The Size of the Mandarin Focal Phrase

國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(iv) The remaining syllables undergo regular foot formation.

In this present study, the pattern which shows that the foot formation starts at the focused element is more common among the informants, whereas the foot formation ends at the focused element occurs less frequently among them. In terms of markedness, the common pattern is unmarked while the less frequent pattern is marked.

3.7 The Size of the Mandarin Focal Phrase

This section examines the size of the Mandarin focal phrase and tries to see whether the Mandarin focal phrase is identical to foot, phonological phrase, or intonational phrase.

This study adopts the phrasing principles proposed by Hsiao (1991, 1995) to examine the size of the focal phrase in Mandarin. The principles are shown below:

(22) Mandarin Phrasing Principles a. Intonational Phrase (IP):

[…x…]^SU, where x = phonological phrase b. Phonological Phrase (Ph):

‧

As (22) dictates, IPs are sense units (SUs) phrased by grouping together Phs, which are marked at the left edge of an Xmax only if the Xmax is branching. Xmax means the maximal projection of certain parts of speech, such as VP, NP, etc.. When an IP consists of two or more syllables, it must be subject to the Sense Unit Condition (SUC)³

Consider Type C first. In this study, it is found that there are unmarked and marked focal patterns in Type C, as shown in (23) and (24), respectively.

. An intonational phrase is built upon a phonological phrase so an intonational phrase could contain several phonological phrases. In other words, a phonological phrase cannot cross the boundary of an intonational phrase.

(23) ‘Dogs bite women.’ (unmarked reading)

3 Selkirk (1984) proposes a sense unit to condition intonational phrasing in English, assuming that two constituents Ci, Cj form a sense unit if (a) or (b) is true of the semantic interpretation of the sense:

a. Ci modifies Cj (a head)

b. Ci is an argument of Cj (a head).

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

In (23), the Ph boundary #S, #VP, and #NP are placed before guo, yao, and nu, respectively because these words are maximal projections and have syntactic branching. Thus, in (23), there are three phonological phrases: guo, yao, and nu ren.

Since the focal phrase is yao nu ren, the size of the focal phrase is proved to be larger than that of the phonological phrase. However, the IP boundary (%) reveals that the focal phrase matches the intonational phrase. The tone sandhi pattern does not cross the IP boundary.

(24) ‘Dogs bite women.’ (marked reading)

focal phrase{guoL yaoL} nuL renLH

#^S[ #^VP[ #^NP[ Ph [ ] % [ ] IP

(LH L) (L LH) tone sandhi pattern

In (24), the focal phrase guo yao is built upon two phonological phrases but is identical to an intonational phrase. In short, the examination shows that when focal phrasing occurs in Type C, the focal phrase is identical to the intonational phrase.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

Following are example of the adverb zhi from Type A.

(25) ‘Small Li only wants to buy umbrellas.’

xiaoL liL {zhiL xiangL maiL yuL sanL}focal phrase

#^S[ #^ADVP[ #^VP[ #^NP[ Ph [ ] % [ ] IP

(LH L) ((LH LH) L) (LH L) tone sandhi pattern

In (25), there are four phonological phrases: xiao li, zhi, xiang mai, and yu san. The focal phrase begins at the focused element zhi and ends at NP yu san. Clearly, the focal phrase is bigger than the phonological phrase. However, since zhi modifies the verb phrase xiang mai and yu san is the internal argument of xiang mai, they together form an intonational phrase. As a result, the focal phrase matches the intonational phrase.

In conclusion, the size of the focal phrase in Mandarin is bigger than the phonological phrase, which seems to support the findings of Cho (1990) and Kanerva (1990) in Korean and Chichewa, respectively. However, though bigger than the

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

phonological phrase, the focal phrase does not belong to an intermediate prosodic hierarchy between phonological phrase and intonational phrase because the focal phrase matches the intonational phrase. In other words, no mismatch between the focal phrase and the intonational phrase could be found in the present data. Thus, I suggest that the focal phrase in Mandarin matches the intonational phrase.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

55

CHAPTER 4

OT ANALYSIS

This chapter analyzes Mandarin focal tone sandhi under the framework of Optimality Theory. Two sets of constraints are proposed in this chapter, the prosodic constraints, which evaluate the prosodic candidates and determine the tonal domain, and the tonal constraints, which evaluate the tonal candidates. It will show that with these two sets of constraints, focal tone sandhi in Mandarin can be accounted for successfully.

4.1 Prosodic Constraints for Mandarin Regular Tone Sandhi

According to the theory proposed by Shih (1986), the best units of prosodic structure in Mandarin are disyllabic feet. The immediate constituents in a sentence are joined into disyllabic feet. In non-focal speech, there is no monosyllabic foot; it joins with a neighboring foot; assuming that each of the syllables should be parsed into feet.

With these basic properties, four prosodic constraints are listed in (1-5).

(1) PARSE-σ: parse syllables into feet. (Prince & Smolensky 1993)

(2) ALIGN-L(IC, Ft): the left edge of every pair of terminal immediate constituents (IC) coincides with the left edge of a foot (Ft).

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(3) ALIGN-R(IC, Ft): the right edge of every pair of terminal immediate constituents (IC) coincides with the right edge of a foot (Ft).

(4) FTMIN: at least two syllables per foot. (Hsiao 2006) (5) FTMAX: no more than two syllables per foot. (Hsiao 2006)

The idea of Foot Binarity comes from the widely attested observation that feet ideally consist of exactly two elements, morae or syllables (cf. Prince 1980, McCarthy &

Prince 1986, Kager 1989, Hayes 1994). In other words, metrical feet can be either syllabic or moraic. The foot types laid out in (6) below is proposed in McCarthy and Prince (1986) and Hayes (1987) to account for Hayes’ (1985) typological findings. L stands for light syllable and H stands for heavy syllable:

(6) Foot Types

Iambic Trochaic Syllabic LH H, LL σσ LL, H

The Prosodic Hierarchy and Foot Binarity, taken together, license the notion of

‘‘Minimal Word’’ (Prince 1980, Broselow 1982, McCarthy and Prince 1986, 1990, 1991a, 1991b). According to the Prosodic Hierarchy, any instance of the category

‧

Prosodic Word (PrWd) must contain at least one foot (Ft). Under Foot Binarity, every foot must be either bimoraic or disyllabic. And, under transitivity, then, a Prosodic Word must contain at least two moras or syllables. Since Mandarin is a quantity-insensitive language, which does not distinguish syllable weight, the minimal word is disyllabic.

In addition, Hsiao (2010) built a corpus of Chinese gu-shi and found that the number of disyllabic nouns is more than that of monosyllabic nouns. Duanmu (2012) used The Lancaster Corpus of Mandarin Chinese and examined the word-length references in Chinese. One of Duanmu’s findings is that for NN compound, the total number of disyllabic compounds accounts for roughly 60 percent in the corpus, which is more than that of trisyllabic and quadrisyllabic compounds. The result of VO compound is similar to that of NN compound. In sum, the typological evidence supports that foot in Chinese is disyllabic.

(7) zhao xiao li ‘Looking for Small Li.’

zhao xiao li

‧

ALIGN-R(IC, Ft), and FTMAX in trisyllabic sentences. The PARSE-σ constraint requires that every syllable must be parsed into a foot; candidates that contains unparsed syllables, such as (7b) and (8b), is ruled out. Candidates (7c) and (8c) violate the FTMIN constraint because the first syllable forms a monosyllabic foot. A monosyllabic foot does not occur in regular reading; it is more marked than a disyllabic foot. FTMIN, which sets a minimum of binarity, must dominate FTMAX to prevent a monosyllabic foot from occurring.

The ALIGN-L(IC, Ft) constraint requires the left edge of the terminal immediate constituent to align with the left edge of the foot. Take zhao xiao li in (7) for example.

The right branching structure contains a verb zhao and an object xiao li. The disyllabic object xiao li is a pair of immediate constituents, so both syllables should be phrased into the same foot. The foot structure will be (σ(σσ)). Therefore, candidate

‧

(7d), ((σσ)σ), violates the A^LIGN-L(IC, Ft) constraint because the left edge of the immediate constituent does not align with the left edge of the foot. On the other hand, The ALIGN-R(IC, Ft) constraint requires the right edge of the terminal immediate constituent to align with the right edge of the foot. Take ju qi shou in (8) for example.

The left branching structure contains a verb ju qi and an object shou. The disyllabic verb ju qi is a pair of immediate constituents, so both syllables should be phrased into

the same foot. The foot structure will be ((σσ)σ). As a consequence, candidate (8a), (σ(σσ)), violates the A^LIGN-R(IC, Ft) constraint because the right edge of the

immediate constituent does not align with the right edge of the foot. Finally, PARSE-σ and FTMIN do not conflict with each other; therefore, these two constraints are at the same ranking. Since unparsed syllables all violate the Parse-σ constraint, I will not list this constraint in order to save space.

(9) gou yao xiao bao ‘Dogs bite Small Bao.’

gou yao xiao bao

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

In (9), the syntactic bracketing of gou yao xiao bao is [σ[σ[σσ]]], but the prosodic phrasing is (σσ)(σσ). Candidates (9b) and (9c) contain a monosyllabic foot so they violate the FTMIN constraint. Thus, candidates (9b) and (9c) are rejected.

Candidate (9d) is ruled out since the prosodic phrasing takes all the syllables as a single domain, which violates the lower ranked FTMAX constraint. It also violates the ALIGN-L(IC, Ft) constraint since the left edge of the pair of immediate constituent xiao

bao does not align with the left edge of foot. Consequently, candidate (9a), the

unmarked reading, is selected as the optimal output successfully and naturally. In short, the present prosodic constraint ranking for Mandarin tone sandhi is listed below.

(10) Prosodic Constraint Ranking for Mandarin Tone Sandhi (in regular speech):

PARSE-σ, F^TMIN >> FTMAX, ALIGN-L(IC, Ft), ALIGN-R(IC, Ft)

4.2 Prosodic Constraints for Mandarin Focal Tone Sandhi

In Chapter 3, I have described how there are three types of focal phrasing patterns: Type A includes the focused auxiliary verbs and adverbs in a simple sentence, Type B the flat structure, and Type C the others. We will propose two alignment constraints, ALIGN-R(F, FP) and ALIGN-L(F, FP), to ensure that the focused element is

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

at the edge of the focal phrase, and two additional alignment constraints, ALIGN-R(FP, Ft) and ALIGN-L(FP, Ft), to align the edge of the focal phrase with the edge of foot.

4.2.1 A

^LIGN

-L(F, FP) & A

^LIGN

-L(FP, Ft)

According to the generalization in Chapter 3, the focused element is at the left edge of the focal phrase when the focal phrasing occurs in Type A and the unmarked reading of Type C. Hence, to capture this fact, an alignment constraint is needed, as defined in (11).

(11) ALIGN-LEFT(focused element, focal phrase): the left edge of the focused element (F) coincides with the left edge of the focal phrase (FP).

In Type A and the unmarked Type C, the candidates which do not align the left edge of the focused element with the left edge of the focal phrase will be ruled out by ALIGN-LEFT(F, FP). ALIGN-L(F, FP) must outrank ALIGN-R(F, FP) to avoid the situation that the right edge of the focused element aligns with the right edge of the focal phrase. Tableau (12) presents Type A for illustration.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(12) xiao li zhi xiang mai yu san ‘Small Li only wants to buy umbrellas.’

xiao li zhi xiang mai yu san

[σσ[σ[σ[σ[σσ]]]] ALIGN-L(F, FP) ALIGN-R(F, FP)

☞ a. {σσ}{σσσσσ}FP *

b. {σσσ}FP{σσσσ} *!

The focused element is the adverb zhi. Candidate (12b) does not align the left edge of focused element zhi with the left edge of the focal phrase so it is ruled out by ALIGN-L(F, FP). Instead, candidate (12a) incurs no violation in ALIGN-L(F, FP) because the left edge of zhi aligns with the left edge of the focal phrase.

In addition, the left edge of the focal phrase should coincide with the left edge of the foot. Thus, another alignmentconstraint is needed, as defined in (13).

(13) ALIGN-LEFT(focal phrase, foot): the left edge of the focal phrase (FP) coincides with the left edge of the foot (Ft).

The ALIGN-L(FP, Ft) requires that the left edge of the focused element must align with the left edge of the focal phrase in Type A and the unmarked Type C. Notably, the alignment of the focal phrase and the foot at the left edge indicates that foot formation starts at the left edge of the focal phrase instead of at the end of foot formation. These

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

two constraints do not conflict with each other because the former is about the formation of focal phrase and the later is about the formation of foot.

The following tableau shows how ALIGN-L(FP, Ft) operates in Type A. The adverb zhi and auxiliary verb gan are shown as instances in (14) and (15), respectively.

Here we discuss only those candidates that incur no violation in ALIGN-L(F, FP). That is, all the candidates have correct form of focal phrase during the evaluation.

(14) xiao li zhi xiang mai yu san ‘Small Li only wants to buy umbrellas.’

xiao li zhi xiang mai yu san

[σσ[σ[σ[σ[σσ]]]] ALIGN-L(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {σσ}{σσσσσ}FP

☞ a. (σσ)((σσ)σ)(σσ) *

b. ((σσ)σ)(σσ)(σσ) *! *

(15) xiao li ken mai yu san ‘Small Li is willing to buy umbrellas.’

xiao li ken mai yu san

[σσ[σ[σ[σσ]]]] A^LIGN-L(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {σσ}{σσσσ}FP

☞ a. (σσ)(σσ)(σσ)

b. ((σσ)σ)(σ(σσ)) *!

As shown above, both candidates (14b) and (15b) violate ALIGN-L(FP, Ft). The left

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

edge of the focal phrase does not align with the left edge of the foot. To be concrete, the left edge of the focal phrase aligns with the right edge of the foot so that foot formation ends at the focused zhi and gan. As a consequence, (14b) and (15b) are ruled out by ALIGN-L(FP, Ft). The candidates (14a) and (15a) are successfully selected as the optimal output, respectively.

(16) gou yao xiao bao ‘Dogs bite Small Bao.’

gou yao xiao bao

[σ[σ[σσ]]] ALIGN-L(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {σ}{σσσ}FP

a. (σσ)(σσ) *!

☞ b. (σ)(σ(σσ)) * *

c. (σ)((σσ)σ) * * *!

d. (σσσσ) *! * *

As for the unmarked reading of Type C, ALIGN-L(FP, Ft) also plays an important role in focal phrasing. What’s more, ALIGN-L(FP, Ft) must outrank FTMIN to allow for the appearance of a monosyllabic foot, as shown in (16) above. Both candidates (16a) and (16d) incur one violation in the ALIGN-L(FP, Ft) constraint because the left edge of the focal phrase does not align with the left edge of the foot. Thus, (16a) and (16d) are ruled out. Although (16b) and (16c) satisfy ALIGN-L(FP, Ft), (16c) violates

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

ALIGN-L(IC, Ft) because the left edge of xiao bao does not align with the left edge of the foot. As a result, (16b) is successfully selected as the optimal output.

So far, one may ask why ALIGN-L(F, FP) and ALIGN-L(FP, Ft) operate in Type A and the unmarked Type C. To be specific, what is the theoretical implication of ALIGN-L(F, FP) and ALIGN-L(FP, Ft) with regard to Mandarin typology? A possible answer would be the interleaving between focus and Mandarin lexical tones.

Phonetically, when focus occurs, the pitch range of the focused element is greatly expanded. In Mandarin, Tone2 is a LH tone while Tone3 is a L tone. A LH tone is more prominent than a L tone because it ends with a H pitch. Therefore, the focus would prefer to fall on the position where the pitch is prominent in surface representation. To be concrete, ALIGN-L(F, FP) and ALIGN-L(FP, Ft) let the focused element be at the sandhi position since a L tone always undergoes tone sandhi and changes to a LH tone if followed by another L tone. Thus, the sandhi position at which pitch raises from a L tone to a LH tone is where the focused element prefers to be.

4.2.2 A

^LIGN

-R(F, FP) & A

^LIGN

-R(FP, Ft)

The ALIGN-R(F, FP) is proposed to capture the fact that the focused element is at the right edge of the focal phrase. The definition is given below.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(17) ALIGN-RIGHT(focused element, focal phrase): the right edge of the focused element (F) coincides with the right edge of the focal phrase (FP).

The ALIGN-R(F, FP) constraint aligns the right edge of the focused element with the right edge of the focal phrase. In Type B and the marked reading of Type C, every candidate which observes this requirement would incur a fatal violation in the ALIGN-R(F, FP) constraint. Take Type B, for example.

(18) wu wu jui wu wu ‘55955’

wu wu jui wu wu

[σσσσσ] ALIGN-R(F, FP) ALIGN-L(F, FP)

☞ a. {σσσ}FP{σσ} *

b. {σσ}{σσσ}FP *!

Further, to ensure that the foot formation ends at the focused syllable, another alignment constraint is necessary.

(19) ALIGN-RIGHT(focal phrase, foot): the right edge of the focal phrase (FP) coincides with the right edge of the foot (Ft)

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

The alignment of the focal phrase and the foot at the right edge means that the foot formation ends at the right edge of the focal phrase. The ALIGN-R(FP, Ft) constraint comes into play in Type B and the marked Type C. With regard to its ranking, the ALIGN-R(FP, Ft) constraint must dominate FTMIN to allow a focal syllable to form a monosyllabic foot. First, consider the example of the marked Type C.

(20) gou yao xiao bao ‘Dogs bite Small bao.’

gou yao xiao bao

[σ[σ[σσ]]] ALIGN-R(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {

σ}

FP{σσσ}

a. (σσ)(σσ) *!

☞ b. (

σ)(σ(σσ))

* *

c. (

σ)((σσ)σ)

* * *!

d. (

σσσσ)

*! * *

As can be seen in tableau (20), the focused syllable is the monosyllabic subject gou.

Both candidates (20a) and (20d) violate ALIGN-R(FP, Ft) because the right edge of the focal phrase does not align with the right edge of the foot. Instead, gou continues associating with the following syllables to form a disyllabic foot and a quadrisyllabic foot, respectively. Thus, (20a) and (20d) are ruled out. Other the other hand, candidates (20c) and (20b) do not violate the ALIGN-R(FP, Ft) constraint because the

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

focal phrase gou aligns with the right edge of the foot. As a result, the focal phrase is isolated as a monosyllabic foot. Although (20b) and (20c) violate FTMIN, they are not rejected because of the full satisfaction of the higher ranked ALIGN-R(FP, Ft).

However, in (20c), the left edge of xiao bao does not align with the left edge of the foot. It causes a fatal violation in the ALIGN-L(IC, Ft) constraint. Therefore, candidate (20b) is selected as the optimal output.

(21) gou yao xiao bao ‘Dogs bite Small Bao.’

gou yao xiao bao

[σ[σ[σσ]]] ALIGN-R(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {σσ}FP{σσ}

☞ a. (σσ)(σσ)

b. (σ)(σ(σσ)) *! * *

c. (σ)((σσ)σ) *! * * *

Tableau (21) demonstrates the situation where the focus shifts to the verb yao.

Candidates (21b) and (21c) fail to satisfy the highest ranked ALIGN-R(FP, Ft) constraint owing to the mismatch of the focal phrase and the foot. That is, the foot formation does not end at the focal syllable yao; instead, yao proceeds to form a trisyllabic foot with the object xiao bao. As a result, (21a) is the optimal candidate among all of the candidates.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(22) gou yao xiao bao ‘Dogs bite Small Bao.’

gou yao xiao bao

[σ[σ[σσ]]] ALIGN-R(FP, Ft) FTMIN FTMAX ALIGN-L(IC, Ft) Focal Phrase: {σσ}{σσ}FP

☞a. (σσ)(σσ)

b. (σ)(σ(σσ)) *! *

In (22), it seems that (22a) and (22b) may be grammatical output since there is no violation in the ALIGN-R(FP, Ft) constraint. The right edge of the focal phrase accurately aligns with the right edge of the foot. However, in (22b), the subject gou is isolated as a monosyllabic foot. Candidate (22b), therefore, is ruled out due to the violation of the FTMIN constraint. The optimal candidate is (22a). Now consider the focal phrasing in Type B.

(23) wu wu jui wu wu ‘55955’

wu wu jui wu wu

[σσσσσ] A^LIGN-R(FP, Ft) FTMIN FTMAX

Focal Phrase: {σσσ}FP{σσ}

☞a. ((σσ)σ)(σσ) *

b. (σ)(σσ)(σσ) *!

c. (σσ)((σσ)σ) *! *

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

Tableau (23) demonstrates the focal phrasing of a list of digits. Candidate (23c) violates ALIGN-R(FP, Ft) since the right edge of the focal phrase does not align with the right edge of the foot. Candidates (23a) and (23b) fully satisfy the ALIGN-R(fp, ft) constraint because the phrasing ends at the focal digit. However, (23b) incurs one violation in the FTMiIN constraint because the first syllable forms a monosyllabic foot.

Consequently, candidate (23a) is successfully selected as the optimal output.

Then, what is the correlation between ALIGN-R(F, FP), ALIGN-R(FP, Ft), and Type B? Type B refers to the flat structure. By nature, a flat structure is composed of more than two constituents of the same type. The constituency of each constituent is not so tight as that of other types of sentence. Under such circumstances, the duration of the constituent would tend to be lengthened when it is under focus. The ALIGN-R(F, FP) and ALIGN-R(FP, Ft) constraints thereby regulate the focused element to be at the lengthening position.

4.2.3 A

^LIGN

-E(FP, IP)

Finally, the edges of the focal phrase and the intonational phrase should coincide; in

other words, the focal phrase is equal to the intonational phrase. One more alignment constraint is needed.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(24) ALIGN-EDGE(focal phrase, intonational phrase): the edge of the focal phrase (FP) coincides with the edge of the intonational phrase (IP)

Here we take Type A for demonstration. Notice that the intonational phrase is equal to the focal phrase.

(25) xiao li zhi xiang mai yu san ‘Small Li only wants to buy umbrellas.’

xiao li zhi xiang mai yu san

[σσ[σ[σ[σ[σσ]]]] ALIGN-E(FP, IP) Focal Phrase: {σσ}{σσσσσ}FP

☞ a. {σσ}IP{

σσσσσ}

IP

b. {σσσ}IP{σσσσ} IP *!

As the above tableau illustrates, the focal phrase in (25b) crosses the boundary of the intonational phrase. Thus, (25b) is eliminated in violation of the ALIGN-E(FP, IP) constraint. The focal phrase in (25a) does not cross the boundary of the intonational phrase so it is successfully selected as the optimal output.

4.2.4 [F

T

M

IN

& N

ON

-F

INALITY

]

So far, ALIGN-R and ALIGN-L can successfully account for most of the focal

‧

phrasing data in Chapter 3. However, when the focus is on the penultimate or the last syllable, the use of ALIGN-R and ALIGN-L would lead to a wrong prediction. Examples are given in (26).

Syntactic Structure [gou [yao [gou]]] [gou [yao [gou]]]

Focus

gou yao

Constraint ALIGN-L(FP, Ft) ALIGN-R(FP, Ft) Wrong Phrasing (gou yao) (gou) (gou yao) (gou) Correct Output (gou (yao gou)) (gou (yao gou))

For the unmarked Type C, the focus is on the last syllable gou. One might assume that the ALIGN-L(FP, Ft) constraint would align the left edge of gou with the left edge of the foot. However, gou is not isolated as a monosyllabic foot but forms a trisyllabic

For the unmarked Type C, the focus is on the last syllable gou. One might assume that the ALIGN-L(FP, Ft) constraint would align the left edge of gou with the left edge of the foot. However, gou is not isolated as a monosyllabic foot but forms a trisyllabic

在文檔中 華語焦點變調 - 政大學術集成 (頁 59-0)