The Development and Validation of the Portuguese Speaking Test

Center for Applied Linguistics, Washington, DC

Center for Applied Linguistics, Washington, DC

Georgetown University

Foreign Service Institute

U. S. Government Language School

University of Pennsylvania

During the past decade, thanks primarily to the efforts of the American Council on the Teaching of Foreign Languages (ACTFL), like Educational Testing Service (ETS), and with the assistance of several government language training agencies under the auspices of the Interagency Language Roundtable (ILR), a major theoretical and practical development in the field of foreign language assessment has taken place. This development is the application of a «proficiency» orientation in the testing of foreign language competence. This movement has also brought about the adoption of a proficiency approach to language instruction in many quarters.

Two standards lie at the heart of the oral proficiency testing movement: the Oral Proficiency Interview (OPI) and the ACTFL/ILR speaking proficiency guidelines. The OPI is a direct face-to-face evaluation of the learner's second language competence conducted by trained interviewers and raters. [For information on the direct oral proficiency interview used by the government the reader is referred to Wilds (1975) and Sollenberger (1978). For information on the ACTFL oral interview see Higgs (1984); and Stansfield and Harmon (1987)]. The ACTFL proficiency guidelines (ACTFL 1986), based on the earlier developed ILR guidelines, provide the criteria against which language proficiency is rated. Between 1980 and 1988, ACTFL trained approximately 1,600 oral proficiency interviewers and raters in the major foreign languages taught in the United States: Spanish, French; German, and Russian. As a result, oral proficiency testing is widely available to examinees in need of a rating of their competency in these languages. However for less commonly taught languages such as Portuguese, the scarce number of trained interviewers makes it difficult if not impractical to connect an interviewer in one part of the country with an examinee in another part of the country. This situation impedes the availability of an oral proficiency interview to an individual who may have need for such a rating. The Portuguese Speaking Test (PST), a semi-direct test of oral proficiency, was developed in response to this need.

Semi-direct testing (using recorded and printed stimuli and recording examinee responses) is the most efficient and feasible approach to proficiency measurement in the less commonly taught languages. This approach eliminates the need to sustain a costly and labor intensive face-to-face (direct) Oral Proficiency Interview program for low-volume languages whose enrollment figures may be unstable from year to year. However, semidirect testing does provide the benefits derived from a continual assessment program, and it can serve as the impetus for competency-based learning on the part of students of these languages.

There have been several efforts to develop semi-direct tests of oral proficiency in foreign languages. One of the earliest was the Recorded Oral Proficiency Examination (ROPE), designed and described by Lowe and Clifford (1980). In the ROPE, the examines hears a series of tape-recorded questions in the target

––––––––   642   ––––––––

language and answers them orally during timed pauses following the questions. There are no written or visual stimuli in the ROPE. The range of the ILR proficiency levels covered by the ROPE is 0 + to 5.

Another semi-direct test is the Test of Spoken English (TSE), published by the Educational Testing Service (ETS). The TSE was developed by Clark and Swinton (1979) and is most widely used by universities to determine whether a foreign student applicant has sufficient command of spoken English to serve as a teaching assistant in his or her field while pursuing an advanced degree. Using a picture booklet and 20 tape-recorded questions, the TSE elicits about a 7-8 minute sample of speech. Examinees' response tapes are returned to ETS where they are scored on a scale of 0-300 by trained raters. The results of the test are sent to institutions designated by the examinee. In a validation study, the TSE was found to be an adequate substitute for a face-to-face speaking test, since it showed a correlation of .79 with an oral proficiency interview (Clark and Swinton 1980).

In 1986, the Center for Applied Linguistics (CAL) conducted a project with support provided by a grant from the Department of Education to develop a proto-typical ROPE-like test for Chinese that could be measured on the ACTFL/ILR guidelines yet use visual as well as aural stimuli (as the TSE) [Clark 1986; Clark and Li 1986]. The Chinese Speaking Test (CST) differed from the former tests in that it used a format similar to the OPI. However, following a «warm-up» in the target language, the stimuli are presented in English. This is because the questions following the warm-up can be quite lengthy and complex One problem with the ROPE, which presented stimuli in the target language, was that not all examinees could understand the test questions. In addition, the CST was specifically designed for examinees at a targeted proficiency range of Intermediate-Low to Superior on the ACTFL scale (1-3+ on the ILR scale). Four parallel forms of the CST were developed. Using a representative student sample and ACTFL-certified interviewers in Chinese, a validation study was conducted to statistically compare the OPI with the CST. Using a sample of 32 subjects, correlations of between .96 and .98 were found between student scores on the four CST forms and the OPI.

The Portuguese Speaking Test was developed under a second grant from the Department of Education. The grant also includes funds for the development of similar tests of Hausa, Hebrew and Indonesian in 1989. The Test Development Committee consisted of Charles W. Stansfield (CAL), who wrote the project proposal and served as Principal Investigator, Dorry Kenyon, Test Development Specialist (CAL), and three experienced instructors of Portuguese with training in using the ILR and/or ACTFL proficiency rating scales: Dr. Ricardo Paiva (Georgetown University), Ms. Fatima Doyle (Foreign Service Institute Language School) and Ms. Inés Ulsh (United States Government Language School).

After reviewing earlier semi-direct tests, the Portuguese-speaking members of the committee developed the specific items for the PST, following the item types used on the CST (Clark 1986). After each item was written, it was critiqued by the other members of the Test Development Committee. Items were then either discarded or revised repeatedly until no flaws could be identified. A professional artist was contracted to work with the Committee in the development of appropriate visual stimuli for the test. Audio stimuli for the Brazilian and Lusitanian versions were recorded in a professional recording studio. When all visual and audio stimuli were ready, three forms of the test were assembled and printed for trialing.

The following section identifies each part of the test and gives a brief description of each item type:

This section corresponds to the «warm-up» section of the direct interview. In this section, the examinee listens to conversational questions about his/her family, education, hobbies, in Portuguese and responds to each question as it is asked. There are 12 to 13 such questions on each form. This is the only section in which Portuguese is used on the tape.

Below are some examples (though not actual test questions) of the type of question found in this part of the test.

For each of the following question types, the examinee is given between 15 and 30 seconds to prepare an answer before being required to speak. Time for giving an answer ranges from 45 seconds to 1 minute and 45 seconds.

The examinee is shown a pictorial map in the test booklet and is instructed to give directions between two points on the map. The question is contextualized in that examinees are instructed that they are giving this information to a friend over the phone.

The examinee is shown a drawing in the test booklet and is instructed to describe the picture in as much detail as possible. Each picture contains a variety of objects and actions. This question is contextualized so that the examinee knows the specific audience being addressed and the purpose of the description.

The examinee is instructed to speak in a narrative fashion about a sequence of four or five pictures shown in the test booklet. There are three sequences of this type; one each for past, present and future time narration. The questions are contextualized so that the examinee is given a specific audience and a specific reason for the narration.

The examinee is instructed to talk about selected topics involving different discourse strategies. These strategies include explaining a process, supporting an opinion and talking about a hypothetical situation. There are five such topics, each printed in the test book let.

Below are some examples (though not actual test questions) of this type of item.

The examinee reads a printed description of a real-life situation in which a specified audience and communicative task are identified. The examinee is then instructed to carry out the specified task. The tasks range from making simple requests to giving a formal toast.

Below are some examples (though not actual test items) of this type of item.

Before beginning the test, the examinee listens to general directions on the test tape. These directions are also printed on the front cover of the test booklet.

In order to ensure that the questions were clear, understandable and working as intended and to check the appropriateness of the pause times allotted on the tape for examinee responses,

––––––––   644   ––––––––

the three preliminary forms of the PST, each containing more items than would ultimately be required, were trialed on 15 learners of Portuguese at Georgetown University (1 form each) as well as on 2 native speakers of Portuguese (1 Brazilian, 1 Lusitanian)¹⁰³. Nine of the learners took the test in a language lab setting, the others were administered the test on an individual basis using two tape recorders. Whenever the test was individually administered, one of the authors observed the examinee take the test and recorded notes. The observer paid particular attention to the adequacy of pause times, undue signs of nervousness in the examinee, and examinee blocking on a question; i. e., not responding at the appropriate time. Upon finishing the test, all examinees completed a detailed written questionnaire about the test and when possible, they were also questioned by the observer as to their reactions to the test. The speech samples elicited by the questions were evaluated by the Portuguese-speaking members of the Test Development Committee and an ACTFL rating was assigned to each examinee tape by Ms. Ulsh, an ACTFL-certified water and interviewer trainer.

As a result of the feedback obtained during the trialing, it was discovered that only minor modifications in the questions were necessary; in most cases this involved clarification or simplification of ambiguous items in the pictures. The original pause times were generally confirmed, though they were reduced in the opening conversation section and lengthened or shortened for some of the topics and situations. More time was given to prepare and answer the questions involving supported opinion.

In response to student comments obtained on a questionnaire completed after the trialing, instructions were added to, the general test directions explaining that, although the location of questions varied between Brazil and Portugal, the examinee would only need to answer in the type of Portuguese he/she normally uses. In addition, it was decided to provide two different versions of the tape (one for Brazilian and one for Lusitanian Portuguese) in the warm-up conversation section. This decision did not alter the tests significantly, although the wording of the scripts for part one varied slightly. The, questions and test booklets remained the same for both versions. Also based on student comments, some items were chosen for deletion from the final forms.

In its final version, the PST lasts approximately 43 minutes and elicits a speech sample of approximately 22 minutes.

A research study was designed and carried out to validate each of the three forms of the test. The study sought to answer the following questions:

1. Can this test, which involves spoken responses in Portuguese, be scored reliably by different raters?

2. Are the three separate forms of the PST interchangeable, i. e., do they produce similar examinee results independently of the particular form administered?

3. Do the recorded responses produce the same score as a regular five interview for any given examinee?

To answer these questions, a research de sign was prepared involving 80 subjects. The subjects were predominantly undergraduate students who had completed a year or more of Portuguese study in the Department of Spanish and Portuguese at the University of Maryland, College Park, and current and former students of the Portuguese Department at Georgetown University. One subject was a CAL employee and another worked at the Foreign Service Institute. Each subject received an honorarium of $20.00 for participating in the study¹⁰⁴.

Each subject was administered two forms of the PST and an oral proficiency interview. The research design controlled for any advantages or any effects on student scores resulting from the order of administration of the tests, with half of the subjects receiving the oral interview first and half the taped test first. The third test for all subjects was a taped test. The design also controlled for the order of the PST tests in that 10 subjects took Forms A and B, 9 subjects took A and C, and 11 subjects took B and C. Within each group half of the subjects took one form first and then the other form while the other half took the PST forms in the reverse order. The design also attempted to control for proficiency level. Based on teachers' observations, subjects were assigned into a «high» group and a «low» group. Half of each of the forms A-B, A-C, and B-C groups were selected from the high group, and half from the low group.

Two raters were used in the validation study. They were Fátima Doyle (rater 1), an ILR-certified rater from the Foreign Service Institute and María Antonia Cowles (rate 2), ACTFL-certified rater from the University of Pennsylvania. Ms. Doyle conducted 16 of the face-to-face interviews and Ms. Cowles conducted 14.

The majority of the subjects were tested at their respective university in one session, between 21/4 and 3 hours in duration, except in a few cases where conflicts in the subject's schedule made this impossible. In such instances, testing was done on contiguous days.

The live interview was tape recorded. Upon completing the live interview; the interviewer assigned a rating to the subject. After all, interviews were completed, the recorded Rape of the five interview was given a second, independent rating by the other water. Each rater also rated each of the recorded PST test tapes, anonymously and in random order. All the ratings were completed during a period of six weeks, after which subjects were sent six test scores in the mail: the scores of each of the raters on the five interview and the scores of each of the raters on the two PST forms taken by each examinee.

Ratings on both the five interview and the tape-based semi-direct tests were based on a scale combining both ACTFL and ILR rating scales. For statistical analysis, these ratings were converted to a scale with weights as signed as follows:

ACTFL/ILR Level	Coded as:
Novice-Low	0.2
Novice-Mid	0.5
Novice-High	0.8
Intermediate-Low	1.0
Intermediate-Mid	1.5
Intermediate-High	1.8
Advanced	2.0
Advanced-Plus	2.8
Level 3	3.0
Level 3+	3.8
Level 4	4.0
Level 4+	4.8
Level 5	5.0

The system of score coding above is based on the ILR 0 to 5 rating scale and is intended to assign an appropriate numerical value to the proficiency level descriptions. For example, proficiency at an Advanced-Plus level is characterized by many of the same features as at the Superior/3 level, though the examinee cannot sustain the performance. Thus, the numerical interpretation falls closer to 3.0 than mid-way between the two, as may be expected.

Table 1 shows the mean score, standard deviation and other basic statistics for the ratings assigned by each of the two raters to subject performances on each of the semi-direct test forms and on the live interview.

In answer to research question 1 (Can the test be scored reliably?), it was found that different raters could score the test with a

high degree of reliability. Interrater reliabilities (Pearson product-moment correlations) between the ratings assigned by Rater 1 and those assigned by Rater 2 for each of the semi-direct test forms and for the five interview are shown in Table 2 below.

These interrater reliabilities are all uniformly high across the three test forms and the five interview. Interrater reliability was not adversely affected by the semi-direct test format as opposed to the OPI. This suggests that the PST elicits as ratable a sample of speech as the five interview.

In response to research question 2 (Are the forms interchangeable?), the three separate forms of the PST were found to produce similar examinee results. Parallel-form reliabilities for the same subject taking two different test forms, with the same rater scoring both forms, are shown in Table 3.

The statistics indicate that the parallel form reliability of the PST is very high. With the first rater, the parallel-form reliability was nearly perfect .99. With a different rater, rater 2, the parallel-form reliability was also a very high .94. Such favorable statistics provide strong support for the proposition that each form of the PST elicits a sample of speech that is uniformly challenging to the examinee. The fact that the parallel-form reliability was high for two different raters supports the claim that the sample of speech elicited by different forms is equally ratable.

Table 4 shows parallel-form reliabilities for subjects taking two different test forms, with each form scored by a different rater.

This type of parallel-form reliability takes into account measurement error that can be attributed to natural variation in examinee speech, error that can be attributed to differences in test form, and error that can be attributed to differences in raters. Thus, it may be viewed as a lower-bound, conservative estimate of the reliability of PST scores. Again the reliabilities here are all high, and the average reliability under these severe conditions (different forms and different raters) is .95.

To answer the third research question (Does the PST produce the same rating as the oral interview?), correlations of semi-direct test scores with the five face-to-face interview were calculated. These are given in Table 5 below. These correlations are evidence of the validity of the PST as a surrogate five interview. (Note that in pairing the two ratings, each examinee's interview rating was used twice to correspond to each of that examinee's two PST form ratings.)

Again, the correlations are all high. The average correlation based on 120 pairs of ratings (30 subjects x 2 PST forms x 2 ratings, correlated with the rating assigned for the five interview) was .93. Such results support the claim that the PST is a valid measure of oral language proficiency that can be substituted for a five interview.

In addition to analyzing examinee test scores, we also compared the two testing formats (semi-direct versus live interview) through an analysis of feedback from the examinees themselves. To do this, a questionnaire containing nine questions about both for mats was given each examinee after the last taped test. Examinees had to mark choices

––––––––   647   ––––––––

on the questionnaire; they were provided with space for further written comments in response to each question, if desired. The results of the questionnaire are represented graphically in nine figures in the appendix and are referred to in the text below. (For more information on the results of the questionnaire, see Stansfield and Kenyon 1988). All 30 subjects in the validation study completed the questionnaire for a 100% participation rate.

The questionnaire revealed that the same percentage of students (73%) felt that their maximum level of Portuguese speaking ability had been probed by each test (Figures 1 and 2). This suggests that for the most part the students felt their speaking ability was being adequately tested and they found no difference in the ability of the two test formats to test the depth and thoroughness of their present Portuguese speaking ability. None of the students felt there were unfair questions in the five interview (Figure 3); however, for the taped test, 5 students (17%) felt there were unfair questions on the PST (Figure 4). This may be due to the fact that some of the students tested were beneath the recommended ability level of the test (i. e., their level was discovered to be 0+ on the ILR scale, Novice on the ACTFL scale). For such low proficiency students, the five interviewer can adapt the interview or cut it short, whereas in the taped test, the weaker examinee is asked every question. In any case, only a very low percentage of students felt there were «unfair» questions on the taped test.

Students were also asked in which test they felt more nervous. It was expected that a large percentage of the students would feel more nervous in the taped test, since that mode of testing speaking ability is more unfamiliar and perhaps «unnatural». Indeed, about 70% of the students did feel more nervous in that mode, while over 30% of the students felt more nervous or the same amount of nervousness in the five interview (Figure 5).

Another question focused on perceived difficulty. Despite the fact that students did approximately the same on both tests (see correlations above), the vast majority (90%) perceived the taped test as more difficult (Figure 6). Perhaps some of the individual comments are enlightening; these seem to revolve around the timed pauses, the length of the test (for students who were not at the recommended level), and discomfort in talking to a machine. It appears the «unnatural» format contributed heavily to perceived difficulty.

Regarding the technical qualities of the taped test, students were asked whether the pauses were too long, too short or about right. The majority (70%) of the students had no problem with the timed pauses in general (Figure 7). From the individual comments and the fact that 3 students marked in more than one category in response to this question, pause time was in general an individual concern, varying from person to person and from question to question. Given the individual variation mentioned above, in a taped test with timed pauses such as the PST, the fact that 70 per cent of the students felt the pauses were about right is significant. It is also significant that 100% of the students felt the PST directions were clear (Figure 8). This is gratifying as there is no possibility in the taped-test mode for examinees to ask questions if something is misunderstood once Part One of the test is begun.

The last question served as a «catch-all» summary question. It asked the students which type of test they preferred taking. The majority (86%) choose the five interview (Figure 9). From the comments, it can be seen that this is probably a reflection on the five interview testing mode, which seemed more natural, rather than a reflection on the technical quality of the taped test. However, 14% of the students either preferred the taped test or had no preference, and two comments suggest that the taped test was more interesting and challenging than the five interview.

In summary, the responses of the examinees indicate that though they were very positive about the content, technical quality and ability of the taped test to probe their speaking ability, the unfamiliar mode of testing and perceived «unnaturalness» of speaking to a machine caused a greater perceived difficulty and more nervousness than the five interview. Thus, the majority of the students said they preferred the five interview to the taped test. Nevertheless, given the extremely high correlations between the two types of tests and the positive response to the taped test quality it appears that the taped test may confidently be used as an alternative, albeit «second choice» in the examinee's eyes, to oral proficiency testing via the five interview. Moreover, it is expected that future examinees who

––––––––   648   ––––––––

are more prepared for the test through the Examinee Handbook which is distributed to all examinees registered for the test may find the testing mode less threatening than the students participating in the validation study who went to both the live interview and the taped tests without any special advance preparation in order to avoid any biasing.

The type of test instrument discussed in this article can be called a simulated oral proficiency interview (SOPI). While such an instrument is a semi-direct test, the SOPI differs to some degree from other semi-direct tests (such as the TSE and the ROPE) that have been developed, in that the SOPI follows very carefully the OPI format. The SOPI used here begins with a warm-up, then introduces more complex functions such as giving directions and giving a detailed description. It then tests narration in the present, past, and future. Next, it tests the ability to speak on a variety of topics involving different content, and finally, it tests the ability to handle several interactive situations through simulated role playing tasks. Following the warm-up conversation, in Part 1, the difficulty of tasks increases within each part. Questions at the end of each part are typically more difficult than the questions at the beginning of the next part. Therefore, the SOPI also follows the OPI format in that it moves from level check to probe several times during the assessment. Although no «wind-down» (ending the live interview with each questions or praise for the examinee's performance) was used here, there is no reason why a wind-down should not be incorporated in future forms of the PST. Indeed, the raters who participated in the PST validation study made this suggestion, which will be implemented in the development of other SOPI tests.

Extrapolating from our experience in developing the PST and the results obtained in this validation study, it is interesting to speculate on the possible advantages of the PST (and other SOPI test instruments that might be developed with equal care in other languages) over a five interview. This study found that the PST produces scores that are comparable to those obtained on the five interview. The scores were based on speech samples that are equally ratable by interviewers; that is, the speech samples were about equally useful to the interviewers in helping them as sign a rating within the Novice to Superior ACTFL levels. Thus;. there is evidence that the two formats (SOPI and OPI) produce results that are commensurate, both quantitatively and qualitatively.

However, under certain circumstances the PST may be preferable, even when an interviewer is available. One such situation may be when accuracy of rating is a concern. The PST is consistent from form to form and the raters used in the PST operational program¹⁰⁶ are highly reliable. Thus, when the quality of a rater is a major concern, such as when an important decision will be made based on the oral proficiency rating assigned an examinee, it may be preferable to use the PST.

Another situation when the PST and similar tests in other languages maybe preferable is when many examinees need to be tested within a short span of time. For instance; if only an hour were available for testing a group of students; the PST could be administered to all examinees simultaneously in a language laboratory and the examinee response tapes scored subsequently by raters.

The PST and similar SOPI tests in other languages may offer psychometric advantages as well. It is possible that the PST produces results that are more consistent than those obtained through a live interview; even when expert interviewers are used. This is because the PST forms were carefully developed and trialed; thereby ensuring that the questions are of equivalent difficulty. On the other hand, live interviews are always different in content, even when expert interviewers are used.

The average correlation between a live interview and a -PST form (.93) is about equal to the correlation between a single interview scored by two different raters (.94). It would be interesting to determine the correlation between two different interviews scored by different raters and compare it to the average correlation between two different PST forms scored by different raters (.95).

Although this study included only one live interview (which was rated twice), we suspect that the very consistent results obtained here on different forms of the PST might not have been duplicated for two different five interviews. Differences in examinee performance could find their origin in the many factors related to a possible «interviewer effect», such as variations in the degree, of nervousness on

––––––––   649   ––––––––

the part of the examinee or different interpersonal relations between the examinee and the interviewer. This causes us to speculate that the semi-direct SOPI format used on the PST might ordinarily produce more consistent results than those that would ordinarily be produced by the Oral Proficiency Interview. Further research focusing specifically on this issue is in order.

In conclusion, on the basis of the experience of the Portuguese Speaking Test, simulated oral proficiency interviews employing a semi-direct method of test administration appear to be an appropriate and workable alternative to the Oral Proficiency Interview, especially in cases where using the OPI is economically or essentially unfeasible, as in many of the less commonly taught languages.

American Council on the Teaching of Foreign Languages. 1986. ACTFL Proficiency Guidelines. Hastings-on-Hudson, NY American Council on the Teaching of Foreign Languages.

Clark, John L. D. 1986. Handbook for the Development of Tape-mediated, ACTFL/ILR Scale-based Tests for Speaking Proficiency in the Less Commonly Taught Languages. Washington, DC: Center for Applied Linguistics.

Clark, John L. D. and Ying-chi Li. 1986. Development, Validation, and Dissemination of a Proficiency-based Test for Speaking Ability in Chinese and an Associated Assessment Model for Other Less Commonly Taught Languages. Washington, DC: Center for Applied Linguistics. (Alexandria, VA: ERIC Document Reproduction Service No. ED 278 264)

Clark, John L. D. and Spencer Swinton. 1979. An Exploration of Speaking Proficiency Measures in the TOEFL Context. TOEFL Research Report 4. Princeton, NJ: Educational Testing Service.

_____.1980. The Test of Spoken English as a Measure of Communicative Ability in English-medium Instructional Settings. TOEFL Research Report 7. Princeton, NJ: Educational Testing Service.

Higgs, Theodore V., ed. 1984. Teaching for Proficiency, the Organizing Principal. Skokie, IL: National Text book Co.

Lowe, Pardee, Jr. and Ray T. Clifford. 1980. «Developing an Indirect Measure of Overall Oral Proficiency». Measuring Spoken Language Proficiency. Ed. James R. Frith. Washington, DC: Georgetown University Press. 31-39.

Sollenberger, Howard E. 1978. «Development and Cur rent Use of the FSI Oral Interview- Test». Direct Testing of Speaking Proficiency: Theory and Application. Ed. John L. D. Clark Princeton, NJ: Educational Testing Service. 1-12. (Alexandria, VA: ERIC Document Reproduction Service, No. ED 172 523).

Stansfield, Charles W. and Chip Harmon eds. 1987. ACTFL Proficiency Guidelines for the Less Commonly Taught Languages. Washington, DC: Center for Applied Linguistics and Hastings-on-Hudson, NY: American Council on the Teaching of Foreign Languages. (Alexandria, VA: ERIC Document Reproduction Service, Na. Ed 289 345)

Stansfield, Charles W. and Dorry M. Kenyon. 1988. Development of the Portuguese Speaking Test (Final report to the U. S. Department of Education). Washington, DC: Center for Applied Linguistics. Alexandria, VA: ERIC Document Reproduction Service No. ED 296 586.

Wilds, Claudia P. 1975. «The Oral Interview Test». Testing Language Proficiency. Eds. Randall Jones and Bernard Spolsky. Arlington, VA: Center for Applied Linguistics. 29-44.

Book Reviews

EDITORIAL POLICY: Publishers and authors are invited to submit books for review in Hispania; in general, journal numbers will not be reviewed. Hispania cannot accept unsolicited reviews nor honor requests to review specific books. Members of AATSP who wish to be considered as reviewers may send copies of curricula vitae to the Book Review Editor. Those assigned books for review will receive a stylesheet and a statement of editorial policy.

The seventy articles that comprise this collection of proceedings demonstrate the efforts of scholars -not exclusively academicians- in the province of Guadalajara (Spain) to come together for the exchanging of ideas concerning the historical and Cultural make-up of their region. Taking its name from the Henares River as the unifying geographical element of the province, the conference focused on events, people, and places in the cities of Guadalajara and Alcalá de Henares. The topics presented range in interest from archaeology, art, and biography to economics, history, and sociology.

While many of the articles offer engaging reading, the proceedings as a whole can only appeal to the American Hispanist who knows the area exceedingly well. A case in point concerns the articles that focus on the descriptions of specific artistic and archaeological findings. As a result, the unfamiliar and highly specialized material makes the volume of limited interest. Some of the articles, however, seem especially noteworthy because of their wider range.

Medievalists may find that «Sobre la conquista cristiana de Guadalajara y Sigüenza (de las tradiciones a la historia)» offers a bit of trivia related to the Cid: The author tells that legends abound concerning the Christian reconquest of these two cities. It appears that in the sixteenth century the powerful Mendoza family sought to link its name to that of the Cid by claiming that Alvar Fáñez played a decisive role in the reconquest of Guadalajara (1085). «Mosé Ben Sem Tob de León, autor del 'Zohar', ilustre vecino de Guadalajara en el siglo XIII». Perhaps the best-written article in the collection, concerns this Jewish mystic whose «Libro del Esplendor» influenced later Spanish writers, namely Santa Teresa and Unamuno.

Two articles, «Estudiantes alcalaínos y seguntinos en los arzobispados y obispados de Nueva España en la época de los Austrias (1535-1700)» and «Virreyes americanos de origen alcarreño donate el período de la casa de Austria», demonstrate the involvement of the region in American affairs. Of the 140 bishops and archbishops to be appointed to the New World, forty attended the region's two universities (Alcalá and Sigüenza); seven viceroys belonged to noble families from the area.

Cervantes, Alcalá's most famous native son, is the subject of two articles. Golden Age specialists will recognize the name Dominick Finello, whose «Cervantes y los dramas de pastores del siglo de oro» discusses the pastoral in Spain as a game in which literature became life-like and where the reading of a text led to the reenactment of that text in everyday life: Finello cites two examples from the Quixote and mentions several comedias in which Cervantes demonstrates this imitation. The article leaves the reader with a desire to explore further the question of the pastoral during the Golden Age: «Miguel de Cervantes, Antonio de Sosa y Africa» treats the theme of captivity, especially in the writings of Sosa, friend and companion of Cervantes in captivity. To fully understand and appreciate the state of captives in Africa during the sixteenth century one should consult Sosa's diálogos which would appear to be an excellent source book.

As one would expect, many articles deal with Cardinal Cisneros and 'his' university «La documentación pontífica de la Universidad Complutense en el período fundacional» traces the efforts of Cisneros from 1495 to obtain Papal Bulls to charter the university. The prelate's efforts were rewarded in 1512 with a Bull that served as the charter. «El municipio y la universidad de Alcalá de Henares: Dos instituciones tradicionalmente enfrentadas» points up the functions that occurred because academic, not municipal authorities held control over the civil and criminal discipline of the students. As a result, public functions were especially bother some because they precipitated great tensions between the two entities.

The most fanciful of the articles is entitled «Cristóbal Colón, alcarreño». It would appear that Columbus was one of a set of illegitimate twins born to Aldonza Mendoza, stepsister of the Marqués de Santillana: Evidently the other twin was murdered by family members and Columbus was given over to the family of his wet nurse. Columbus did not reveal the secret because he needed money and the help of his relatives. According to this article, the admiral lies buried by his mother's side in the church of Nuestra Señora de los Remedios in Cogolludo. What appears to be conjecture, and the lack of scholarly notes and documentation leads one to consider it imaginative.

While this volume offers some informative reading,

––––––––   654   ––––––––

as a whole it is far too specialized in scope for the use of most Hispanists.

David G. Burton

Ohio University

This anthology comprises excerpts from thirty five previously-published texts selected to recount from the Christian perspective the Moorish presence in Spain, 711-1150; two proposed companion volumes are to gather texts from the later centuries and those written in Arabic. In keeping with the goal of the Aris & Phillips Hispanic Classics series to make texts accessible to the layman and the student while remaining acceptable to the specialist, a parallel English translation faces the original document written in Latin, Old Spanish or Old French. The general Introduction high fights the thematic orientation and style of the selections, acknowledges editorial liberties with the published sources, and explains the use of proper names, coinage and the system of dates. Three maps delineate respectively the political division of the peninsula in about 1130, the frontiers of Christian and Muslim Spain from 914-1480 (after Lomax), and locate selected cities and rivers in the peninsula. A Booklist of nineteen entries includes general reference works written in Spanish and English. A brief introduction to each reading sketches pertinent historical and literary features and provides a citation for the immediate published source.

Counterbalancing Castro's view of the relatively peaceful coexistence of Christians and Moors, Smith's mosaic of texts from cultural history portrays enmity and hostility with some background shadowing of the medieval opponents' mutual affective regard. Smith laments the absence of contemporary Christian texts about the XII c. schools of translators in Toledo, the arts and craft products of al-Andalus, libraries and literacy, and social practices, and conjectures that the writers either «disdained» them or were more concerned «with the struggle of faiths and campaigns on a large scale...» (vi).

Raising his own doubts about which might be the better method, the distinguished British Hispanist has ordered the documents according to the date to which they refer rather than to their date of composition. This overriding concern with content can explain in part the editor's uneven treatment of source material and means of textual transmission. In 1, the reader meets Lucas de Tuy's Chronicon mundi through the Castilian translation used in Alphonso X's Estoria de España found in the Primera crónica general (= PCG). In 2, Rodrigo de Rada's De rebus Hispaniae appears in the Latin original; the second time that author is cited (5), it is through the amplified Alphonsine Castilian translation in the PCG; the third time (18), the text appears in the Latin original. The Toledano's Historia arabum is not included.

Because the editor proposes to present the reconquest from the Christian perspective, he feels the need to justify the inclusion of portions of the lost history of Ben Alcama (20, 22, 23) preserved through the intermediary of the PCG. In the introduction of 20, he appeals to the text's «great importance» (99); in 22, he argues unconvincingly for its inclusion because the Arabic original is «lost and more positively, that what he [Ben Alcama] wrote has been mediated through Castilian of a period only slightly later than his own» (109). One wonders if Ben Alcama also will be included in the companion volume dealing with the Arabic perspective.

Epic-legendary characters and the sources for our knowledge of them receive the editor's unique treatment. Pondering Ben Alcama's report of the Cid's speech to the Moors at Valencia, Smith suggests that the Campeador spoke in Arabic (115). He rejects the plausibility of a poetic source for the legend of the mora Zaida (21), although he is aware of the numerous factual liberties in the text and the Alphonsine filters. He posits a XIII c. prose source because the text «shows no obvious poetic traces in the form of remaining rhymes or turns of phrase» (104). The portion cited contains the prosification of a variant of the traditional, oral based epic formula «sonando... las nuevas» (cf. PMC 1154, 1156, 1206) -«sonando la su muy grande fama», «la su buena fama et del su buen prez que cresçie coda dia et sonaua mas»- as well as the formulaic commonplace of being love-struck «de oidas, que no de vista», well-studied by Menéndez Pidal, Gautier and others. Not surprisingly, given the editor's critical position, he rejects the epic origins of the Poema de Fernán González (13) and the traditional roots of the Cid epic (24). Student and lay users of the anthology should be alerted that the prefatory material represents that editor's distinctively individualistic views.

The edition contains few misprints; however, it consistently restores the ç in the PCG material. All the printed sources of the anthology's contents do not appear in the Booklist, nor are they always cited in complete form in the introduction to the relevant excerpt. This well-edited collection provides a wide panorama of significant, fascinating texts from the cultural and intellectual history of the early years of Moorish occupation of the Iberian peninsula. Although the original writers may have been removed from the events (by up to five centuries), and their texts may have been translated and re-translated, prosified and re-poeticized, their testimony to the ardor of the Reconquest has been delivered to a broad audience by this commendable collection of readings.

Nancy Joe Dyer

Texas A&M University

Con extraordinario acumen, nos brinda la emérita catedrática Ruth H. Webber una colección de ensayos preparados por expertos de enjundia a fin de entretejer un mosaico panorámico de un género vivido, amén de vívido, en la secular conciencia ibérica.

El trabajo discurre a lo largo de dos coordenadas: en el tiempo, evalúa la tradición romanceril hasta la actualidad; en el espacio, sigue el manantial poético desde la Península hasta otras regiones atadas a su cultura.

Vistos a vuelo de pájaro, los diez ensayos de este libro tienen mucho paño que cortar: Diego Catalán se ocupa por la «artesanía literaria» (1) y columbra la proyección sintagmática a través de los nexos paradigmáticos de unos romances ejemplares; Ana Valenciano reseña la problemática de las expediciones del Seminario Menéndez Pidal para enriquecer el repertorio vigente; Antonio Sánchez Romeralo estudia la difusión del género gracias a los pastores trashumantes; Suzanne Petersen se apoya en los medios electrónicos para el análisis geográfico y cronológico del mecanismo de reproducción y cambio de la narrativa poética enmendándole la plana al eternamente desconforme Daniel Devoto (77-79); Maximiano Trapero reseña el estado del romance dialogado en las Canarias; Manuel da Costa Fontes enfoca su estudio en las Azores, Portugal y entre los grupos emigrados al Canadá y a los Estados Unidos; Judith Seeger analiza las versiones brotadas en el medio rural de la costa brasileña; Mercedes Díaz Roig se encara a la tradición del romancero en México; Samuel Armistead y Joseph Silverman presentan la vertiente sefardita con riqueza de atisbos; Beatriz Mariscal de Rhett analiza la correlación entre las estructuras narrativas del romancero y su cambiante manera de captar la realidad.

Por ser obra de estudiosos especializados, este trabajo destila y condensa los resultados más recientes y certeros sobre las distintas facetas del tema. La esmerada traducción al inglés, casi siempre llevada a cabo por la Dra. Webber, no relega el libro simplemente al lector angloparlante, sino que ensancha su impacto entre cuantos se ocupan por un género poético que germina en las postrimerías de la Edad Media y se acerca ahora a su propia postrimería tras una confusa alianza de intereses hostiles a su rebrote originario. Se acerca el ocaso, es cierto; pero el ocaso de la tradición oral no deja de avivar la tradición literaria, añadiendo así una validez sui generis a este bien presentado tomo colectivo.

Carmelo Gariano

California State University, Northridge

Como el editor mismo confiesa (10), el presente estudio no pretende ser más que un estudio comparativo, o sea, un cotejo de un manuscrito anónimo del siglo XV (el propio Tratado) con varios otros documentos didácticos de la Alta Edad Media. El contenido del estudio de Ramírez será de menos interés para los filólogos y críticos literarios que para los que quieran estudiar las influencias políticas, históricas y sociológicas de aquella época.

En su Introducción, capítulo que abarca las páginas 9-83, Ramírez nos ofrece los frutos de sus extensas investigaciones con respecto al manuscrito conservado en El Escorial. En la sección que trata de la descripción física del documento, Ramírez apunta las muchas y variadas lagunas que ha descubierto en el texto. A raíz de estas lagunas Ramírez colige que el Tratado conservado no es el original sino una copia de alguna versión incompleta anterior. Felizmente, los críticos modernos disponen de otro manual para príncipes del siglo anterior, la Glosa castellana al «Regimento de príncipes» (ca. 1350) de Juan García de Castrojeriz, documento que había sido el patrón del Tratado y que ahora sirve de buena guía para la restauración del contenido perdido en dichas lagunas.

Uno de los móviles principales de este estudio es el de refutar y corregir la edición del Tratado que preparó el padre Bonifacio Difernan en 1962. Ramírez cuenta hasta 45 graves errores de transcripción en la obra de Difernan, sobre todo en la ortografía (rregular por reglar), palabras equívocas (tísicos por físicos; la mas por jamás), y varios casos de palabras o frases faltantes (15-17). El editor también rechaza la fecha de composición que ha propuesto Difernan para el Tratado (a fines del siglo XII o el siglo XIII). Ramírez, basando su fallo en las muchas y obvias semejanzas con la Glosa de Castrojeriz, considera «patentemente absurda» la fecha dada por Difernan (17).

Los capítulos del Tratado de más interés son las secciones dedicadas a las mujeres y a los tribunales de justicia en el reino. Tres distintos capítulos tratan de la enseñanza de la mujer (LXVIII), de las viudas (XCVI) y de la virginidad (XCVII). La actitud conservadora, casi monástica, del Tratado se ve en el papel tan limitado que tiene la mujer en estos capítulos. Nunca se refiere a los dotes intelectuales de la mujer en el Tratado. Se le niega toda posibilidad de influir en los hombres; queda suprimido el papel de consejera al marido que señala la Glosa.

El doble asunto de las leyes y la justicia vuelve a aparecer en muchos capítulos donde la justicia se alaba abstractamente o donde se aplican casos concretos a la práctica de uno u otro oficial real. En cuanto a los buenos consejos ofrecidos en el Tratado, Ramírez cita varios ejemplos paralelos en otros tratados medievales como el Libro de cien capítulos, las Flores de filosofía y las célebres Siete Partidas de Alfonso el Sabio.

Además de la transcripción modernizada -gracias a la ayuda de un microfilm- del texto, lo más valioso del estudio de Ramírez es el inventario de asuntos y ejemplos paralelos que aparecen en el Tratado, en la Glosa, y en varias otras obras didácticas clásicas o medievales, incluso la Biblia (66-71). También vale señalar el cotejo de ejemplos que parecen haber sido tomados directa o indirectamente de la Glosa de Castrojeriz (72-77). Otros elementos utilísimos son la bibliografía de obras citadas y consultadas, el glosario, y el índice onomástico.

A pesar de la mucha minuciosidad con que se ha estudiado el Tratado y la fuerza y claridad con que Ramírez ha presentado sus argumentos, me parece difícil que la presente obra les despierte el interés a los hispanistas de hoy. Vale preguntarle al editor: ¿A qué público se dirige este estudio? Es un tomo que, como un débil eco de la más conocida Glosa de Castrojeriz, no aumenta mucho nuestros conocimientos de la Edad Media. Alguno que otro especialista en la época de los Reyes Católicos, quizá, querrá pedir un ejemplar de este estudio para la colección de textos medievales de su biblioteca universitaria. Fuera de esos casos especiales, es francamente dudoso que muchos otros vayan a añadir este tomo, por excelente que sea, a su propio inventario.

E. T Aylward

University of South Carolina

James R. Stamm's study presents an insightful, methodical reading of La Celestina. As the title suggests, the work is approached through its structure, examined progressively in four chapters focusing respectively on «El marco textual» (such peripheral material as the title, «Carta», «Prólogo», verses, íncipit, and «Argumento»); «El Auto o fragmento» (the first act); «La Comedia: Un arte de continuación» (Rojas's ostensible continuation of the Auto); and «La Tragicomedia» (the further continuation, the material interpolated throughout the work and from, acts 16, through 21). Each of these structures is considered through an analysis of its individual components or acts, in the order of their appearance. A final chapter of «Conclusiones» summarizes the current status of critical inquiry concerning the inviting but thorny questions of the author(s)' identity, possible interpretative «keys» to the work, and the work's unity, and further develops conclusions that have arisen in the course of the study.

Basic to the critical approach is an inward focus, a concentration on the text itself, details of plot and language, relationships among characters, internal consistencies and inconsistencies, specific content of individual speeches, the inherent thematic and dramatic internal structuring of the acts, and what all of these reveal about the work. It is only as a result of such analyses that externals such as relationships with cancionero poetry, the novela sentimental, or Petrarchan influences arise. Neither does Stamm allow himself to be lured into the blind alleys of such past critical disputations as the supposed converso status of the work's author or characters, although the largely bibliographical notes give due recognition to such issues. Especially sensitive is the detection of linguistic echoes throughout the work, the recurrence of images, phrases, metaphors, and their themes, as one speech recalls or prefigures another.

Based on the internal evidence that he brings to light, Stamm reaches some sound conclusions concerning the work's authorship. While linguistic and thematic echoes between the Auto and the Comedia create an artistic unit