correctly stressed on the penultimate syllable. High achievers’ performance in trisyllacbic noun test was as consistent as they did in the disyllabic noun test. As to low achievers, their preference for the penultimate stress is observed again in the XV-ending test. The main stress consistently fell on the penultimate syllable no matter whether the penult was light or heavy. The result echoes low achievers’
performance in disyllabic nouns, where non-XOV ultima were generally unstressed.
On the other hand, compared with the result obtained in the XVO-ending test in Table 12, it is confirmed that stress assignment in XVN/L-ending and XV(V)-ending nouns were more alike. A possible reason is that, as Huang (1990) indicated, in some quantity insensitive language like Bunun, nasal or liquid codas are not included in syllable weight, which makes XVL or XVN as heavy as light as HV(V). Low achievers’ stress assignment in disyllabic and trisyllabic simplex nouns has revealed a similar way of weight-counting.
Table 13
L2 Learners’ Stress Assignment of XV(V)-Ending Nouns
Stress Location Syllable types of
penult and ultima
Low-achievers High-achievers
Stimulus antep. penult ultima antep. penult ultima
XV + XV(V) retina 11.67% 88.33% 0% 75% 25% 0%
XVN/L + XV(V) agenda 21.67% 78.33% 0% 5% 95% 0%
XVO + XV(V) kerygma 11.67% 88.33% 0% 6.67% 93.33% 0%
*Average accuracy and inaccuracy rates are not provided as the actual stress falls on different syllables.
Extended EDCD Analysis
The extended Error-Driven Constraint Demotion (EDCD) algorithm proposes that interlanguage stress grammar follows a bidirectional model. To demonstrate how the algorithm is applied to stress assignment in English simplex nouns, this section will illustrate the operation of the proposed model based on the elicitation data.
The extended EDCD claims that the initial state of L2 acquisition is L1 setting.
At the initial state, no L2 stimuli have been exposed to yet, and thus the state is static.
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
To demonstrate how the interlanguage grammar of English simplex nouns is constructed step by step, the first task is to explore the initial state of it, namely the grammar of nominal stress in Chinese.
The core constraints in stress grammar concern a language’s preference for foot structure, foot type, footing direction, and so on. As far as foot structure is concerned, Duanmu (2000) observed that light syllables cannot be stressed in Chinese and claimed the existence of moraic feet is not sufficiently supported in Chinese. The existence of stressed light syllables and moraic feet in Chinese can be regulated by FOOT-BINARITY(σ) in (2).
(2) FOOT-BINARITY(σ) (McCarthy & Prince, 1986/1996; Prince, 1993).
Assign one violation mark for every foot that does not contain at least two syllables
Besides, to avoid a flat structure, PARSE in (3) is indispensable.
(3) PARSE (McCarthy & Prince, 1986/1996; Prince, 1993).
Assign one violation mark for every syllable that is unparsed into a foot.
As for foot type, Duanmu (2000) suggested that in all Chinese dialects, disyllabic words are always 1-2 (or 1-04) and 2-1 is transformed from 1-2 in final position. The preference for syllabic trochee in Chinese can be captured by TROCHEE(σ) in (4).
(4) TROCHEE(σ)
Assign one violation mark for every foot whose head is not initial.
Quantity-insensitive languages, like Chinese, should be blind to syllable weight and accepts only disyllabic foot. Therefore, [HL] are [LL] both considered compliance with TROCHEE(σ), yet [H] will incur a violation of TROCHEE(σ).
Furthermore, Feng (1998a) claimed that Chinese footing direction is left-to-right based on the evidence from some multisyllabic nouns in (5).
(5) a. Five syllable string (jin mu) (shui huo tu) *(jin mu shui) (huo tu)
4 1 represents primary stress, 2 represents secondary stress, 0 represents lack of stress or lexical tone.
•‧
‘fuel, rice, oil, salt, (soy)sauce and tee, daily necessities.’
(Bu-yi) (-nuo-si) (ai-li-si) *(Bu-yi-nuo) (-si-ai) (-li-si) ‘Buenos Aires’
Feng (1998) observed the stray syllable in the syllable strings can only be grouped with the last foot, which suggests that Chinese footing is from left to right.
The preference for a rightward footing directionality can be captured by ALLFTL in (6).
(6) ALLFTL (McCarthy & Prince, 1993a)
Assign one violation mark for every foot whose right edge is not aligned with the left edge of a prosodic word
Since TROCHEE(σ) is violated in the final position, it is assumed to be ranked lower than the other constraints. The constraint ranking for Chinese stress is listed in (7).
(7) Constraint ranking for nouns in Chinese PARSE, FT-BIN(σ), ALLFTL >> TROCHEE(σ)
It is true that Chinese stress grammar for nouns involve other requirements on tones, stress clash, and so on. However, the following analysis will focus on the core constraints listed in (7) due to the limit of space.
Tableau (8) demonstrates how the word da4xue2 ‘university’ is assigned primary stress in Chinese using the constraint ranking obtained above. Here, candidate (a) represents a trochaic foot type; candidate (b) has both syllables stressed, and candidate (c) is iambic. The evaluation shows none of the candidates violates PARSE. As can be seen, candidate (b) is seriously punished by FT-BIN(σ) and ALLFTL.
Each monosyllabic foot is against FT-BIN(σ) once. Besides, the right foot fails to
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
coincide with the word on the left edge, which incurs a violation of ALLFTL.
Candidate (a) and (c) are equally good in terms of foot structure and footing direction.
However, candidate (c)’s iambic foot type is punished by TROCHEE(σ) and results in its elimination. Candidate (a), a syllabic trochee, surfaces as the optimal output.
(8) Chinese stress assignment of daxue ‘university’
daxue FT-BIN(σ) ALLFTL PARSE TROCHEE(σ)
→ a. [1-2]
b. [1][1] *!* *
c. [2-1] *!
According to the extended EDCD, the constraint ranking in tableau (8) serves as the initial state of the acquisition of L2 stress for Chinese speakers. Now that the L1 setting is ensured, we can proceed to L2 stress grammar. Pater (2000) shares Prince &
Smolensky (1993)’s view of alignment-based approach, assuming that ALIGN-HEAD
is fulfilled both by the foot bearing the stress in English, as defined in (9).
(9) ALIGN-HEAD (PrWd-R, Head(PrWd)-R) (hereafter ALIGN-HEAD
(word, head(word))
Align the right edge of the prosodic word with the right edge of the head of the prosodic word
In Pater’s analysis (2000), ALIGN-HEAD assigns different numbers of violation marks based on the distance between the head syllable and the right edge of the word.
For example, an output like [Cána]da is assigned two violation marks, while an output like Ca[náda] is punished only once. Placing ALIGN-HEAD in the undominated position will cause the ultimate syllable to bear the main stress. However, in English noun stress assignment, stress does not lie in the ultimate position but in the antepenultimate or penultimate position. According to Pater, that is the result of the dominance of NON-FINALITY over ALIGN-HEAD in (10).
(10) NON-FINALITY (NON-FIN)
The head of the prosodic word must not be final
It is noteworthy that in Pater’s analysis, NON-FIN punishes a word whose head
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
foot coincides the word on the right edge, such as Ca[nada], but forgives a word whose head foot coincides the word on the left edge, such as [Cana]da. In other words, the so-called “final” in the constraint requirement is defined by word edge rather than sequence.
In English, nouns with light penultimate, such as Canada, America, cinema, are assigned an antepenultimate stress. Besides NON-FINALITY(head(word)), Pater (2000) indicates that antepenultimate stress can be captured by the constraints on binarity and left-headedness, represented by FT-BIN and TROCHEE(µ) in (11) and (12) respectively.
(11) Foot-BINARITY(µ) (hereafter FT-BIN(µ))
Assign one violation mark for every foot that does not contain at least two morae.
(12) TROCHEE(µ)
Assign one violation mark for every foot whose head is not initial.
It is noteworthy that, according to Hayes(1995), TROCHEE(µ) can be satisfied by left-headed disyllabic feet regardless of syllable weight, including [LL] or [H]. Since English primary stress is either penultimate or antepenultimate, ALIGN-HEAD is assumed to be violable, and ranked lower than the other constraints. (13) shows the constraint ranking for English simplex nouns. Again, some other constraints like
*CLASH or *LAPSE are as crucial. Yet, due to the limit of space, constraints not determinant to the current issue will be disregarded here.
(13) FT-BIN(µ), TROCHEE(µ),NON-FINALITY(head(word)) >> ALIGN-HEAD
Tableau (14) demonstrates how constraint ranking proposed by Pater(2000) assigns antepenultimate stress to Canada. The winner, candidate (a), violates ALIGN-HEAD twice, for the head of the prosodic word is two syllables away from the right edge of the prosodic word. Nevertheless, candidate (a) still surfaces as the optimal choice because the violations of ALIGN-HEAD are minor. Candidate (b) adheres to ALIGN-HEAD more faithfully than candidate (a); however, it is ruled out owing that footing ultimate syllable is considered lethal in English. Candidate (c) fails to fulfill the requirement of FT-BIN(µ), for the head foot is monosyllabic. Candidate (d) is ruled out mainly because the iambic head foot is against TROCHEE(µ).