Intervocalic rhotic pronunciation by adult learners of Spanish as a second language

Please download to get full document.

View again

of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Intervocalic rhotic pronunciation by adult learners of Spanish as a second language
Document Share
Document Tags
Document Transcript
    Intervocalic Rhotic Pronunciation by Adult Learnersof Spanish as a Second Language   Timothy L. Face University of Minnesota   1. Introduction The amount of work on the second language acquisition of Spanish phonology is relatively smallin comparison to other areas of Spanish second language acquisition. Of the work on the secondlanguage acquisition of Spanish phonology, much focuses on the non-linguistic factors that influence pronunciation as a whole (i.e., without focusing on particular sounds or features in that pronunciation;e.g., Elliott 1995a, Reeder 1997, Simões 1996) or on the effects of instruction on progress toward anative-like pronunciation (e.g., Castino 1992, Elliott 1995a, 1995b, 1997, González-Bueno 1997).Work that has focused on the acquisition of specific Spanish sounds has shared a couple of characteristics. First, studies have tended to cluster around certain types of sounds and issues.Certainly the most studied sounds are the voiceless stops (e.g., González-Bueno 1997, Nathan 1987,Reeder 1997, Zampini 1994, 1998), where much attention has been paid to the fact that the same phonological stops, /p,t,k/ , exist in both Spanish and English, but with the considerable phoneticdifference that voice onset times are much shorter in Spanish than in English. Another characteristic of  studies that have focused on the second language acquisition of particular Spanish sounds is that often there is a focus on how non-linguistic factors (e.g., context of learning, fielddependence/independence, age, attitude) lead to improvement in the pronunciation of particular sounds(e.g., Díaz-Campos 2004, Elliott 1995a, 1995b, 1997, Rosenman 1987). Finally, a third characteristicof studies on the acquisition of particular Spanish sounds is that many studies tend to consider accuracy in achieving the target sound, and in some cases improvement in accuracy, but not thedevelopment of pronunciation en route to consistent accuracy. That is, even those studies that examineimprovement in the pronunciation of particular Spanish sounds often consider only the increase inaccuracy in achieving the target over time without considering the changes in error types and whatsuch development might tell us about the acquisition process.The present study diverges from each of the three common characteristics mentioned above for studies on the second language acquisition of specific Spanish sounds. First, by considering the pronunciation of rhotics, I examine sounds that have received considerably less attention in studies of Spanish second language phonology than have other sounds. Second, rather than look at the effects of non-linguistic factors on the pronunciation of intervocalic rhotics, I consider the effects of level of  proficiency in the second language (though certainly other studies have included this as well) in order to provide a cross-sectional view of development in the production of these sounds. Lastly, while the present study certainly considers the increase in accuracy of producing the target sounds across proficiency levels, of just as much interest are the non-target productions in order to gain insight intothe developmental process of learners in acquiring the Spanish rhotics.The present study investigates the acquisition of Spanish rhotic pronunciation by speakers of American English, and therefore it is pertinent to briefly consider the rhotic systems of these twolanguages. American English has a voiced alveolar approximant [  ] that varies in the details of thespecific articulation, being retroflex for some speakers and not for others, being truly alveolar inarticulation for some speakers while post-alveolar for others, having varying degrees of lip rounding, * I am grateful to two anonymous reviewers for their useful comments on an earlier version of this paper. Anyerrors are, of course, my own. © 2006 Timothy L. Face. Selected Proceedings of the 7th Conference on the Acquisition of Spanish andPortuguese as First and Second Languages, ed. Carol A. Klee and Timothy L. Face, 47-58. Somerville, MA:Cascadilla Proceedings Project.  etc. But all of these articulatory variants represent mechanisms to achieve the acoustic characteristicsassociated with American English [  ] (Ladefoged & Maddieson 1996). For the purposes of the presentstudy, where the details of the articulation are of less importance than the employment of the AmericanEnglish rhotic in attempting to pronounce the Spanish rhotics, I shall refer to this consonant simply asa voiced alveolar approximant. The voiced alveolar approximant is the only rhotic in AmericanEnglish, and is a sound that is not present in any variety of Spanish. 1 Spanish falls among the relatively small number of languages to have more than one rhotic. Allvarieties of Spanish have two rhotics, though there is some variation as to the nature of one of them.All varieties have a voiced alveolar tap [  ] and a second rhotic, which most commonly is a voicedalveolar trill [ r ]. While some varieties of Spanish have another sound instead of the voiced alveolar trill, such as a voiced uvular trill or an assibilated trill (Canfield 1981, Lipski 1994), the voicedalveolar trill is not only by far the most common across varieties, but is also uniformly the soundtaught to students studying Spanish as a second language in the United States. As the voiced alveolar tap [  ] and the voiced alveolar trill [ r ] are the two Spanish rhotics presented to American Englishspeaking students studying Spanish, these are the rhotics that will serve as target sounds in the secondlanguage and, therefore, that will be discussed here. It should be noted, however, that even in varietiesof Spanish that have these two rhotics, other sounds are sometimes substituted for them, especially inthe case of the trill (e.g., Blecua 2001, Hammond 1999).The Spanish alveolar tap [  ] is nearly identical to the American English alveolar tap produced asan allophone of  /t/ and /d/ in post-tonic position (e.g., later  , ladder  ). But while native speakers of American English have an alveolar tap in their first language, they do not associate this tap with arhotic, and this could provide some difficulty of re-categorization of this sound in acquiring Spanish.Furthermore, in American English the alveolar tap occurs exclusively intervocalically in post-tonic position, while in Spanish it occurs intervocalically in both pre-tonic and post-tonic positions, and italso occurs word-finally, syllable-finally, and as the second member of a complex syllable onset.Therefore, while American English speaking learners of Spanish should be able to produce thealveolar tap with no difficulty, given its existence also in the first language, they must learn to both produce this sound in contexts in which it does not occur in their first language and also associate thesound with a rhotic rather than view it as an allophone of  /t/ and /d/ as in American English.The Spanish alveolar trill is unlike any sound in American English. The lack of similarity canmake sounds easier to acquire in a second language since there is no need to re-categorize a sound thatexists in the first language or to recognize a small and not very salient distinction between a sound inthe first language and a similar sound in the second language (e.g., Flege 1995). However, while thelack of similarity of the Spanish alveolar trill to any sound in American English might seem to favor itsrapid acquisition, the alveolar trill is articulatorily difficult, requiring very precise control of apertureand airflow with minimal deviation. The articulatory precision required to produce the alveolar trill(e.g., Ladefoged & Maddieson 1996, Recasens 1991, Solé 2002) leads not only to difficulty for thesecond language learner in acquiring this sound, but also to it being acquired quite late, and from sixmonths to two years later than the tap depending on the criterion used in determining acquisition, bychildren acquiring Spanish as their first language (e.g., Goldstein 2000).The two Spanish rhotics contrast with each other in intervocalic position, with numerous minimal pairs existing (e.g.,  pero ‘but’ vs.  perro ‘dog’ and caro ‘expensive’ vs. carro ‘cart’, where inorthography <r> represents the tap and <rr> represents the trill). In all other positions there isneutralization of the contrast between the two rhotics, with context determining which of the twooccurs. While second language learners of Spanish, then, must learn to produce both the alveolar tapand the alveolar trill in other positions as well, intervocalic position is the only position in which theymust learn to produce both rhotics, and to do so consistently so as to adequately communicate thecontrasts signaled by the difference between these two sounds. It is for this reason that the presentstudy focuses on the production of rhotics specifically in intervocalic position. 1 Blecua (2001) documents alveolar approximant realizations for both the tap and, to a lesser extent, the trill inPeninsular Spanish. This is not the same type of alveolar approximant as English has, however, as there aredifferences both in duration and r-coloring. 48  2. Previous studies of the second language acquisition of Spanish rhotics There are two primary studies dealing with the second language acquisition of Spanish rhotic pronunciation by native speakers of American English. Major (1986) tracks four beginning learners intheir pronunciation of the Spanish rhotics throughout an intensive eight week, seven hours per day,Spanish course. The subjects read a word list and a sentence list containing the target sounds duringseven recording sessions throughout the span of the intensive course. The data from the two tasks werelumped together since there were potential confounding factors if task type were to be considered.Major presents the results in terms of the number of correct productions of the target sounds, and alsoreports the number of non-target productions that can be classified as transfer errors and the number that can be classified as developmental errors.For the target intervocalic tap, there was improvement for three of the four subjects. Two subjectsshowed considerable improvement over the span of the study, with one increasing from 3% accuracyin achieving the target in the first session to 79% accuracy in the final session, and another increasingfrom 10% to 73% accuracy. A third subject showed a slight improvement, increasing in accuracy from43% to 57%. The fourth subject showed no improvement at all over the course of the study. For theintervocalic trill, Major reports improvement for two of the four subjects. One subject improved from48% accuracy to 71% accuracy from the first to the last recording session, while another improvedfrom 52% to 100% accuracy. The remaining two speakers not only showed no improvement, but alsono accuracy in producing the target trill sound. Over the span of seven recording sessions and acombined 347 opportunities to produce the target trill, these two subjects combined to produce onlyone trill.The focus of Major’s study is on the interaction between accuracy in achieving the target, transfer errors (i.e., non-target sounds produced due to the influence of the learner’s first language), anddevelopmental errors (i.e., non-target sounds not attributable to the influence of the learner’s firstlanguage), as he attempts to support his Ontogeny Model, which makes predictions about theinteractions of these three elements in interlanguage phonology. While he does provide a list of  principle developmental errors (21 different sounds produced for the target trill and 4 for the target tap)at the end of his study, he provides no indication of the speakers’ progression in using these.The second study to consider the second language acquisition of Spanish rhotics by nativespeakers of American English was carried out by Reeder (1998). Reeder conducted a cross-sectionalstudy of the acquisition of the Spanish intervocalic trill by 40 learners, with ten at each of four stagesof Spanish language study. His subjects included university students in a first semester Spanish course,a third semester Spanish course, upper division undergraduates and graduate students, and full timefaculty teaching Spanish at the university level. The subjects read words in carrier phrases, identifieditems represented in pictures, and produced a 30-second narration in response to a written cue. Reeder,however, pools the data from all tasks together and does not consider the possible effects of task-typeon the production of the Spanish alveolar trill.Reeder presents his results in terms of accuracy in achieving the target trill, which was examinedacoustically where an accurate production consisted of multiple alveolar closures. There is nodiscussion of the nature of non-target-like productions. The results clearly show higher accuracy in producing the trill as the learner’s level of Spanish study increases. First semester students showed 7%accuracy, third semester students 13%, upper division undergraduates and graduate students 37%, andfaculty 83%. While this increase in accuracy as the level increases is what one would expect to find, itis noteworthy how low the accuracy is for all levels except faculty. What we cannot determine basedon the manner in which the results are reported is whether there are individuals with relatively highaccuracy and others with almost no accuracy, as in Major’s (1986) study. If this is so, it could explainwhy the percentages are so low when subjects are considered together. Nonetheless, it is interestingthat two of Major’s beginning level subjects showed accuracy levels nearly as high as the faculty inReeder’s study. One certainly wonders whether explicit pronunciation instruction was given in theintensive Spanish course from which Major’s subjects were selected.While the two aforementioned studies are the only ones to look in any sort of detail at the pronunciation of Spanish rhotics by native speakers of American English, other studies have 49  occasionally reported that there is more improvement for the target trill in Spanish than for soundsmore similar to ones that exist in American English (e.g., Elliott 1995b, 1997). While this mention of improvement of the trill is certainly of interest, one must wonder how much improvement there is andwhat the developmental path to accuracy in trill production looks like. None of the studies carried outto this point examine the development of pronunciation of the Spanish rhotics other than to commenton the percentage of accuracy in achieving the target. In addition, none of these studies make directcomparison to actual production patterns by native Spanish speakers, an important point given that thetarget sounds are not always produced even by native speakers. 3. Experimental methods   The present study is a cross-sectional study involving 41 native speakers of American Englishstudying Spanish at the university level, as well as a control group of 5 native speakers of Spanish. Of the native English speakers, 20 were enrolled in a fourth semester university language course – thefinal course required to complete a foreign language requirement – and 21 were Spanish majors or minors enrolled in an upper division elective course. In order to be enrolled in the fourth semester course, students were required to achieve a certain score on a placement test or to have completed thethird semester language course. These learners would be considered to be at an intermediate level of university-level language study. The more advanced group of majors and minors had reached asignificantly higher level of proficiency in Spanish. To reach that level, they were required to advance beyond the fourth semester level to a sequence of advanced language skills courses and then beyondthat to a set of core upper level courses reserved for majors and minors. Having completed those twoadditional levels beyond the fourth semester in order to reach the upper division elective courses, thestudents were considerably more advanced in their Spanish abilities than fourth semester students. Thisgroup of Spanish majors and minors would be considered to be at an advanced level of university-levellanguage study. All of the native English-speaking subjects were born and raised in the midwesternUnited States. None have parents who speak a language other than English at home, nor do any of thesubjects speak any language other than English and Spanish. Each of the subjects began studyingSpanish after adolescence, beginning their language study either in high school or at the universitylevel.The five native Spanish-speaking subjects represent both Peninsular and American varieties of Spanish, and are pursuing graduate study in Spanish in the United States. As discussed above, thevoiced alveolar tap [  ] and the voiced alveolar trill [ r ] are the target rhotics for which AmericanEnglish-speaking learners of Spanish aim, yet the trill does not exist in all varieties of Spanish.Therefore, in order to make a valid comparison with native speakers, the native speakers were selectedfrom varieties of Spanish that do make use of the voiced alveolar trill.Subjects were recorded reading a short story of approximately 1750 words in Spanish. The storyoffers a text of considerable length with a developing plot so that subjects could not focus closely ontheir pronunciation as they might in a word list where all attention can go to a single word. While thereading of the story cannot be considered to be identical to spontaneous speech, we know that there isa continuum of speech style that can be affected by the task that subjects perform and the amount of attention the task allows them to pay to form (e.g., Major 2001, Tarone 1979, 1983). The story provides a less guarded style, nearer on the style continuum to the style of spontaneously occurringspeech than to that of closely guarded speech, while still providing the target sounds in identicalcontexts for all subjects, and thus assuring that differences between subjects are not due to contextualdifferences in which the tokens occurred. The recording sessions took place in a digital languagelaboratory, with the recordings saved as .wav files. Subjects were told that they were part of a studyexamining how speakers of American English learn Spanish by comparing the Spanish of learners atdifferent levels of language study with each other and with that of native speakers of Spanish. Whileno particular mention was made of pronunciation, it is certainly possible that some subjects wereaware that their pronunciation was the target of investigation since they were reading a story rather than creating language on their own. Nonetheless, it was felt that this realization would be offset by thetask, since students could not focus heavily on form. 50  Ten occurrences in the story of each of the two target rhotics (i.e., the voiced alveolar tap and thevoiced alveolar trill) in intervocalic position were selected, and production of these same target rhoticswas analyzed in each subject’s reading of the story. Cognates were avoided to the extent possible inorder to avoid enhancing transfer of the American English rhotic to pronunciation of the Spanishrhotics beyond the extent that this would occur in words that are not cognates. 2 Words containing thesounds to be analyzed were isolated and examined both aurally and acoustically through a waveformand/or a spectrogram within the PCquirer computer program for acoustic analysis. Tokens where amispronunciation led to the rhotic not being intervocalic were excluded since the present studyinvestigates only the pronunciation of the target rhotics in intervocalic position. In the analysis it wasnoted whether the sound produced was an accurate production (i.e., successful production of the targettap or trill, depending on the specific case) and, when not, what non-target sound was actually produced. A successful tap was considered to be produced when a brief closure was evidenced in thewaveform or spectrogram as a result of the tongue tapping the alveolar ridge. A successful trill wasconsidered to be produced when voiced airflow was interrupted by a series of brief obstructionsresulting from the vibrating tongue approaching the alveolar ridge multiple times. It was fairlycommon for the tongue to create an alveolar obstruction but not a full closure on one or more of itsapproaches to the alveolar ridge during a trill (though in all but one case there was at least onecomplete closure as well). These cases were counted as trills as the tongue did trill in creating themultiple alveolar obstructions and results in a canonical-sounding production of the trill as determinedthrough consultation with three native speakers. Such lack of closure was not considered a target-like production of the tap, however, since the tongue did not tap the alveolar ridge and did not result in acanonical-sounding production of the tap. 4. Results Table 1 reports the accuracy for achieving each target rhotic by subjects at each of the three levels(i.e., the two learner levels and the native speakers). As can be seen in the table, the native speakers produced the target tap 92% of the time and the target trill 86% of the time. 3 While the more advancedlearner group produced the tap with fairly high accuracy (78.7%), the same is not the case for the trill.Overall, there is a highly significant difference in accuracy across levels for both the target tap ( χ  2 (2) =60.53,  p < .0001) and the target trill ( χ  2 (2) = 138.01, p < .0001). Furthermore, pairwise comparisons between the native speakers and the advanced learner group shows a statistically significant differencefor both the target tap ( χ  2 (1) = 5.44, p < .05) and the target trill ( χ  2 (1) = 61.48, p < .0001). For thelearners, there is considerably higher accuracy for both the tap and the trill at the more advanced level.The differences by level are statistically significant for both the target tap ( χ  2 (1) = 41.02, p < .0001)and the target trill ( χ  2 (1) = 37.18, p < .0001). The tap was produced by 14 of the 20 fourth semester learners in the present study, and 13 of those produced it accurately at least 50% of the time. The tapwas produced by all 21 advanced subjects, and 20 of those produced it accurately at least 50% of thetime. The trill, on the other hand, as might be inferred from the combined numbers in Table 1, was produced by fewer learners and at a lower rate of accuracy when it was produced. The trill was produced by only 6 of the 20 fourth semester learners, and 4 of those produced only one trill each.While the trill was produced by 17 of the 21 advanced learners, only 4 of those produced it accuratelyat least 50% of the time. This is in stark contrast to the native speakers, where 4 of the 5 subjects produced the trill at least 80% of the time, and the fifth subject produced the trill 60% of the time. Avery small percentage of the more advanced learners, then, even approach the frequency with whichthe native speakers produce the trill.Table 2 presents the non-target productions for the target tap by level. Only four productions of the target tap were not produced as taps by native speakers. In each of these four productions an 2 In order to include ten intervocalic instances of target trills, some cognates were included among the wordscontaining the ten target trills. 3 Recall from Section 1 and the references cited there that native speakers produce sounds other than the tap andtrill, though these are the most common productions. As can be seen by the percentages here, the native speakersin this study produced the target sounds in the vast majority of cases.   51
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks