NOM wait-C-YO
‘(The one) that the man was waiting for!’
Last, but not least, the impossibility of -yo-attachment involves quantifier floating (Lee and Park 1991:377). Attaching -yo to classifiers leads to ungrammaticality in nonelliptical contexts, as presented in (9b–c).
Kim-i Lee-eykey cangmi-rul-yo paek songi-rul
Kim-NOM Lee-to rose-ACC-YO hundred CL-ACC
‘Kim bought a bunch of one hundred roses for Lee.’
Kim-i Lee-eykey cangmi-rul paek songi-rul(*-yo) sacwuesse- yo.
Kim-i Lee-eykey cangmi-rul-yo paek songi-rul(*-yo) sacwuesse-yo.
It is important to note, however, that -yo can occur on classifiers in FAs.
Kim-i Lee-eykey mwues-ul sacwuess-ni?
Kim-NOM Lee-to what-ACC bought-Q
‘What did Kim buy for Lee?’
*Cangmi-rul-yo paek songi.
rose-ACC-YO hundred CL
‘A bunch of one hundred roses.’
Cangmi-rul paek songi-yo.
Cangmi-rul-yo paek songi-yo.
The sentences in (9) contrast sharply with the FAs in (10) because the ill-formedness of attaching -yo to the classifiers in the nonelliptical sentences in (9b–c) substantially improves in the FAs in (10b–c). On the other hand, the well-formed nonelliptical sentence in (9a) that lacks -yo on the classifier becomes ungrammatical in the FA in (10a).
In a nutshell, when nonelliptical contexts that do not host -yo become elliptical, -yo-attachment becomes grammatical. Merchant’s (2004) PF deletion analysis of ellipsis straightforwardly accounts for this phenomenon.4 As illustrated in (11), the fragment undergoes fronting to clause-peripheral position, followed by deletion of TP; -yo, occupying a position higher than the elided portion, survives the deletion process, resulting in an illusory surface “fragment-yo.” From this perspective, -yo in FAs is the left-peripheral remnant of ellipsis. That is to say, it is a stranded sentence-final particle.5
In summary, we have seen that Korean allows -yo in FAs, but not in nonelliptical contexts. This fact is direct morphological evidence in favor of the PF deletion theory of ellipsis. From this perspective, the -yo in FAs is best understood as an ellipsis remnant on the clausal left periphery; the surface “fragment-yo” is merely illusory.6
I would like to thank Jason Merchant and two anonymous LI reviewers for helpful comments and suggestions. All remaining errors are mine.
1 While the particle under discussion appears on the right periphery in Korean, I use the term left periphery to refer to the CP layer in the sense of Rizzi 1997.
2 The following abbreviations are used here: ACC = accusative; C = complementizer; CL = classifier; NOM = nominative; PRPS = propositive marker; Q = question marker.
3Key-derived manner adverbs such as sinna-key ‘with gusto’ allow -yo-attachment.
4 An anonymous LI reviewer notes that the data attested here can be consistent with the nonconstituent ellipsis approach as well.
5 In this analysis, the exact position that the sentence-final -yo occupies is of major importance. For current purposes, it suffices to say that the sentence-final particle takes a position at least higher than the ellipsis site, TP. In fact, there is reason to believe that the relevant position for the particle in question may be even higher than the Force position in Rizzi’s (1997) clausal left periphery system. Consider the following sentences:
Kapang-i yeki iss-e-yo.
bag-NOM here exist-C-YO
‘Here is the bag.’
Kapang-i eti iss-ulkka-yo?
bag-NOM where exist-Q-YO
‘Where is the bag?’
The sentence-final -yo in (ia–b) follows the declarative marker -e and the interrogative marker -ulkka, respectively, both of which are clause-type elements in the sense of Cheng 1991 and therefore most likely to head the Force projection in the sense of Rizzi 1997.
Indirect evidence emerges from language acquisition. Children learning Korean make an error of the sort that Korean-speaking adults do not with respect to -yo-attachment. They produce sentences in which -yo follows the propositive marker -ca ‘let’s’.
It is clear that since the propositive marker is a clause-type marker, it occupies a position higher than the T head position, most likely, the Force head position.
But then a question arises: why can no other left-peripheral particle than -yo be stranded by the ellipsis process that produces FAs? I leave this for future research.
6Van Craenenbroeck and Lipták (2008) observe a similar yet different phenomenon found with the Hungarian suffix -e in sluicing. I thank Jason Merchant for pointing this out to me.