< Previous420∙Cho, Keeseok ․ Kim, Sung-hun ․ Bae, Sang-heeThe Journal of Studies in Language Vol. 34, No. 3, 2018d.The options should be chosen in accordance with economy principle.Option (18a) demonstrates the case of attracting another head to satisfy the affix feature of W. This option is divided into two cases. The first case is that X has a base-merged lexical item. If X has a base-merged lexical item that can bear the imperative feature of W, X raises to W instead of W lowering to X, since option (18a) has priority over option (18b).The second case is that X does not have a base-merged lexical item, but Y has a base-merged lexical item. Suppose that the base-merged lexical item at Y cannot raise to X. Then the imperative feature of W has no other option but to be lowered to Y. This is the case in positive imperative sentences with main verbs in English. Suppose that the base-merged lexical item at Y can raise to X. If the lexical item that has raised to X from Y can also raise to W, the imperative feature of W is borne by the lexical item by successive raising. If this successive raising is not available, the imperative feature of W is borne by the lexical item by lowering of the imperative of W to the lexical item at X that has raised from Y.Option (18c) is used as a last resort when neither raising nor lowering is available. For example, if there can be a head that is c-commanded by W and c-commands X, then a certain lexical item can be base-merged at W to host the imperative feature of W. Otherwise, option (18c) cannot be used.Now let us bring up the four essential issues in imperative constructions. They are repeated here as (19a), (19b), (19c), and (19d).(19)a.Why is do-support necessary in English negative imperative constructions?b.Why is the contraction of not and do necessary in inverted negative imperative constructions?c.Why should the negative not be pied-piped along with do in inverted negative imperative constructions?d.Why can’t positive emphatic imperative sentences be inverted sentences?We will first discuss the issue (19a). Consider the following sentences.(20)a.Work hard, will you?b.Do work hard, will you?c.*Not listen to her, will you?d.Do not listen to her, will you?Sentences (20a) and (20b) are positive imperative sentences that show that do-support is optional. Let us explore the syntactic structure of the positive imperative sentences.Structure (21) shows the syntactic structures of positive imperative sentences (20a) and (20b). The abstract imperative feature of Comp should be borne by the lexical item. In positive imperative constructions such as (20a), the imperative feature of Comp is lowered to the lexical item work, bypassing the null T. In positive emphatic imperative constructions such as (20b), the verb do, which is base-merged at T, can be raised to Comp to host the emphatic imperative feature of Comp, or the emphatic imperative feature of Comp can be lowered to the verb do. The first option, which is T-to-C movement, is syntactically vacuous for two reasons. First, it does not A Unified Analysis of Imperative Sentences in English: Adjacency Approach between Lexical Bearers of Imperative Features and Functional Locus of Imperative Features∙421The Journal of Studies in Language Vol. 34, No. 3, 2018produce any NP-Aux inversion effect. Second, it does not produce extra semantic effects since the sentence is already an emphatic positive imperative sentence before the movement takes place. The T-to-C movement of do violates the economy principle (18d). Therefore, the emphatic imperative feature of Comp is lowered to do, which is option (18b). Sentences (20c) and (20d) are negative imperative sentences that show do-support is obligatory. Let us explore the syntactic structure of negative imperative sentences. (22)Structure (22) shows the syntactic structures of negative imperative sentences (20c) and (20d). The abstract imperative feature of Comp should be hosted by the lexical item. In negative imperative sentences such as (20c), the main verb listen cannot raise to T let alone raise to Comp since English Tense does not have the strong V-feature to attract the main verb. The second option is lowering the imperative feature of Comp to the main verb listen. This is also not possible because the intervening not, which is syntactically a head, blocks the lowering of the imperative feature of Comp, which is a kind of head movement. The lowering the imperative feature of Comp to the lexical item is possible only when the former and the latter are adjacent to each other. Therefore do-support is used as a last resort to rescue the sentence, as shown in (20d). This is the option (18c).Now let us go to the second issue (19b) and the third issue (19c). Consider the following sentences.(23)a.You do not talk here.b.You don't talk here.c.Don’t you dive here.d.*Do not you dive here.e.*Do you not dive here.(21)422∙Cho, Keeseok ․ Kim, Sung-hun ․ Bae, Sang-heeThe Journal of Studies in Language Vol. 34, No. 3, 2018Sentences (23a) and (23b) show that the contraction of do and not is optional in negative imperative sentences with a pronoun subject. However, in inverted negative imperative sentences such as (23c) and (23d), the contraction of do and not must take place. Let us look into the syntactic structure of inverted negative imperative sentences.(24)Structure (24) shows the syntactic structure of inverted negative imperative sentence (23c). This structure is derived from T-to-C movement, which is syntactically a head movement. Since two heads cannot move together in a single head movement operation, the two heads do and not are contracted into the one head don’t, which in turn undergoes T-to-C movement to derive the sentence (23c). If the two heads do and not undergo head movement to Comp without contraction, two heads are jammed into one head, deriving a geometrically ill-formed sentence. This is why the contraction of do and not must take place in the inverted negative imperative sentence.Sentence (23e) shows that the T-to-C movement of do cannot strand the negative not in the inverted negative imperative sentence. Let us look into the syntactic structure of sentence (23e).(25)In structure (25) the negative not fails to be carried along with do, and is stranded in the negative phrase. The stranded negative not fails to bear the imperative feature of Comp and thus cannot participate in the formation of the imperative meaning of the sentence.4) Therefore, sentence (23e) cannot be the right form for the inverted negative imperative sentence.4) [αβγ]Suppose thatαis the functional locus of the imperative feature, and the verbal element β and the negativeγ are the potential lexical bearers of the imperative features. In the negative imperative constructions, β and γ should both adjacent to form one negative imperative unity. In the case that β and γ undergo raising to the functional locus of imperative feature α, β and γ should be contracted into one head before raising to α, which can host only one head. The negative imperative unity is well-supported from Korean. In the Korean imperative constructions, the imperative maker is ‘-a’ and the negative imperative marker is ‘-ma’, which is a united form of the imperative marker and the negative element.A Unified Analysis of Imperative Sentences in English: Adjacency Approach between Lexical Bearers of Imperative Features and Functional Locus of Imperative Features∙423The Journal of Studies in Language Vol. 34, No. 3, 2018Now let us go to the final issue (19d). Consider the following sentences.(26)a.You do work hard, will you?b.*Do you work hard, will you?Sentences (26a) and (26b) show that the positive emphatic imperative sentence cannot be an inverted sentence. The T-to-C movement of do in (26b) is a semantically vacuous operation in that it does not produce any extra semantic effects since the sentence is already an emphatic positive imperative sentence before the movement takes place. The T-to-C movement of do violates the economy principle (18d). Therefore, the emphatic imperative feature of Comp is lowered to the verb do. 5. ConclusionIn this study, we discussed essential issues in imperative constructions in English and provided licensing conditions for English imperative constructions. The essential issues discussed in this study were (i) Why is do-support required in English negative imperative constructions? (ii) Why is the contraction of not and do necessary in inverted negative imperative constructions? (iii) Why should the negative not be pied-piped along with do in inverted negative imperative constructions? (iv) Why can’t positive emphatic imperative sentences be inverted sentences?The answers for these research questions were provided in terms of the adjacency requirement between the functional host of the imperative feature and the lexical bearers of the imperative feature, as well as the economy principle that bans vacuous syntactic operations. The alternative licensing conditions offered in this study are hierarchical conditions in the sense that they are applied in order.ReferencesBelletti, Adriana and Luigi Rizzi. 1996. Introduction. In Parameters and Functional Heads: Essays in Comparative Syntax. Oxford University Press, Oxford, 3-10.Chomsky, Noam. 2005. On Phase. MA.: MIT.Davies, Eirlys. 1986. The English Imperative. Croom Helm, London, Sydney, Dover, New Hampshire.Han, Chung-hye. 2000. The Structure and Interpretation of Imperatives: Mood and Force in Universal Grammar. Outstanding Dissertations in Linguistics. Garland Publishing, New York.Hornstein, Norbert. 1995. Logical Form: from GB to Minimalism. Blackwell, Oxford.Platzack, Christer and Inger Rosengren. 1997. On the Subject of Imperatives: a Minimalist Account of the Imperative Pronoun and Negated Imperative. Journal of Comparative Germanic Linguistics 1.3, 177-224.Potsdam, Eric. 2007. Analyzing Word Order in the English Imperatives. In Imperative Clauses in Generative Grammar: Studies Offered to Frits Beukema, ed. Wim van der Wurff, 251-272. Amsterdam; John Benjamins.Rupp, Laura. 2003. The Syntax of Imperatives in English and Germanic: Word Order Variation in the Minimalist Framework. New York: Palgrave Macmillan.424∙Cho, Keeseok ․ Kim, Sung-hun ․ Bae, Sang-heeThe Journal of Studies in Language Vol. 34, No. 3, 2018Scontras, G., & Gibson, E. 2011. A Quantitative Investigation of the Imperative-and-Declarative Construction in English. Language, 817-829.Zhang, S. 1990. The Status of Imperatives in Theories of Grammar. Ph.D. Dissertation, University of Arizona.Cho, Keeseok, Professor107 Imun-ro, Dondaemun-gu, Seoul, 13-791, Republic of KoreaDepartment of English, Cyber Hankuk University of Foreign Studies E-mail: d9501001@hanmail.netKim, Sung-Hun, Professor152, Jukjeaon-ro, Suji-gu, Youngin-si, Gyonggi-do, 448-701 Republic of KoreaDepartment of English and American Humanities, Dankook UniversityE-mail: shkim@dankook.ac.krBae, Sang-Hee, Professor119, Dandae-ro, Dongnam-gu, Cheonan-si, Chungnam, 330-714, Republic of KoreaGeneral English Department E-mail: bsh102@dankook.ac.krThe Journal of Studies in LanguageThe Journal of Studies in Language 34(3), 425-438 (2018)1)The Interface between Perception and Production in L2 English VowelsLee, Shinsook* and Cho, Mi-Hui**Korea UniversityKyonggi University*First Author / **Corresponding AuthorABSTRACTThe Journal of Studies in Language 34.3, 425-438. The study explores the interface between perception and production in L2 English vowels focusing on the precedence relationship and the correlation between perception and production. To that end, 34 Korean EFL learners identified 12 English vowels presented in the bVt word structure and their productions of the same English vowels were identified by native English speakers. The results revealed that overall the precedence relationship of perception over production was attested but not across all the target vowels. The results also showed that similar error patterns were observed in perception and production for the same vowels. However, no correlation was found between perception and production, suggesting that L2 speech perception and production may not develop concurrently. The findings of the study are discussed in light of L2 learning, especially in light of the development of L2 perception and production skills. (Korea University · Kyonggi University)Keywords: English vowels, perception and production, precedence, correlation, L2 learningOPEN ACCESShttps://doi.org/10.18627/jslg.34.3.201811.425pISSN : 1225-4770Received: October 01, 2018Revised: November 06, 2018Accepted: November 14, 20181. IntroductionThe Spoken speech communication involves talker-listener parity and thus the relationship between speech perception and production has received much discussion in the language acquisition literature (Bradlow et al., 1997; de Jong et al., 2009a, b; Flege, 1995; Pater, 2004). Specifically, motor theory (Liberman and Mattingly, 1985, 1989) contends that listeners perceive speech sounds according to their own articulatory gestures, indicating that there is a close link between speech perception and production. In a similar vein, the direct-realist approach (Best, 1994; Folwer, 1986) maintains that listeners directly perceive talkers’ articulatory gestures, claiming that speech perception and production are closely connected.This is an Open-Access article distributed under the terms of the Creative Commons Attri-bution NonCommercial License which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.Copyrightⓒ2018 the Modern LinguisticSociety of Korea본인이 투고한 논문은 다른 학술지에 게재된 적이 없으며 타인의 논문을 표절하지 않았음을 서약합니다. 추후 중복게재 혹은 표절된 것으로 밝혀질 시에는 논문게재 취소와 일정 기간 논문 제출의 제한 조치를 받게 됨을 인지하고 있습니다.-The first author was supported by the College of Education, Korea University Grant in 2018. We are grateful to the anonymous reviewers for providing helpful comments on an earlier version of the paper.426∙Lee, Shinsook ․ Cho, Mi-HuiThe Journal of Studies in Language Vol. 34, No. 3, 2018On the other hand, the acoustic-auditory approach of speech perception (Stevens and Blumstein, 1981) contends that listeners perceive speech sounds in terms of the acoustic signals independently of articulation that generated them but does not make any specific claims about the relationship between perception and production, as noted by Bradlow et al. (1997). Concerning the relationship between speech perception and production for L2 learners, several studies reported that there is a moderate correlation between the two (Flege et al., 1997; Flege et al., 1999; Schmidt and Flege, 1995). For instance, Flege et al. (1997) examined the perception and production of English front vowels by L2 learners from different L1 backgrounds. The results showed that the learners’ production accuracy of the vowels was related to the learners’ perception accuracy to some degree (p. 437). However, some studies found no correlation between speech perception and production (Peperkamp and Bouchon, 2011). Specifically, Peperkamp and Bouchon (2011) looked into the perception and production of English front vowels /i/ and /ɪ/ by French-English bilinguals and reported that no correlation was found between the bilinguals’ perception and production of the vowel contrast. Other studies further documented that perception and production skills in L2 do not develop in tandem (de Jong et al., 2009b).As for English vowel studies, Yang and Whalen (2015) investigated the interface between American English speakers’ perception and production of English vowels. Native English speakers identified ideal English vowels from synthetic vowel stimuli and they also produced the same vowel stimuli in a clear speech. The formants of the native speakers’ vowel productions were measured and the native speakers’ produced vowel space was compared to their perceived vowel space. The results showed that the native speakers’ produced vowel space diverged from their perceived vowel space, indicating that the speakers’ vowel productions do not necessarily reflect their perception of the vowels.Hong (2015) examined American and Korean speakers’ English vowel identification using the stimuli from the database of Hillenbrand et al. (1995). He reported that native American English speakers used dynamic spectral cues and duration when identifying English vowels whereas Korean speakers showed a tendency to use duration and static spectral cues. Similarly, in the production study Hong (2016b) found that native American English speakers’ vowel signals showed dynamic spectral properties while those of Korean speakers exhibited static spectral properties. The findings of Hong (2015, 2016b) seem to suggest that native English speakers showed similar patterns when perceiving and producing English vowels. Korean speakers also exhibited similar perceptual and production patterns for English vowels.Given that whether there is a direct link between speech perception and production in L2 is still controversial, one important question that should be addressed in L2 acquisition is the degree to which L2 speech perception and production are linked. However, not many studies have investigated the interface between the two, especially the relationship between vowel perception and production by the same L2 learners. Thus, the present study explores the relationship between perception and production of English vowels by Korean EFL learners. Specifically, the present study investigates the precedence relationship between English vowel identification and production and also the correlation between the two in order to shed light on the long-standing issue of the interface between speech perception and production.The Interface between Perception and Production in L2 English Vowels∙427The Journal of Studies in Language Vol. 34, No. 3, 20182. Literature Review2.1 Precedence Relationship between Perception and ProductionIn first language acquisition it is well-known that speech perception and production do not develop hand in hand. That is, infants' productive abilities lag behind receptive abilities in that infants' receptive skills are a prerequisite to productive skills (Best, 1994; Gnanadesikan, 2004; Jusczyk, 1997; Pater 2004; Schiller and Meyser, 2003). Likewise, Flege (1995) argued for a perceptual basis for foreign accents. Specifically, Flege contended that L2 learners tend to perceive L2 sounds in terms of the closest L1 sound categories, especially when L2 sounds and their L1 counterparts share the phonetic space to a greater extent. This causes L2 learners to fail to notice subtle acoustic and phonetic differences between L2 sound categories and their L1 analogues, which hinders L2 learners from establishing new phonetic sound categories for the L2. This in turn leads L2 learners to have accented speech as their production in L2 is guided by incomplete L2 sound categories. Similar arguments were made by Flege and his colleagues (Flege et al., 1997; Flege et al., 1999).The precedence of perception over production has been documented in the L2 acquisition literature. For instance, Flege et al. (1997) maintained that L2 learners’ errors in producing English front vowels could be attributed to their perceptual errors of the vowels. Bradlow et al. (1997) also reported that Japanese learners’ perceptual training of English /l/ and /r/ alone led to noticeable improvements in pronunciation, even without explicit training to produce the target sounds. Along the similar lines, Lee (2011) reported that advanced Korean EFL learners’ perception abilities were ahead of their production abilities in the acquisition of English fricatives.On the other hand, several studies showed that speech production took precedence over speech perception in L2 acquisition. For example, Sheldon and Strange (1982) reported that Japanese learners produced English /l/ and /r/ far better than perceiving them. Likewise, Korean learners’ production abilities were significantly ahead of their perception abilities in the acquisition of English voiceless coronal fricatives (Joh and Lee, 2001), English consonant clusters (Lee and Cho, 2005), and English /l/ and /r/ (Sung, 2006). However, Cho and Jeong (2013) investigated the acquisition of 11 American English vowels by Korean learners and reported that the precedence relationship between perception and production varied with individual target vowels. Specifically, the Korean learners were better at perceiving English [oʊ] than producing it while they were better at producing English [eɪ] and [u] than perceiving them. Given that two opposite patterns of the precedence relationship between perception and production were reported in the L2 acquisition literature, one of the goals of the current study is to examine the precedence relationship between English vowel perception and production by Korean learners in order to provide another testing ground for this issue. 2.2 Correlation between Perception and ProductionBohn and Flege (1997) contended that perception may precede production in the initial stages of L2 acquisition but perception is more likely to be resistant to pliability than production as L2 learners’ experience with the target language increases, indicating that the two skills may not develop in tandem. Strange (1995) maintained that the degree of alignment between perception and production varies with L2 learners’ experience with the target language. No 428∙Lee, Shinsook ․ Cho, Mi-HuiThe Journal of Studies in Language Vol. 34, No. 3, 2018correlation may exist between perception and production for experienced L2 learners whereas perceptual errors may predict production errors for inexperienced L2 learners.Fabra and Romero (2012) examined the perception and production of English vowels by Catalan learners with different levels of English proficiency; high, mid, and low English proficiency groups. The learners discriminated between English /i/-/ɪ/, /ɛ/-/æ/, /ɑ/-/ʌ/, and /u/-/ʊ/ vowel contrasts and they also produced English words containing one of the following vowels: /i/, /ɪ/, /ɛ/, /æ/, /ɑ/, /ʌ/, /u/, and /ʊ/. The results revealed that the Catalan learners were overall poor at discriminating between the vowel contrasts but the learners, in particular high and mid English proficiency group learners, were good at discriminating between English /i/-/ɪ/ and /u/-/ʊ/ contrasts. Acoustic measurements of the learners’ productions (only female speakers’ productions) indicated that the Catalan learners gave heavy weight to the F2 dimension compared to native English speakers. That is, the learners showed narrowed distances in the vowel space between the two members in the pairs /i/-/ɪ/, /ɛ/-/æ/, /ɑ/-/ʌ/, and /u/-/ʊ/ (p. 502). However, productions of more proficient learners progressed toward native-like norms with respect to vowel expansion and duration. Also, native English speakers’ judgments of the vowels produced by the Catalan learners indicated that their productions of /i/, /ɛ/, /ʌ/, and /ɑ/ were more problematic compared to other vowels but their productions tended to show better identification as their English proficiency became high. The results further revealed no significant correlation between perception and production either at the individual level or at the group level.Peperkamp and Bouchon (2011) tested French-English bilinguals’ perception and production of the English /i/-/ɪ/ contrast. The bilinguals discriminated between the vowel contrast in an ABX discrimination test1) and they also produced the vowels embedded in sentences. The bilinguals’ vowel productions were assessed for global nativeness by native English speakers and also identified by a native English speaker. Peperkamp and Bouchon (2011) reported that there was no correlation between the French-English bilinguals’ perception and production of the English vowel contrast.As for the acquisition of English vowels by Korean learners, most studies focused either on vowel perception or vowel production. For example, Hong (2015) examined 57 Korean university students’ and 5 native American English speakers’ perception of American English vowels /i, ɪ, ɛ, æ, ɑ, ɔ, ʌ, ʊ, u/ in the h_d forms taken from the database of Hillenbrand et al. (1995), using a forced-choice identification test. Hong found that Korean students had a tendency to use duration along with static spectral properties while native American English speakers used dynamic spectral properties and duration when they identified the target vowels. In a follow-up study, Hong (2016a) reported that Korean university students with a high-level of English proficiency tended to use dynamic spectral cues and duration similar to native American English speakers, whereas those with a low-level of English proficiency used static spectral cues and duration.Hong also (2016b) conducted a production experiment where 18 Korean university students produced 9 American English vowels /i, ɪ, ɛ, æ, ɑ, ɔ, ʌ, ʊ, u/ in the h_d forms. The results showed that the Korean students’ productions of English vowels /ʌ/, /ɪ/, /æ/, /ɑ/, and /ʊ/ deviated from the target vowels but those of English /i/, /u/, and /ɛ/ approximated the targets vowels. Specifically, the Korean students' productions of English /ʌ/ were confused with English /ɑ/ and /ɔ/ and their productions of English /ɪ/, /æ/, /ɑ/, and /ʊ/ were also confused with English /i/, /ɛ/, /ɔ/, and /u/, respectively. Hong noted that Korean students’ production difficulties of English /ʌ/, /ɑ/, and /ɔ/ were likely to be related to their 1) The participants were requested to judge whether the stimulus word X was the same as the stimulus words A or B.The Interface between Perception and Production in L2 English Vowels∙429The Journal of Studies in Language Vol. 34, No. 3, 2018perceptual difficulties given that Korean students also showed perceptual difficulties identifying English /ʌ/, /ɑ/, and /ɔ/ as reported in Hong (2014). Importantly, however, the Korean students in the perception experiment and those in the production experiment in Hong (2014, 2015, 2016a, 2016b) were not the same learners. Accordingly, it is not clear whether Korean learners’ production difficulties of English vowels were closely related to their perceptual difficulties of the vowels. Thus, studies that investigate both perception and production of English vowels by the same learners are called for.3. Experiment3.1 ParticipantsThe participants were 34 Korean learners of English enrolled in English language education courses at a university in Seoul. They consisted of 10 male and 24 female students and their mean age was 22.5 years when the experiment was carried out. Their English proficiency was rated either upper-intermediate or advanced in that their average IBT-TOFEL score was 102 with the range from 85 to 115. No participants had the experience of residing in English-speaking countries more than 8 months.3.2 MaterialsThe target vowels were 12 American English vowels: /i, ɪ, eɪ, ɛ, æ, ɝ, ɑr, ʌ, ɑ, ɔ, oʊ, u/. The target vowels were embedded in the b_t forms (beat, bit, bait, bet, bat, burt, bart, but, bot, bought, boat, boot). Due to a possible variation in word frequency among the stimuli, high frequency English words with the same target vowels were also provided (seed, sit, eight, set, hat, world, card, cut, hot, law, go, fruit, respectively), as in Evans and Iverson (2007). The stimulus words were produced by a female native speaker of American English in the frame of “Say _____ again”. The native speaker was from Ohio in the US and 20 years old. The recording was conducted in a soundproof lab at a university at a sampling rate of 16-bit/44.1kHz. Speech Filing System (SFS) and a RODE-NT1A microphone were used in the recording. The recordings of the native speaker were inspected by 2 native American English speakers with several years of phonetic training at University College London (UCL) and also by a phonetician who had been teaching pronunciation training courses at UCL.3.3 ProcedureEnglish words with the target vowels were displayed on the computer monitor and the participants produced the words in the frame of ‘Say _____ again’ three times. Before the test, the participants went over the stimuli for familiarization. The vowel production test was conducted in a sound-attenuated room with a Sony ECM-MS907 microphone using Speech Filing System (SFS). The recordings were digitized at 44,1kHz and stored as wave files at a laptop computer. After the recording, audio files were edited for native English speakers’ identification of the words (i.e., vowels). Specifically, no more than three sequences of front/back or high/mid/low vowels occurred in a row and Next >