You are on page 1of 2

Focus Particles, Scope, and Reconstruction in Korean

Youngjoo Lee, Massachusetts Institute of Technology


Synopsis This paper presents an account of the distribution and interpretation of Korean focus particle man
only. I argue that the focus particle is an agreement morpheme that indicates the presence of a (null) Focus
head, and that this null head carries quantificational/exhaustive meaning that we usually associate with English
only. In order to detect the position of the Focus head, I extend the Mirror Principle (Baker 1985) to nominal
affixes, and claim that the relative order of the focus particle and the case marker/postposition reflects the
hierarchy of functional heads that agree with the focus particle and the case marker respectively. I show that
the proposed analysis accounts for some puzzling facts about the particles scope along with long-standing
observations regarding its distribution.
Two Asymmetries: distribution and scope There is a well-known distributional asymmetry between
grammatical case markers and postpositions (semantic case markers) regarding the position of man. The focus
particle man precedes grammatical case markers, but follows postpositions, as shown in (1a-b). (cf. Cho 2000).
(1) a. Mary-ka John-man-ul
saranghanta.
Mary-N John-only-Acc love
Mary loves only John.

b. Mary-ka Boston-eyse-man salassta.


Mary-N
Boston-in-only
lived
Mary has lived only in Boston.

This asymmetry is correlated with another asymmetry, namely scope asymmetry. An overt Acc. marker
triggers obligatory reconstruction when the man-phrase is scrambled, whereas postpositions do not have such
an effect. For example, in (2a) an every-only sequence only allows a surface scope reading. Scrambling of the
overtly case-marked man-phrase in (2b) does not change this scope relation, which is puzzling given the
assumptions that DP-man is a QP of type <et, t> (Choi 1998) and that scrambled QPs can be interpreted in
their surface position and allow ambiguity (Hoji 1985, Anh 1990). In order for the wide scope reading of only
to be available, the case marker must be absent, as shown in (2c). This pattern is in contrast with that of PPs in
(3). When a PP is in its base position, as in (3a), only the surface scope reading is available. When the PP is in
the S-initial position as in (3b), on the other hand, the sentence becomes ambiguous, which means that the PP
can be interpreted in its surface position. These asymmetries (the contrast between (1a) and (1b), and the
contrast between (2b) and (3b)) require an explanation.

(2) a. motun-saram-i
John-man-ul
saranghanta.
every-person-Nom
John-only-Acc
love
(i) Everyone loves John and no one else.
(every > only)
(ii) *John is the only one whom everyone loves.
(*only > every)
[motun-saram-i
ti
saranghanta].
b. John-man-uli
John-only-Acc
every-person-Nom
love
(i) Everyone loves John and no one else.
(every > only)
(ii) *John is the only one whom everyone loves.
(*only > every)
c. John-mani
[motun-saram-i
ti
saranghanta].
John-only
every-person-Nom
love
(i) Everyone loves John and no one else.
(every > only)
(ii) John is the only one whom everyone loves.
(only > every)
(3) a. motun-saram-i
John-hako-man
akswuhayssta.
every-person-Nom
John-with-only
shook_hands
(i) Everyone shook hands only with John.
(every > only)
(ii) *John is the only one with whom everyone shook hands.
(*only > every)
[motun-saram-i
ti
akswuhayssta].
b. John-hako-mani
John-with-only
every-person-Nom
shook_hands
(i) Everyone shook hands only with John.
(every > only)
(ii) John is the only one with whom everyone shook hands.
(only > every)
Proposal and Analysis I propose that the focus particle man is an agreement morpheme that indicates the
presence of a Focus head, which has the lexical entry given in (4). The Focus head takes two arguments, and
asserts that the second argument (an individual of type e) is the only element that satisfies the first argument (a

predicate of type <e, t>). The focused phrase (man-phrase) either undergoes focus movement to the specifier of
FocP, or is base-generated in the [Spec, FocP], where interpretation takes place. In order to detect the position
of the null FocP, I claim that Bakers Mirror Principle applies to nominal affixes: the linear order of nominal
affixes reflects the hierarchy of the corresponding functional heads. Since postpositions precede man, and man
is followed by case markers, the FocP is above VP (where PP is licensed), and below vP/TP (where Acc/Nom
cases are checked/assigned). This accounts for the distributional asymmetry between case markers and
postpositions. The higher FocP does not involve a movement structure, as will be argued below, and this rules
out the unattested form *DP-Case-man.

(4) p<e,t>.xe.p(x) =1 & zeALT(x): p(z) = 1 z = x


The scope asymmetry also follows from the null FocP analysis. In (5a), which represents (2a), everyone is in
[Spec, TP] and John in [Spec, FocP], which is lower than vP, as evidenced by the order of the particle and
Acc. marker. Since the FocP is lower than TP (Cf. Belletti 2002; Jayaseelan 2001 for low FocP), the scope
relation (every > only) follows from this configuration. In (5b), the man-phrase underwent scrambling to Sinitial position (cf. (2b)). Yet, scrambling does not change the position of the FocP. FocP is still below TP.
Therefore, scrambling does not affect scope relation, and the same interpretation as that of (5a) is
compositionally derived. The ambiguous sentence in (2c) has two derivations. One reading (every > only) has
the same structure as (2b), i.e. the structure in (5b). The other reading (only > every) has the high FocP (above
TP), and the man-phrase is base-generated in the spec of the higher FocP, and co-indexed with a pro within
TP, as shown in (5c). This structure is parallel to that of topic construction (Saito 1985), and is justified by
several tests. One test is the availability of a resumptive pronoun, which is not allowed in case of scrambling.
The trace in (5b) cannot be replaced by an overt pronoun, as shown in (6a), whereas the pro in (5c) can, as
shown in (6b). Once the pro is replaced by a pronoun, only the wide scope reading (only > every) is available.
This shows that the empty category in the wide scope reading is a pro, not a trace.

(5) a. [TP everyone x [FOCP John [y [VP x loves y ] FOC]] T]


Focus Mvt.

(Low FocP)

b. [TP John v [TP everyone x [FOCP v [y [VP x loves y] FOC]] T]]


Scrambling

Focus Mvt.

c. [FOCP Johni [TP everyone x [VP x loves proi ] T] FOC]


(6) a. *John-man-uli
[motun-saram-i
ku-luli
John-only-Acc every-person-Nom
b. John-mani
[motun-saram-i
ku-luli
John-only
every-person-Nom
(i) *Everyone loves John and no one else.
(ii) John is the only one whom everyone loves.

(High FocP)

saranghanta].
love
saranghanta].
love
(*every > only)
(only > every)

On the other hand, the overt presence of postpositions does not fix the position of the FocP. Under the
assumption that postpositions are assigned/checked within VP, the order of hako-man with-only only shows
that the FocP is higher than VP. Since the order of suffixes is compatible with both positions of FocP (High
FocP above TP and Low FocP below TP), the ambiguity is expected. The ambiguous sentence in (3b) has the
following two structures. Unlike DP-man case, PP-man does not allow a resumptive pronoun in any case.

(7) a. [TP John [z [TP everyone [x [FOCP z [y [VP x y shake_hands]] FOC] T]]]]
Scrambling
b. [FOCP John [y [TP everyone x [VP x

(Low FocP)

Focus Mvt.
y shake_hands] T ]] FOC]

(High FocP)

Focus Movement
To sum up, I show that the null FocP analysis is superior to the alternative approach that identifies the
particle man as a quantificational element. The proposed analysis accounts for the long-standing
distributional properties of the particle along with the hitherto unnoticed scope facts and the peculiar role
of overt case marking.

You might also like