In our account,
shì is analyzed as a focus-domain marker, rather than a focus marker (contra
Teng 1979;
Huang 1982;
Xu 2004); namely, only constituents that fall in the c-command domain of
shì will have a chance to have a contrastive focus reading if prosodically prominent. To account for the so-called Adjacency Condition in
shì … (de) patterns, we argue for two types of syntactic structures, one involving a Focus head in the left periphery, the other not. As schematized in (36), (36a) and (36b) represent two specific structural configurations that exploit the focus head in the left periphery, whereas (36c) does not involve the focus head. In the structure (36a), since only XP falls in the c-command domain of
shì, it can have a focus reading. YP cannot be focused, because it does not fall in the c-command of
shì. As argued in
Section 4.1, the structure in (36a) is concretized by
Pan’s (
2017,
2019a) FocP, in which it is the null focus head that assigns a focus reading. [
Shì XP] is positioned inside the spec of FocP. This structure thus gives rise to the apparent Adjacency Effects.
10 (36b) resembles (36a) in that FocP is implied in both structures; however, these two structures differ in one important aspect: in (36b), both XP and YP fall under the focus domain of
shì. Since both XP and YP fall into the c-command domain of
shì, both can receive a contrastive focus reading if assigned prosodic prominence, giving rise to the lack of Adjacency Effects. We propose that it is FocP that assigns a focus reading to XP and YP, but it is prosody that brings out a contrastive reading. Unlike (36a/b), the structure in (36c) does not involve a FocP in the left periphery. In the structure (36c),
shì takes the rest of the sentence as complement, including XP, YP and ZP. Since XP, YP and ZP all fall in the c-command domain of
shì, all of them can receive a focus reading if assigned prosodic prominence, giving rise to the lack of Adjacency Effects. In
Section 4.2, we will argue that
shì is the matrix verb and that the contrastive focus reading is assigned through prosodic prominence, such as stress assignment, to any material in the c-command domain of
shì. As a result, there is no Adjacency Condition in our analysis.
(36) | a. | [FocP [shì | XP], [Foc‘ [… YP …]] |
| b. | [FocP [shì | [ XP … YP …]], [Foc‘ [… ZP …]] |
| c. | [TP [vP shì | [ XP … YP …ZP]]] |
4.1. Structure (36a): FocP
As shown in
Section 3.1, there is no genuine Adjacency Condition in all the
shì …(de) patterns. The adjacency-like behavior is in fact due to a limited scope for focus assignment. For this situation, we adopt the structure of (36a). It concerns ‘bare initial-
shì +
ex-situ object’, ‘initial-
shì … final-
de’ and ‘initial-
shì … non-final-
de’. We extend
Pan’s (
2017,
2019a) analysis of
ex-situ cleft focus structure to this group of patterns.
As argued in
Pan (
2019a), this object moves and remerges with
shì, resulting as part of the specifier of FocP.
11 The null Foc head provides its Spec with a focus reading. We clarify that
shì never moves; instead, it is always base-generated. More importantly,
shì alone can never occupy Spec, FocP; it is the entire constituent [shì + XP] that occupies Spec, FocP. Given the requirement imposed by the domain marker
shì, the focus must fall in its c-command domain, which only includes the
ex-situ object
niúròu ‘beef’.
(37) | | Initial-shì + ex-situ object | | | | |
| a. | [FocP [TP2 shì [niúròu]j, [Foc’ [Focº ∅] [TP1 | tā | bù | xǐhuān | chī tj ]]]]. |
| | be beef | 3sg | neg | like | eat |
| | ‘It is beef that he does not like eating.’ |
| b. | |
We extend this analysis to the two types of “initial-
shì …
de” patterns. To account for the ‘’initial-
shì … final-
de’’ pattern, we assume that
de is in C (
Cheng 2008;
Hole 2011). Recently,
Pan and Xu (
2022) argued that the final
de is in an Assertion head, which is higher than the S.Asp head and lower than the iForce head (see detailed discussion below). The null Foc head takes as complement a lower CP headed by
de.
12 The subject
tā ‘he’ is dislocated and remerges with
shì. The result of this merging is part of the Spec of FocP. The subject
tā ‘he’ receives a focus reading from the Foc head. Since
tā ‘he’ is in the c-command domain of
shì, the focus assignment is rendered possible. The rest of the clause is not in the c-command domain of
shì; therefore, no focus can be assigned to it. Consequently, the so-called Adjacency Condition is not a genuine condition that regulates the distribution of foci. The presence of this ‘restriction’ as in (38) is due to the fact that the c-command domain of
shì is inside the Spec of FocP.
13(38) | | Initial-shì … final-de |
| a. | [FocP [TP2 shì | tāj ] [Foc’ [Focº ∅] [TP1 tj | zài | Wēinísī | xué-le | hànyǔ | de]]]. |
| | be | 3sg | at | Venice | study-perf | Chinese de | |
| | ‘It is he that studied Chinese in Venice.’ |
| b. | |
Regarding the ‘’initial-
shì … non-final
de’’ pattern, we follow
Paul and Whitman (
2008) in assuming that
de is in Asp head. Recall that the structure between
shì and non-final verb-adjacent
de excludes any material above
vP. To account for the Adjacency Condition, we apply the FocP analysis to the case with non-final
de. As shown in (39), the AspP headed by
de is part of the complement of Foc head. Again, the merging of
shì with the dislocated subject
tā ‘he’ is within Spec of FocP. The domain marker
shì limits the focus reading, which is assigned by the Foc head, to its c-command domain. To explain the verb-adjacent position of
de, either
de undergoes some morphological movement, or the direct object moves out of AspP, followed by the remnant movement of AspP.
(39) | | Initial-shì … non-final de | | | | | | |
| a. | [FocP [TP2 shì | tāj ] [Foc’ [Focº ∅] [ [AspP tj | zài | Wēinísī | xué | de | hànyǔ]]]]. |
| | be | 3sg | at | Venice | study | de | Chinese |
| | ‘It is he that studied Chinese in Venice.’ |
| b. | |
4.2. Structure (36b): FocP
Like the structure (36a, repeated below) shown in the previous section, (36b) involves a Focus projection in the left periphery, in which the c-command domain of
shì is inside the Spec of FocP. Recall our claim that only the constituents that fall in the c-command domain of
shì can be assigned a focus reading. In addition, the structure in (36b) concerns the same group of initial-
shì …(de) patterns, namely, ‘bare initial-
shì +
ex-situ object’, ‘initial-
shì … final-
de’ and ‘initial-
shì … non-final-
de’. However, unlike the type of (36a) presented in
Section 4.1, (36b) seems to lack the so-called Adjacency Effects, that is, both XP and YP, falling in the c-command domain of
shì, can receive a contrastive focus reading if assigned prosodic prominence (see
Section 3.1 for a detailed demonstration).
(36) | a. | [FocP [shì | XP], [Foc‘ [… YP …]] |
| b. | [FocP [shì | [ XP … YP …]], [Foc‘ [… ZP …]] |
We use an example of the ‘bare initial-
shì +
ex-situ object’ to illustrate this. As shown in (40), both the adjacent
nǐ-de ‘your’ and the non-adjacent NP
tàidù ‘attitude’ of an
ex-situ object can be contrastively focused. We argue that it is FocP that assigns a focus reading, but it is the prosody that brings out a contrastive meaning. We use (40c) to illustrate the domain of
shì. Again, our claim is that any material in the c-command domain of
shì can be targeted by a focus reading, be the whole complement of
shì ‘be’, that is,
nǐ-de tàidù ‘your attitude’, or part of it, that is,
nǐ-de ‘your’ or
tàidù ‘attitude’. While the null Foc head is responsible for the focus reading, the prosodic prominence brings about a contrastive meaning.
(40) | | Initial-shì + ex-situ object |
| a. | [FocP [TP2 shì | [nǐ-de tàidù]j, | [Foc’ [Focº ∅] [TP1 gōngsī-de | lǎobǎn | bù | xīnshǎng | tj ]]]. |
| | be | you-de attitude | company-de | boss | neg | appreciate | |
| | ‘It is your ATTITUDE that the boss of the company does not appreciate.’ |
| b. | [FocP [TP2 shì | [nǐ-de tàidù]j, | [Foc’ [Focº ∅] [TP1 gōngsī-de | lǎobǎn | bù | xīnshǎng | tj ]]]. |
| | be | you-de attitude | company-de | boss | neg | appreciate | |
| | ‘It is YOUR attitude that the boss of the company does not appreciate.’ |
| c. | |
4.3. Structure (36c): No FocP
In contrast with (36a) and (36b), repeated below, the structure of (36c) shows a ‘freer’ focus assignment to any element to the right of
shì, that is, XP, YP and ZP, because, as argued above, any material that falls in the c-command domain of
shì can be assigned a focus reading. Furthermore, we argue that both the focus reading and the contrastive reading are assigned through prosodic prominence such as stress assignment in the case of (36c). The structure of (36c) concerns all the medial
-shì patterns and the ‘initial
-shì + TP’ pattern. We follow
Pan (
2019a) in assuming that the
shì in these patterns takes as complement the entire string to its right. Unlike (36a) and (36b), there is no FocP in (36c).
(36) | a. | [FocP [shì | XP], [Foc‘ [… YP …]] |
| b. | [FocP [shì | [ XP … YP …]], [Foc‘ [… ZP …]] |
| c. | [TP [vP shì | [ XP … YP …ZP]]] |
Pan’s (
2019a) analysis was originally made to explain the sentence as in (41), in which the negator
bù scopes over the sentence-final aspect particle
le, which has a change-of-state meaning. He argued that
shì is the matrix verb and takes a clause CP as its complement. Here, the split CP projects until S.AspP (the projection of sentence-final aspectual particles), headed by
le in S.Asp head on the right edge. Inside the complement of
shì, there is a null subject
pro controlled by
wǒ ‘I’ (cf.
Huang’s (
1989) Generalized Control Rule). In addition,
wǒ ‘I’ is in the matrix topic position (see also
Huang 1988 and
Cheng 2008 for topic analysis of pre-copula elements in this context).
(41) | TopP > Neg > shì ‘be’ > le > TP-miss home |
| [TopP wǒj | [TP1 bù | shì | [S.AspP [TP2 proj. xiǎng jiā] | le]]]. |
| I | neg | be | miss home | le |
| ‘As for me, it is not the case that I start missing home.’ |
| (Pan 2019a, p. 25, ex. (28)) |
Example (41) is in fact a bare medial-
shì sentence. We apply Pan’s analysis directly to another example of the same pattern as in (42). As shown below, the entire string to the right of
shì is in the c-command domain of
shì. Given our prosodic strategy, any item in that domain can be associated with focus by means of prosodic prominence and obtains a contrastive meaning. As a result, we observe the lack of the so-called Adjacency Effects.
14(42) | | Bare medial-shì |
| a. | [TopP | tāj | [TP1 [vP shì | [AspP proj zài Wēinísī | xué-le | hànyǔ]]]]. |
| | | 3sg | be | at Venice | study-perf | Chinese |
| | ‘It is in Venice that he studied Chinese.’ |
| b. | |
For the ‘’medial-
shì … final-
de’’ pattern, the particle
de is merged in C, which in turn is taken as the complement of
shì. The entire CP headed by
de, which falls into the domain delimited by
shì, can be targeted by prosodic prominence, giving rise to the contrastive focus reading.
(43) | | Medial-shì … final-de |
| a. | [TopP | tāj [TP1 shì | [CP [AspP proj zài Wēinísī | xué-le | hànyǔ | [Cº de ] ]]]]. |
| | | 3sg be | at Venice | study-perf | Chinese | de |
| | ‘It is in Venice that he studied Chinese.’ |
| b. | |
As shown in (44), the ‘’medial-
shì … non-final
de’’ pattern has a similar structure in which
shì is the matrix verb. Recall the restriction that only the
vP structure can occur in this pattern. The particle
de is merged in Asp, taking
vP as its complement (see
Paul and Whitman 2008). The entire AspP is in the c-command domain of
shì, to which the focus can be assigned by stress.
15(44) | | Medial-shì … non-final-de |
| a. | [TopP | tā [TP1 shì | [AspP [vP proj zài Wēinísī | xué | de | hànyǔ]]]]. |
| | | 3sg be | at Venice | study | de | Chinese |
| | ‘It is in Venice that he studied Chinese.’ |
| b. | |
At last, we tackle the case in which the initial-
shì precedes a TP. Recall that in this pattern, the focus can be assigned to any element c-commanded by
shì. As shown in (45),
shì takes the entire string to its right as complement. As a result, the entire TP, which is the c-command domain of
shì, is legitimate to receive a focus reading via prosodic prominence (see
Section 3.1 for examples).
(45) | Bare initial-shì + TP | | | | |
| [shì | [TP | táifēng | xíjuǎn-le | Àomén]]. |
| be | | typhoon | sweep-perf | Macau |
‘It is the typhoon that swept through Macau.’ Or ‘It is Macau that was swept through by the typhoon.’ |
In all the
shì … (de) patterns,
shì is treated as a verb (
Paul 2015;
Pan 2017). In addition, to account for the exhaustivity effects, we follow
Pan (
2019a) in assuming that
shì contributes to the exhaustive meaning, supported by a crucial diachronic argument. We have argued that
shì is a domain marker rather than a focus marker. Its occurrence signals the domain to which the focus reading can be assigned. We have shown that
shì can be merged either in a position inside the Spec of FocP or in the position of a matrix verb. Consequently, the c-command domain of
shì varies, giving rise to the presence or absence of the so-called Adjacency Effects respectively.
Let us turn to root/non-root status of the sentence-final
de. As argued by
Paul and Whitman (
2008), the sentence-final
de in propositional assertion (that is, their ‘NP shì V O de’ pattern) is a non-root C head, which is associated with a [-finite] feature. Since the complement of the non-root C is non-finite, the subject must undergo A-movement to the matrix subject position.
An alternative view was proposed by
Pan and Xu (
2022). These authors used (47) to show that the sentence final
de is structurally higher than the sentence-final aspectual particle
le (in S.Asp head), and is lower than the
yes-no question particle
ma (in iForce head). They argue that
de is in an Assertion head. The overall hierarchy they proposed is: AttP > SQP > iForceP > AssertionP > S.AspP > TP.
(47) | a. | [iForce ma] > [Assertion de] > | [S.Asp le]; ’>’ | means ‘is structurally higher than’ |
| | [iForceP [AssertionP [S.AspP [TP | zuótiān | shì xià-guò yǔ] | le ] de ] ma]. |
| | | yesterday | be fall-exp rain | le de ma |
| | ‘Is it the case that it rained yesterday?’ |
| b. | *[iForce ma] > [S.Asp le] > [Assertion de] | |
| | *[iForceP [AssertionP [S.AspP [TP | zuótiān | shì xià-guò yǔ] | de] le] ma] ? |
| | | yesterday | be fall-exp rain | de le ma |
| | Pan and Xu (2022, p. 111, ex. (53)) |
Very importantly, this
de in Pan and Xu’s framework is exactly the
de in propositional assertion in the sense of
Paul and Whitman (
2008); however, different from Paul and Whitman’s view, we observe that this
de can perfectly be used in a root sentence, thus it does not show non-root property at all. More examples can be seen in (48), where
de can only have a root construal.
(48) | a. | [AssertionP [TP | wǒ | chū | mén | qián | běnlái | kěyǐ | xǐ | gè | -zǎo] | de]. |
| | | 1sg | go.out | door | before | originally | can | wash | clf | shower | de |
| | ‘It is indeed the case that I could have taken a shower before going out.’ |
| b. | [AssertionP [S.AspP [TP | wǒ | běnlái | hái | xiǎng | pāi | zhāng | -zhàopiān] | láizhe] | de]. |
| | | 1sg | originally | still | want | take | clf | picture | laizhe | de |
| | ‘It is indeed the case that I intended to take a picture.’ |
The original argument in support of the non-root status of
de is that
de must cooccur with the copular
shì and
shì takes the clause involving
de as its complement. Importantly,
shì can be omitted. In fact, this argument is problematic. Take (49) with the
yes-no particle
ma as an example. The copular
shì can also appear freely in (49) and its presence is never obligatory. However, no one would use (49) to claim that the
ma-sentence involves an obligatory presence of shì and that
shì is phonetically omitted. Therefore, the same argument does not hold for de in
shì…(de) constructions.
(49) | tā | (shì) | míngtiān | qù | Bālí | ma? |
| 3sg | be | tomorrow | go | Paris | ma |
| ‘(Is it) tomorrow (that) he will go to Paris?’ |
From this perspective, we have no reason to claim that de is a non-root only SFP. Therefore, it is more reasonable to treat sentences involving shì…de and those only involving the final de without shì as two different structures.
A reviewer raised a question concerning what mechanism governs the association between the intonational peak and the Focus constituent. Here are our clarifications: first, there is not necessarily a cause–consequence type of association between prosody and focus; rather, they are independent marking/interpretation strategies. For Chinese, contrastivity should be disassociated from topic and focus. A topic can be, but is not necessarily always, contrastive; the same applies to focus. A focus can be information focus or contrastive focus. According to
Erteschik-Shir (
2007), contrastivity should be treated as an independent component, which is differentiated from focus. Adopting this view,
Pan (
2015) further demonstrates that in Chinese, contrastiveness can be marked either syntactically or prosodically, and prosodic marking can only be activated as a last resort when the syntactic marking is not available. Second, in the study of interpretive ambiguities of wh-phrases
in situ,
Pan (
2019b) showed that different combinations of stress with intonation on sentences can disambiguate the readings. He argued that those prosodic forms that have semantic effects at LF can be analyzed as phonological features in the feature bundles associated with a given lexical item in the Lexical Array. Each specific prosodic pattern (i.e., stress combined with sentence intonational pattern) corresponds to one and only one semantic interpretation at the C–I interface. Therefore, one-to-one mapping between prosody and semantics is ensured, even before the relevant derivation starts. As a result, the superficially observed ambiguity is only an illusion. Adopting this view, we argue that the focused element associated with prosodic prominence in
shì sentences can have a focus feature as part of its feature bundles. See also
Bianchi et al. (
2015)
16 for a similar proposal in their analysis of Italian corrective and mirative focus.