Methods That Work (ตอนที่ 2)

(ตอนที่ 2)

CHAPTER 13
Immersion Teachers: What Can We Learn from Them?
Ann Masters Salomone
EDITOR'S INTRODUCTION
A
ccording to Swain and Lapkin (1989), in their thorough review of Canadian immer¬sion experiments, the integration of content and language is essential, and the "prin¬ciples should generalize to any second language instructional setting." Salomone, in this chapter, takes this generalization seriously. This will come as no surprise to well-informed language teachers, but they may take heart from seeing that what immersion teachers often do intuitively is apt to involve the very methods that are known to be gen¬erally applicable in language teaching. Another outcome of the Canadian research, one that Salomone does not stress but which will also be heartening to many foreign lan¬guage teachers who work with subjects beyond the early grades—a fact noted by Swain and Lapkin (1989)—is the finding that adults quite generally have the advantage over children in acquiring a new language. Older children and adults simply have more cogni¬tive muscle to apply to the task and generally excel except with respect to the mastery of subtle aspects of surface form such as the phonology and morphology of the new lan¬guage. Literacy skills and related knowledge bases afford older students a marked advan¬tage over younger children. Owing to what is known of the interrelatedness of language skills and semiotic capacities in general (Cummins, 1981,1983, 1984; Oiler, 1979; Oiler et al., 1991), it ought not to be surprising that more-mature learners are relatively better off with respect to transferable elements of their skills and knowledge than less-mature ones are. Another possible surprise from the Canadian research, and one noted here by Salomone, is that immersion teachers do emphasize surface grammar. According to Swain and Lapkin (1989), they overtly correct about 20% of student errors and also use some explicit teaching of "grammar," as Salomone also shows (in addition see Johnson, Chapter 25). Most importantly, however, Ann Masters Salomone details, perhaps for the first time, that methods that work elsewhere in literacy and language teaching are pre¬cisely the ones that immersion teachers typically employ.
nersion can then be viewed not only as con- For over 20 years, French immersion research
:-based language learning, but as a key exam- has concentrated on the achievement of immersion
of the integration of content and language students. In Canada, where French immersion has
:hing, the principles of which should generalize flourished since 1965, some researchers have
my second language instructional setting found that students attain near-native levels in sec-
.ain &c Lapkin, 1989). ond language receptive skills while maintaining
Chapter 13 / Immersion Teachers 129

we're going to do today"), and eventually
res:
s brun (brown bear)
- dinde multicolore (a multicolored turkey)
jattne orange (yellow) (orange)
fouge brun (red) (brown)
vert (green)
th
mat
she writes, however, she discusses in French tivity with the children, thereby encourag-ir second language production. "On va livre poisson rouge?" (""We're going to do dfish book?") They respond negatively and ■ that the book is "ours brun" ("brown which she then lists as number one. As the presentation of the programme, it is ting to note that teachers at Glenwood often suggest an incorrect response to -a correct one. This technique serves as a to elicit student speech while providing an )US second language model. n va commencer faction de graces. Qu'est-cest?" ("We're beginning Thanksgiving. s this?") Denise shows a ditto of a turkey olor names marked on its feathers and s her own question: "C'est une dinde. Cette >a avoir beaucoup, beaucoup de couleurs. le comment je I'appelle." ("It's a turkey, rkey will have many, many colors. Look at call her.") She writes "une dinde multicol-1 the easel, has the class repeat, and then .em to read with her as she points to the To illustrate number 3, Denise shows the Le math ditto they will complete; number 4 daily journal.
the more advanced third-grade level, uses restroom lining-up activities to expand nd reinforce the second language. First, she les filles qui portent un pantalon" ("girls I pants"), "les filles qui portent une robe" wearing dresses"), and "les filles qui por-:e jupe" ("girls wearing skirts"). Then she ie boys: "les garcons avec les yeux bleus" with blue eyes"), "les garcons qui portent ■zussures blanches" ("boys wearing white ), and "les garcons qui portent des pan-bleus" ("boys wearing blue pants").

Any second language teacher can expand lan¬guage instruction during daily "necessary" activi¬ties; textbook exercises should not be the only vehicle for language practice. Although high school/college classrooms have no circle time, daily warm-ups can be conducted in the L2. Presenting the programme is also possible in a traditional sec¬ond language classroom: The teacher can list the day's activities while discussing them in the L2 with the students. Even "lining up" activities occa¬sionally occur in secondary classrooms (for fire drills and special assemblies). Other tasks that can be conducted in the L2 include calling roll, assign¬ing homework, announcing future academic events, and directing classroom activities. Not using the language for these daily tasks implies to students that the second language is reserved for textbook exercises. Conducting these activities in the L2 signifies to students that the language is a tool for meaningful and useful communication.
Researchers assert that between 65% and 90% of all communication is nonverbal, with some esti¬mates exceeding 90% (Burgoon, Burgoon, &c Woodall, 1989; Mehrabian 6c Ferris, 1967). While attempting to communicate in a language that is unfamiliar to their students, teachers nearly auto¬matically resort to some form of nonverbal com¬munication. Because Glenwood immersion teachers often spend six hours per day communi¬cating in a non-familiar language, they use con¬crete materials and body movements extensively.
The need for concrete materials to reinforce and clarify early language learning is widely accepted: "Language environments rich in con¬crete referents appear to be a necessary environ¬mental characteristic for beginning second language learners" (Dulay, Burt, &: Krashen, 1982). At Glenwood School, for example, Pierre holds up a giant blue crayon to ensure students' comprehension of the sky's color; Denise uses large posterboard cutouts to illustrate various col¬ored animals during reading lessons; Nadine, a second-grade teacher, uses students' leaf collec¬tions as a basis for science and math lessons; and Estelle, a fifth-grade teacher, gives a body-parts quiz that demands that students draw their answers: "11 faut dessiner un menton. Il faut faire une image. N'ecris pas le mot." ("You must draw a chin. You must make a picture. Do not write the word.")

Chapter 13 / Immersion Teachers 131

Afraid of the Big Bad Wolf?" Pierre** irtners sing a song that he created usir T vocabulary:
peur du petit fantome? est pas nous, ce n'est pas nous!
peur du petit fantome? est pas nous du tout! - peur de la sorciere? (etc.)
peur du grand squelette? (etc.)
peur du gros chat noir? (etc.)
's afraid of the little ghost? Not us, not afraid of the little ghost? Not us at al': aid of the witch? [etc.] Who's afraid of eleton? [etc.] Who's afraid of the fc* [etc.]")
ding on the song and integrating lan-
idies and math, Pierre asks a student to
la sorciere. Camille points to the witches
:rom the ceiling, and the class counts to
-nch. Pierre then asks for the "gros chat
black cat"), the "petit fantome" ("little
and the "grand squelette" ("tali skele-
ich the children identify.
;h school/college second language classes,
ips more often in the language-for-spe-
)ses setting, thematic vocabulary can lend
to grammar-based lessons. According to
Novak, and Hanesian (1978), retention
d by learning vocabulary items as a
1 and active cognitive process that
the student's cognitive structure. By
related vocabulary items together and
n in varying contexts, teachers provide
sary concept-relatedness required for
tl learning.
nguage and Content Mohan (1986) second language learners can be peer [ six Glenwood immersion teachers capi-TIS concept. Marie has students teach one he hallway routine: She states, "On ." Some children answer "Devant." a les bras..." Children: "En bos." "On ..'' "Fermee." "On ne marcbe pas..." s canards, cornme les pingouins, comme ("We look..." "Ahead." "We have our "Down." "We have our mouths. . ." "We don't walk. . ." "Like ducks, like ike a horse.") In this ca.se, by reiterating

.' e hallway rules, students socialize one another in r' per Glenwood School behavior. Classroom regu-:-rions and subject-matter content are often trans-„" ;tted from child to child, and all six Glenwood 'jachers encourage students' sharing of information, "- either the first language or the second.
Individual peer teaching, in which one student -, jdresses the class, is also common at Glenwood: Vadine has students take turns leading the class in
■ ">ne while they point to the words on the black-
"-Dard. Estelle names one student as leader for
- ath activities: This student is responsible for
-rading the mathematical statements presented by
-"jree other students, two of whom hold poster-
-aard numbers and one of whom decides which
." jection to face his greater-than/less-than symbol.
Such involvement is also possible with older ■udents: Peer teaching is sometimes more effective
■ ian teacher-centered instruction. According to
" ca and Doughty (1985a), more L2 utterances
:cur in a small-group situation than in a teacher-
■■ onted classroom; and results of a study by Porter
986) show that grammatical errors are no more
- itnmon in student-to-student activities than in
. iose conducted by the teacher. Not only can
der students lead traditional classroom activities, uut they can also teach one another in pair or small-group situations.
Various organizational structures enhance stu¬dents' awareness of the daily lesson. According to Wong-Fillmore (1985), consistency in organization is a structural characteristic of successful L2 class¬es. Carefully articulating a daily routine, specifying sub-activities, and using boundary markers give the students an idea of what to expect, both lin-guistically and instructionally.
Glenwood teachers agree: Denise believes that without a familiar routine, her students would question her constantly about what to do next. Marie specifies the sub-activities of her "ours brun" lesson by writing the following on the board:


And she then explains the order of activities: "Numero un, je regarde; numero deux, je decoupe; numero trots, je les place; numero quatre, je les colls." ("Number one, I look; number two, I cut; number three, I position them; number four, I glue them.") Wong-Fillmore (1985) noted that teachers use lesson scripts that they have adopted for each subject. Once they learn the sequence of sub-activ¬ities for each subject, students can follow the les¬son without having to figure out afresh what is happening each day.
Pierre often uses boundary markers between activities to provide familiar clues and "frame" the event for the students: To mark the end of circle time, Pierre and the class recite and perform the rhythmic finger play:
Frappe, frappe, frappe, les doigts croises; Frappe, frappe, frappe, les bras croises; Frappe, frappe, frappe, les mains fermees; Frappe, frappe, frappe, les pouces leves.
("Clap, clap, clap, fingers crossed; clap, clap, clap, arms folded; clap, clap, clap, hands closed [into a fist]; clap, clap, clap, thumbs raised.")
At a more advanced level, teachers may state simply, as did Patrice, "Les matbematiques main-tenant. Quelle multiplication?" ("Mathematics now. What multiplication?") to mark the end of one activity and the beginning of another. Or a simple "Take out your pencil and paper," spoken in the L2 can indicate to secondary school/college students that a writing activity will soon begin.
By providing as much structure as possible and depending on visuals and formulaic clues, lan¬guage teachers can eliminate confusion and make it possible to use the L2 more hi their classrooms. Recognizing that classroom directions are some¬times the most meaning-based communications of their day, second language teachers should exploit these inherently meaningful situations by struc-turally clarifying their L2 use as much as possible.
Immersion teachers often integrate different subject-matter areas. Pierre moves easily from lan¬guage studies (singing a Halloween song) to math (counting Halloween characters). Patrice integrates reading and math: After lunch, she begins a story, "Casquettes a vendre" ("Caps for sale"), and asks comprehension questions while showing the illus¬trations. In answer to how many hats of each

Chapter 13 / Immersion Teachers 133

T
the secondary/college level, askir
t" questions is even more necessary, giVl
ents' intellectual maturity. Languaj
at these levels should be careful not:
'oning because of students' limit!
ability. Using nonverbals, offering tu
answers for students to choose from, ar
with an incorrect answer are techniqu
lenwood School that can easily be adap
condary/college classrooms to facilita
L2 responses. In spite of their restrict*
abilities, mature students can I
ed to respond creatively to intellectual
ng questions.
practices and underlying principle
d above have one common them.
e is integrated into the teaching of coi
, Met, and Genesee (1989) list sever
s for the effectiveness of integrating coi
language teaching, notably that contei
a cognitive basis for language learning :
rovides real meaning that is an inherent
naturalistic language learning. Further,
ways that their "conceptual frame-
an accommodate different types of ele-
school second language learning
. However, in second language classes at , teaching concepts should be the top pri-concepts can be linguistic, as is the otion that grammar rules can be modified ote fluent pronunciation (e.g., "Vas-y! ie. . . Parle-t-elle?") or sociological, such ench sense of privacy (e.g., walls around ouses, reluctance to discuss personal mat-s important to address basic concepts in
avoid superficiality and rote memoriza-inguage. Not addressing underlying con-1 lead students to believe that language is a mere spelling exercise, six Glenwood immersion teachers who ted in this study are dedicated to teaching
integrated with the teaching of a foreign . Because they nearly always communi-
language that is unfamiliar to their stu-:y have refined many techniques that are ible to all levels of language teaching.


NOTES
1 This research was supported by United States Department of Education grant #R168F80060.
j,jCting housekeeping tasks in the L2, using -•verbals extensively, involving students as peer ■ ■ che'S) structuring classroom routines, integrat-!."" various subject matter areas, and asking high-.'r.order questions can improve any second f ■.igua'.;e classroom.
OISCUSSION QUESTIONS
In addition to the fact that adults generally know more than children, what other factors differentiate the experience of children versus adults in acquiring a new language? For instance, suppose an adult sz\h something that other adults do not understand. How will they respond? But, suppose the speaker ■i ?. child and the hearers are also children. How will the children respond? Also consider the difference bsrv, sen the way adults react to children versus children to adults. How might these differences impact the acquisition of the subtleties of surface forms (e.g., a native-sounding accent, perfect surface rnor-o'ro ogy and syntax, etc.)?
; "VC hr:: results should be expected if the opportunities for language students to produce comprehensible oLtpat were to be increased? Or suppose students have lots of opportunities to understand input but |:ttk or no opportunity to produce output. If comprehended input always must involve an active pro-CL.c::on of an interpretation by the person doing the comprehending, in what sense, exactly (from the s-jcl.-'nt's point of view), is comprehended input less difficult than comprehensible output?
„. \X ha: sorts of social factors contribute to the stabilization or destabilization of the sort of system char¬acterized by Hammerly (1987) as a "terminal classroom pidgin"?
4. In what ways are immersion classroom activities scaffolded to aid comprehension?

Chapter 13 / Immersion Teachers 135

14
Qguage Instruction
rite Ann Snow, i Wesche
UCTION
rinton, Snow, and Wesche define particular content with language-tguage is most effectively learned in ts and pieces of linguistic discourse net as to create an authentic text, hat naturally occur in real contexts, the true narrative type (see Chapter nstruction. Very sensibly, Brinton, ■) should "dictate the selection and ught rather than vice versa." Both nty of negotiating the forms and ztions) in such a manner as to put ., to the facts at hand). They stress, zxtualizing" the target language.
rourse features more appropriate than oth-particular language-teaching purposes? re the implications of using varied subject us opposed to developing the same theme i? How can we know which topics will be :st to our students? How can a given con-: be integrated with language-teaching pur-the language classroom? se and other related questions have led slogists to carefully examine the role of in language teaching. This investigation :o a widely shared belief that simply "con¬ing" language lessons which are organized structures or functions is not enough, theorists and practitioners have suggested ting point the use of authentic texts which ant to the learners' second language needs ritten or oral texts which were created for

lose other than language teaching. These ' e in concrete form the structures, functions, scourse features to be taught. One view is icse features, once identified, can then be at least partially in isolation, with lessons i on particular language forms, functions, ttterns. A second view is that the emphasis informational content itself provides an re means for incidental acquisition of the ge features it presents. Content-based lan--eaching approaches, in fact, often combine ,n form with experiential techniques, this volume, we define content-based don as the integration of particular content ■ith language-teaching aims. More specifically, ncc we are dealing primarily with postsecondary jucstion, it refers to the concurrent teaching of -adcmic subject matter and second language :ills. The language curriculum is based directly n the academic needs of the students and general-follows the sequence determined by a particular ibject matter in dealing with the language prob-ms which students encounter. The focus for stu-:nts is on acquiring information via the second ngur.ge and, in the process, developing their aca-;m:c language skills. Ultimately, the goal is to lable students to transfer these skills to other :ademic courses given in their second language, hus, both in its overall purpose and in its imple-entcicion, content-based instruction aims at elimi-lting the artificial separation between language struction and subject matter classes which exists most educational settings.
In a content-based approach, the activities of c language class are specific to the subject matter ■ing taught, and are geared to stimulate students think and learn through the use of the target nguage. Such an approach lends itself quite natu-!ly to the integrated teaching of the four tradi-tional language skills. For example, it employs authentic reading materials which require students not only to understand information but to inter¬pret and evaluate it as well. It provides a format in which students can respond orally to reading and lecture materials. It recognizes that academic writ¬ing follows from listening and reading, and thus requires students to synthesize facts and ideas from multiple sources as preparation for writing. In this approach, students are exposed to study skills and learn a variety of language skills which

prepare them for the range of academic tasks they will encounter. This type of approach has impor¬tant implications for course design. The course design must indicate the means by which the con¬tent is to be integrated with language objectives. Accordingly, the curriculum and materials must reflect this overall design.
A content-based instructional approach has a number of implications for language teachers as well. First, language instructors are asked to let the content dictate the selection and sequence of lan-guage items to be taught rather than vice versa. They are asked to view their teaching in a new way, from the perspective of truly contextualizing their lessons by using content as the point of departure. They are almost certainly committing themselves to materials adaptation and develop¬ment. Finally, with the investment of time and energy to create a content-based language course comes even greater responsibility for the learner, since learner needs become the hub around which the second language curriculum and materials, and therefore teaching practices, revolve.
A RATIONALE FOR CONTENT-BASED LANGUAGE TEACHING
At least five different rationales for integrating the teaching of language and content are implicit in content-based approaches. First, proponents of English for Specific Purposes (ESP) note that for successful language learning to occur, the language syllabus must take into account the eventual uses the learner will make of the target language. Thus, focus is on the language forms and functions which will best serve the learner, based on system¬atic description. Second, even though learner lan¬guage needs and interest may not always coincide, the use of informational content which is per¬ceived as relevant by the learner is assumed by many to increase motivation in the language course and thus to promote more-effective learn¬ing. Third, content-based approaches apply the pedagogical principle that any teaching should build on the previous experience of the learner, as they take into account the learners' existing knowledge of the subject matter and of the acade¬mic environment as well as their second language knowledge. A fourth rationale is that language

Chapter 14 / Content-Based Second Language Instruction 137

T
things are known knowledge of words fol-. . we cannot hope to learn words we do
IOW unless we have grasped their meaning, not achieved by listening to the words, but
:ing to know the things signified.1
■e is, in fact, ample historical evidence < lan ability to learn a second languat meaningful exposure to its use. For cei pper-class European families have recoj :he utility of direct contact an ication needs in effective language lean ending their children to live and study ] where the language of interest is spokei rting governesses and tutors who woul anguage with the children. Likewise, so iders, immigrants, inhabitants of bord< ' "isoners of war, footloose students, an rists have throughout history proven ab ; re the second language skills they nee ' contacts with speakers of the languag | the contact is extensive enough and the i >n sufficient. i
withstanding the above, educators ha\ ! wed both the goals and the approptiai logy of school second language learnin it differently. Language study in scho< considered to be training in mental disc well as a key to foreign literatures an_ for the educated classes. Such attitudes st in many foreign language-learning situ-Q schools and universities, although .es which advocate language learning for cative purposes through meaningful lan-e have become influential in situations language to be learned is a second lan-clear functional utility for the learner, in which learners need to improve d language skills for purposes of univer-unites formal second language instruc-real and present functional needs. Here mmunicative language teaching and d, content-based instruction merge in ective objectives. -1980S: THE ROOTS OF CONTENT-TEACHING nterest in language-teaching approaches lasize the mastery of particular infor-content as an integral element of second |:.,guage instruction represents a coming together -a o^jcrical experience and theory. In light of the ^prion that language can be effectively taught ' u„h the medium of subject matter, these proaches view the target language largely as the „; _je through which subject matter content is ■mcd rather than as the immediate object of ■dv At the same time they recognize that impor-']t'"ains in language proficiency occur "inciden-lv" ialbeit purposefully in a methodological ise'. as language is used in the understanding -d exoression of meaning. Thus, the traditional njs of both native and second language classes - awareness of linguistic form is largely subordi-tcd ro a focus on acquiring information through ' second language. Such an approach has been oeasingly espoused by teachers and researchers ■olvcd in the teaching of native language skills .ross the curriculum, and it is also the corner-me of such innovations in second and foreign -iKii-ge teaching as immersion language pro- ■i-'.i.. In the following section, we will present ■nc examples of how content-based language ;truction is currently being implemented in □ve, foreign, and second language settings.2 Language Across the Curriculum "lae language across the curriculum movement as a means of first language development received a major impetus in 1975 from the report of a com¬mittee commissioned by the British government to consider "all aspects of teaching the use of English, including reading, writing, and "speech" ;cf. Bullock Report, 1975). A major finding of the committee was that first language instruction in the schools should cross over all subject matter domains. Accordingly, there followed a recom¬mendation for a policy of teaching language as a part of instruction in other curricular areas in British schools. The committee's report and the experimentation which followed it have been very influential in North American schools as well. The principles of language across the curricu¬lum for native speakers of English have been fur¬ther specified in instructional programs at the postsecondary level (Griffin, 1985; Thaiss, 1987). In programs for reading and writing across the curriculum, the perspective taken is that of a recip¬rocal relationship between language and content learning. Students are not only given opportunities to learn to write and learn to read but are also encouraged to write to learn and read to learn in order to fully participate in the educational process. These objectives, by definition, necessitate cooperation between language teachers and sub¬ject matter teachers, with the language teachers emphasizing instruction focused on language prob¬lems in reading and writing English and the con¬tent teachers complementing this with activities requiring reading and writing in the learning of subject matter. As a consequence of this cross-cur-ricular focus, the language across the curriculum movement has resulted in new developments in teacher training and materials development. For example, several publications describing strategies for cross-curricular teaching at the secondary and postsecondary levels deal with such issues as designing effective writing assignments and essay questions, improving the writing process, and eval¬uating student work (Anderson, Eisenberg, Holland, Wiener, & Rivers-Kron, 1983; Simmons, 1983). In short, the language across the curriculum movement is an attempt by both British and North American educators to offer native English speak¬ers access to the full range of educational activities in which language and content are inextricably woven together. It provides students who have highly developed functional skills in their own lan¬guage with an opportunity to further refine the language skills required for advanced schooling— those needed in understanding and expressing complex content using the more-decontextualized skills required in academic reading and writing. This movement has influenced second language instructional theory and practice. It is particularly relevant to the development of content-based sec¬ond language instructional programs of the type treated in this book, in that it provides a method¬ology for students with varying degrees of profi¬ciency—those who need to develop their skills in precisely those areas required for successful learn¬ing of academic content. Language for Specific Purposes Language for specific purposes (LSP) is perhaps the best known and most documented of the con¬tent-based language models. LSP courses generally Chapter 14 / Content-Based Second Languase Instruction 139 other content-based models in which omponent is experiential language learnb :xt. lion Education major example of experimentation wit. -based language teaching can be found i and foreign language instruction in Canad United States. The St. Lambert, Quebec immersion project, begun in 1965, wa 1 in accordance with the theory of the da ' second or foreign languages are bes d (Lambert &c Tucker, 1972). Intensiv e to the target language through natur; rication with a native speaker was consic mtial, as was starting at a young age. Th snt began with a classroom full of English. ; kindergartners receiving the regular hal: riculum entirely through the medium c xom a French-speaking teacher. As yeai research results demonstrated the succes rsion for the mastery of both function; .nd school subject matter, programs wet r other school boards. Two decades latei sion" has become a mass education; ;nt in Canada, where over 240,00u i children in all provinces and territories nuch or all of their school instruction a second language. This instruction takes a variety of immersion formats for differ-proups, and while it primarily involves t also is used for instruction in heritage es such as Hebrew, Ukrainian, and i Indian. s 1971, with the establishment of a Immersion Program in Culver City, i, immersion education has also spread in :d States as an alternative to traditional anguage in the Elementary School (FLES) . While the student enrollment in immer-jrams is still far smaller in the United m in Canada, the American immersion reflect both geographical and language There are approximately 25 immersion around the country offering foreign lan-rruction in Cantonese, Spanish, German, and French (Rhodes, 1987). nrnersion model is a carefully researched >f content-based second language instruc-

- -be elementary and early secondary levels, ites the effectiveness for kindergartners to ." : its of instruction which focuses on teach-:ct matter through the medium of the see¬page. While the language itself is largely ncidentally, there is also, as in first lan-istruction in the schools, some explicit language rules and forms. Since the suc-the immersion model have been achieved ■ :xclusively with language majority chil-ose whose home language is that of the mmunity and is thus not at risk—its main on is to foreign language teaching. For dren, constant out-of-school exposure to anguage, as well as first language instruc-chooi, ensures its continued development iut their school years, crsion-type education through the second has been notably unsuccessful with lan-inority children where there is inadequate d support for continued mother tongue :iopment, such as in some programs in English native Americans and Canadians on reserva-s. It is also unlikely to be an appropriate alter-ve for young immigrant children learning the linant language of their new country in school if „ _ not accompanied by first language instruction.
For language majority children who wish to scd a second language in school, 20 plus years of immersion studies in both Canada and the United 5ta:es have yielded remarkably consistent findings (Genesee, 1987). After several years of intensive exposure to the second language, immersion stu-der.ts achieve a high level of functional ability in the second language, with near-native proficiency in receptive skills by the time they graduate from an elementary school immersion program or a late immersion high school program. While their speaking and writing skills fall short of native speaker performance, such students normally achieve functional skills at a level far beyond what other second language instructional approaches have been able to produce in the schools. Immersion is thus highly successful in developing advanced second language speakers who can sub¬sequently perfect their skills through real-life inter¬actions with native speakers of the language. As for first language and subject matter learning, pri¬mary immersion students catch up to their mono¬lingual peers in scholastic achievement, and they rapidly catch up in native language skills after first

language instruction is introduced. The immersion approach is used successfully with complex subject matter in bilingual secondary school programs as well, yielding impressive proficiency gains together with mastery of the content. In Canada, for exam¬ple, these programs generally follow an early or late immersion sequence, with students taking 40 to 50 percent of their courses through the medium of French. Some forms of content-based instruc¬tion at the postsecondary level are in fact an exten-sion of these secondary school programs to the university setting.
CONCLUSION
The above discussion suggests that content-based instruction can be a very effective way to teach both first and second language skills in school con¬texts. Practical experience with language teaching, systematic thought about language teaching and learning, and empirical research on classroom lan¬guage learning all suggest that content learning through the medium of the target language can serve well as a major component of instructional programs for all age groups. It further suggests that such instruction is particularly appropriate where learners have specific functional needs in the second language.
How, then, does it work? Classroom experi¬ence and second language acquisition theory both tell us that rich second language input in relevant contexts is the key, where the attention of the learner is focused mostly on the meaning rather than on the language. This experiential component appears to be a vital element in the development of functional second language skills, with contextual-ized analytical activities which focus explicitly on language forms, functions, and patterns playing a complementary role in the development of accura¬cy and precision in language use (Allen, Carroll, Burtis, & Gaudino, 1987; Stern, 1978, 1992).
NOTES
1 St. Augustine 60: XI, as cited in Kelly (1969: 36).
2 We have selected these three examples of content-based instruction in different settings in part because of our own personal experiences and in part out of the clarity of the three models and their well-documented implementation with fairly homogeneous populations.

Chapter 14 / Content-Based Second Language Instruction 141

T
ent for content-based instruction, why ^ ... t to teaching a non-primary language (all ej"
2 the fulcrum of successful language insert h ro the separation of skills? If we can list." ut them, write about them, dramatize thei hem to music, interpret them aesthetically mating listening and speaking activities in Qg just composition, or just conversation , -tiveness of a Dr. Pepper ad that appeals on .ot of action accompanied by snappy dialo c plot, etc., suppose we just heard against h is apt to be more effective, the advertiset
' reflect the natural ways in which "languai Snow, and Wesche put it?

CHAPTER 15
Sheltered Subject-Matter Teaching
Stephen D. Krashen
EDITOR'S INTRODUCTION
K
rashen says that "subject matter teaching in a second language, when it is compre¬hensible, is language teaching, because it provides comprehensible input." Sheltered teaching can occur when all the students are more or less at the same level of develop¬ment and when the focus is on the subject matter. Krashen distinguishes two kinds of research showing the effectiveness of some form of sheltering: research that focuses on the acquisition (1) of the subject matter and (2) of the target language. Both sources show that sheltered language teaching works. Another alternative is the "adjunct (or pull-out) model," in which non-native speakers of the primary language system are put in regular classes (along with natives) for subject matter but have one or more additional classes to help them master the language of each particular subject area. In addition, Krashen recommends supplementary activities such as role playing and board games to enrich opportunities for negotiating and increasing the amount of comprehensible input in the target language.



• - i by the success of Canadian immersion TIS (see e.g., Lambert & Tucker, 1972), :d subject matter-teaching (SSMT) derives le important concept: Subject-matter teach-, second language, when it is comprehensi-language teaching, because it provides
: hensible input.1 A history class, given to
second language acquirers, if it is comprehensible,
is;: language class.
There are several crucial characteristics of
SSMT:
1. In SSMT, only second language acquirers are allowed in the class. When all students are sec-ond language acquirers, when all students are in the same linguistic boat, it is easier for the teacher to make the input comprehensible .
2. In Sheltered Subject Matter classes, the focus of the class is on subject matter, not lan-guage. This encourages a focus on meaning, not term, and results in more comprehensible input,

and thus more language acquisition. Sheltered sub-ject-matter classes are thus not "ESL Math" or "ESL History" but are "math" and "history."
If possible, the tests and projects also focus on subject matter and not language. When the test is on subject matter, students will listen to lectures, participate in discussions, read the required and recommended texts, and obtain a great deal of comprehensible input. When the tests are on lan-guage, students will be tempted to conjugate verbs and memorize nouns, and little language acquisi-tion will take place. Similarly, when projects and papers deal with subject matter, students will read extensively in the second language and will obtain comprehensible input.
3. In SSMT, teachers attempt to make input comprehensible. This is done in several ways, including frequent comprehension checking, which indicates to teachers when they need to adjust the input they are providing, and the use of extra-lm-

Chapter 15 / Sheltered Subject-Matter Teaching 143

>ND LANGUAGE MEDIUM STUDIES
iest known of the second language medi -:s are the many reports of Canadian-st rsion (summarized in Lambert & Tucl —
Swain 8c Lapkin, 19S2). It has been shoo :edly that children in these programs acoui--ssive amounts of the second language ar-a great deal of subject matter. In addition ■" lmersion studies, a number of research nr :onfirm that SSMT works for older studen > H.
he Ottawa studies (Edwards, Wesch : len, Cle ment, &c Kruidenier, 1984; at- ' tman, Wesche, &; Ready, 1988) showed th ■sity students could learn both subject matt- i lology) and make progress in a second la !
at the same time. Participants, who were v( ' ■s, had already studied one semester of colk- •
ology in their first language (English . l), and had at least low intermediate know af the second language (French or Englisl leltered course was second semester psychol -
Hauptman et al. one experimental group d red psychology for two semesters), and w anented by a half-hour weekly session with I age teacher, who did no direct gramm - ! ng, but focused on comprehension of conte . developing strategies for effective reading ai nterventions" (Hauptman et al, 1988: 445) . general, subjects made progress in secoi ige acquisition equivalent to students in rcu- i :cond language classes, and acquired subicct : just as well as students who took the sa me : in their first language, o (1982a) (see also Ho, 1982b) reported tha: graders in Hong Kong who had had second age medium instruction for three of their years of EFL were far more proficient in h than comparison students with eight years ditional EFL. (The English medium student iave had more total exposure to English as lowever.) Ho also reported that second l:.a-
medium students learned as much physics gh English over a three-month period zs. irison students did in their first language.
Ho (1985), eighth graders in Hong Kong Dok courses in English learned as much sjb-latter as comparison students who took :s in their first language in four out of five

Second language medium instruction did
C°,r "«r to result in additional second language
rt' , Both experimental and comparison
1L* this study, however, did all subject
""',). ig in English, which reduced the treat-
^...V ices. (Swain, 1988, in discussing this
;uggests that the fact that nonnative
"'..." Jght the Second Language Medium
\\~ ,ave been a factor, as well as the
"rsrhodology used.)
Buch and de Bagheera (1978) found that ESL r-jchcrs who were not native speakers of English Jr.ide significant gains on the Michigan Test and " .significant gains on a cloze test and writing ■ ■ r after taking eight applied linguistics courses in ■: clish. No comparison groups were used in this s-jdv for either language acquisition or content-■s- owledge learning.
Two studies (Saegert, Scott, Perkins, 6c '"_cker. 1974; Gradman &C Hanania, 1991) found significant relationship between years of subject > irtcr instruction through a second language and -;cc:id language proficiency among students of Z- glish as a foreign language. In both studies, ■ars of subject-matter instruction through . giish was a better predictor of English profi-: *ncv than was years of formal instruction in - -glish.3
CONTENT-BASED SECOND LANGUAGE TEACHING
i-.'hlcpsegrell (1984) reported that EFL students — :de significant gains on an essay test and test of renmg comprehension after a five week content-"■ sed economics course. No comparison groups were used. (These subjects outperformed a com¬parison group that did a sheltered economics course in which the emphasis was on output rather than input. Comparison subjects took the essay test only.)
Lafayette and Buscaglia (1985) reported that fourth-semester university level students of French as a foreign language who studied French civiliza¬tion and culture did just as well as a traditional fourth-semester class on several measures of French proficiency (listening and reading), and made better gains on a speaking test. On a writing test, however, the comparison class was slightly

better, gaining about 5 points (pre = 169.3; post = 174.54) as compared to the experimental class's 3-point gain (pre = 165.29; post = 168.47). Lafayette and Buscaglia noted that the writing test was real¬ly a grammar test, with more than 20% of the items on the subjunctive. The comparison group focused on grammar, with two units on the sub¬junctive, while the sheltered classes relied exclu¬sively on acquisition. Since it is quite likely that the French subjunctive is late-acquired (for evi-dence from Spanish, see Stokes, 1988; Stokes & Krashen, 1990), it is no surprise that the sheltered class did not do quite as well on this test.
A very impressive finding is that more of the sheltered students intended to enroll in additional French courses, and more students in this class reported that their interest in studying French had increased as a result of taking the course.
Peck (1987) found that students of Spanish as a foreign language (second-semester college level) made significant gains on an oral test and a listen¬ing comprehension test after taking a seven-week course on social work, which included some direct grammar instruction. There was no comparison group.
Sternfeld (Chapter 18) is unique because it involved beginning foreign language students. First-year college Spanish students who studied Latin-American history, geography, and culture did as well as traditionally taught students on tests of reading comprehension and listening com¬prehension. Comparison students did better on a writing sample, however. Sternfeld noted that this may have been due to the fact that the topic of the writing test was familiar to the comparison students, but was not included in the sheltered class.
Milk 1990 provided content-based second lan¬guage teaching as part of a teacher training pro¬gram to 17 bilingual and ESL teachers at the University of Texas at San Antonio. Students par¬ticipated in one of two summer sessions, held for two hours per day for five weeks (42 contact hours). Students read and heard mini-lectures on second language acquisition theory, designed class¬room activities in groups, and kept dialogue jour¬nals in Spanish. Participants varied considerably in initial Spanish competence, but significant group gains were found on a variety of language tests. No comparison group was used.

Chapter 15 / Sheltered Subject-Matter Teaching 145

cannot be determined which factors of tly t model were helpful, especially since the t classes in the two studies were somewhai it.
TIONS TO SSMT
(198S) maintains that "not all con ten-.g is necessarily good language teaching" (p L content teaching in a second language ing to Swain: 1) students do not product i output, and do not produce enough com itput. More demands for output, accordinc in, will "help learners focus their attentior ticular form-function relationships" (p. 73) e is little correction; 3) the input is "func ■f restricted," that is, "certain uses of Ian may simply not naturally occur, or ma) airly infrequently in the classroom setting'
iave argued (Krashen, 1991) that point! d (2) are not a problem, since languagi tion does not require output or error cor L. In fact, Swain's findings showing tha rade immersion students get little correc-ld produce only modest amounts of Ian-are excellent arguments that output and ion are not necessary, since these children early made excellent progress despite hav-little output and correction. (This is not, :se, to say that output is bad for language tion. I have argued in several places, e.g., :n, 1982, 1985a, that output helps indi-
by inviting comprehensible input, as well
:tively.)
ere are two possible solutions for the third
m Swain mentions, restricted input. One
lity is to "contrive contexts," deliberately
ice contexts that ensure the use of certain
This is difficult to do, since it requires lg what rules students are ready to acquire
")■
:econd possibility is simply to expand activi-i the range of topics and subjects covered, will naturally include more functions and (Swain, 1988: 77). This solution is easier more interesting for teachers and students. I vt some specific suggestions below.

HEWjMRECT10NS IN SSMT
i-MT has been successfully applied to much of the
mentary school curriculum (Swain & Lapkin,
■ _ }2) arid, as we have seen, to subject matter at
."- ■ university level. Students have learned psychol-
- v (Edwards et al, 1985; Hauptman et al., 1988),
- cure and civilization (Sternfeld, Chapter 18;
I fayette 8c Buscaglia, 1985), economics
jchleppegrell, 1984), social work (Peck, 1987)
■ 1 applied linguistics (Buch 6c de Bagheera,
■' 78; Milk, 1990) in SSMT.
I have discussed some other possibilities for
.-.■Itered courses for foreign and second language
.-.dents elsewhere (Krashen, 1982, 1985a). These
.irses would probably provide much of the vari-
- of input that Swain maintains is currendy lack-
v from many content-based courses.
Two of the most promising areas for sheltered . -sses that would be usable for all levels are . rses in popular literature and games. Popular
■ rature and games promise to provide a wide
. iety of input, using activities that students find
merely interesting but often compelling.
POPULAR LITERATURE
Including a sheltered popular literature class may be a good way to combine pleasure reading and sheltered subject-matter instruction, two very effective means of moving beyond conversational language.
There is very strong evidence that pleasure reading is a major source of our advanced linguis¬tic competence (see e.g. Krashen, 1985b, 1989a). In fact, there is evidence suggesting that merely making some popular literature available has a positive effect on literacy development (Rucker, 1932).
The goal of a popular literature class is to introduce students to many kinds of popular liter¬ature so that eventually students will read on their own. This includes comic books (for a review of the research, see Krashen, 1989b), magazines, newspapers, and popular novels.
Such a class will also give students a consider¬able amount of information about the everyday culture of the speakers of the target language, as wcll as linguistic competence.

GAMES
Several kinds of games might be very effective at the intermediate level. Straightforward board games promote interaction, and have the potential of supplying some subject matter knowledge: Britannia (Avalon Hill Game Co.), for example, takes place in Britain in the first century. While playing, participants inevitably learn a great deal of history.
The fullest potential of games is reached in what are termed "role-playing games," extremely complex games which require demanding solitary reading for character creation, and extensive group interaction in playing the actual game. The best known of the role-playing games is Dungeons and Dragons, but many variants exist, including some that set their adventures in actual historical loca¬tions, such as the China and Vikings modules from the GURP (Generic Universal Role Playing) system. Playing these games should result in significant sub¬ject-matter learning as well as language acquisition.
While there has been no evaluation of the value of role-playing games in language acquisi¬tion, it is a safe bet that they will be effective. Role-playing games provide input through reading, as well as input through interaction, and research suggests that interaction is extremely helpful in making input comprehensible (e.g., Pica, Young, &C Doughty, 1987). In addition, Rhoda McGraw and Sian Howells have been offering role-playing games as part of advanced English as a Foreign Language at the fZcole Nationale des Pont et Chassees in Paris, with great apparent success.
An obvious problem with games, as with all interaction activities, is that students hear primari¬ly the speech of other students, or "interlanguage talk" (Krashen, 1981). I have argued that interlan-guage talk probably does more good than harm, "but if students hear only interlanguage talk, there is some chance they may acquire the errors they hear, leading to fossilization (Krashen, 1985a). The cure for this is to include native speakers in the games. Including native speakers as game par-ticipants violates one of the principles of sheltered subject matter teaching, but is consistent with a deeper principle: comprehensible input. When native speakers are in the game, their input can be highly comprehensible and useful; because of the constraints of the game, the students will have

Chapter 15 / Sheltered Subject-Matter Teaching 147


CURRICULAR ISOLATION OF FOREIGN
LANGUAGE LEARNING
Despite three decades of programs designed to promote widespread language learning in the United States, only a small fraction of students achieve functional language competence. Even in postsecondary institutions having a one- or two-year foreign language requirement, foreign lan¬guage study still is often focused primarily on the formal aspects of the target language, and the practical proficiency developed is usually limited to that required for the most basic social needs. Smdents acquire an initial vocabulary which they
ES
I am assuming some familiarity with the I :hesis, the hypothesis that we acquire languaE >ne way, by understanding messages. See m (1982,1985a).
SSMT is not the only way of helping studi beyond conversational language. Other technic e encouraging free voluntary reading and ' use of the student's first language (bilingual edi I. (For supporting arguments, as well as ways r SSMT can be combined with bilingual educate ishen, 1985a, 1985b.)
Increasingly Saegert, et al. (1974) also found thj :tion using French as a language of instructio-so a significant predictor of English proficieno /Spanish as a language of instruction was not ant predictor in Gradman and Hanania's stud' :r. The strongest predictor of English proficienc idman and Hanania was "extracurncula-
>f progressing in target language acquisitioi Why is mere exposure to such surface form Chinese on the radio not be an inadequat
idling encourage students to go beyond th .ow, for instance, does negotiating meaning advances? Phonological development? And
)ly, with a native, in what ways do the facts Dtiations? Suppose someone wants to buy a n the price of the hotel, its location, its size,

CHAPTER 16
Foreign Languages Across the Curriculum
Wendy Allen, Keith Anderson, and Leon Narvaez
EDITOR'S INTRODUCTION
J
t has been known for a rather long time (Carroll, 1967) that for the great majority of students, foreign language education in the United States generally falls short of the goal of getting students to the point where they can communicate effectively in the target language. More likely, they will learn a few words and get some limited and oversimpli¬fied ideas about the "grammar" of the language, and beyond that they will remain almost as monolingual as most of them were when they started out. One of the reasons for this often lamented fact is, as Allen, Anderson, and Narvaez note, that the connection between the target language and the kinds of subject matter that the foreign language student both needs and often wants to study is rarely made. Content-based instruction is a logical way to make the connection between the foreign language and its meaningful uses. Instead of teaching verb paradigms, a minimal vocabulary, and a few sentence pat¬terns with little or no connection to experience, Allen, Anderson, and Narvaez suggest that foreign language students at a fairly advanced level of development in the target lan¬guage ought to be given a variety of opportunities to study subject matter in the lan-guage. Faculty response, student enrollments, and informal reports indicate a positive reception of such programs at St. Olaf College in Minnesota.
then work to expand; they study the verb tense system of the language; they begin to put the vari¬ous elements of the language together, first in sen¬tences and then in paragraphs. But, by the time they are becoming ready to focus more on the use of a foreign language to acquire and process con¬tent knowledge, i.e., to employ the tool which they have so arduously mastered, the requirement is completed and they move on to other areas of the curriculum. Consequently, students experience a sense of intellectual fragmentation with respect to their foreign language learning experiences.
Our foreign language requirements rest on the premise that as students acquire foreign language

Chapter 16/ Foreign Languages Across the Curriculum 149

T
1 understanding; and (4) instruction is o the cognitive and affective needs of -nd to their proficiency level, inton, Snow, and Wesche (1989 ;-:r 14) provide what is to date the n :hensive attempt to define CBI: "the tr :ocus ... on awareness of linguistic foriu ■ subordinated to a focus on acquire lation through the second language ' this volume). They view interest in CBI ng together of the theory of second |a: acquisition specialists such as Krashi -c), on the one hand, and practical expei i the other. As these authors express it,
;room experience and second language acqui-i theory both tell us that rich second lan-s input in relevant contexts is the key, where .ttention of the learner is focused mostly on neaning rather than on the language. This riential component appears to be a vital ele-in the development of functional second lan-;e skills, with contextualized analytical ities which focus explicitly on language s, functions, and patterns playing a complc-ary role in the development of accuracy and sion in language use (p. 139).
rording to Brinton et al. (Chapter 14), C ly involves the concurrent teaching/learnii :ct matter and of language:
language curriculum is based directly on the :mic needs of the students and generally fol-the sequence determined by a particular sub-natter in dealing with the language problems i students encounter. The focus for students is rquiring information via the second language in the process, developing their academic lan-; skills. Ultimately, the goal is to enable stu-; to transfer these skills to other academic ;es given in their second language (p. 135).
I encompasses a wide range of curricul , including the teaching of native language ;ross the curriculum (e.g., language or writ-oss the curriculum programs), immersion ns (first begun in St. Lambert, Canada, in nd currently experiencing a revival in U.S. tary and middle schools), and language designed to prepare learners for specific, irld demands. Although examples can be

as early as the 1920s, CBI did not play a ant role in college curricula until the 1980s, imary model was the "special purposes" : course in areas such as business, health id technology. Another model described by : (1988) was the "integrative" approach uy Earlham College in an NEH-sponsored which identified and prepared texts written aages other than English for use in courses the curriculum. To suggest that CBI is a "movement'" is problematic; it is more ■cly described as a bifurcated movement, le branch focusing on language and the ier on content.
Bnnton et al. (1989 and Chapter 14) in their Micrit-Based Second Language Instruction ;cribe three models for CBI at the post-sec-dary level. In theme based courses, often used in ESL context, the language class is organized in ms of content "modules," and the language chcr teaches both the subject matter and lan-age- (One of several Monterey Institute of em:.:ional Studies CBI programs is based on this ids). I In sheltered instruction, such as that srs-'C at the University of Ottawa, a subject mat-co-nse is taught to a segregated class of second gjcge learners by a content specialist, some-■.es, accompanied by supplementary language ■r-jciLon and sometimes not. Finally, in adjunct truer ion, as in the UCLA Freshman Summer ■gr'-in, students are enrolled concurrently in a g'.:«"ic;e course and a content course that are lired." While all three models seek to promote z'nmy, language through content, they differ in relative emphasis given to language and to con-t k-arning, in overall structure, and in the levels i situations for which they are appropriate. -*1- represent points along a continuum:
GUAGE Theme-based Sheltered Adjunct MAINSTREAM

-*-CLASS
(Brinton etal., 1989:23)
The two projects described below, while hav¬ing some of the features associated with the adjunct model cited above, nevertheless differ to some extent from it and from CBI in general, in that their goal is not content-based foreign lan¬guage instruction, but foreign language-enriched

content instruction. Here we emphasize the con¬cept of enrichment because, while our projects are multidisciplinary in origin and development, and while, clearly, they foster students' foreign lan-guage proficiency, their primary purpose is not so much to enhance foreign language acquisition as to enrich disciplinary study.
THE ST. OlAF PROJECTS
St. Olaf College, a private, church-related liberal arts college located in Northfield, Minnesota, is attempting to address the problem of language/content integration via a two-year pro¬ject funded by the National Endowment for the Humanities (NEH) and a three-year project sup-ported by the Fund for the Improvement of Postsecondary Education (FIPSE). The St. Olaf programs have all four of the characteristics cited by Leaver and Stryker. They share features with the Earlham project, but go beyond it in terms of level and range of second language use.
The St. Olaf projects have three objectives. The first objective is to identify a core group of faculty across the curriculum who have advanced foreign language proficiency and who are willing to incorporate foreign language texts into their courses. The second objective is to create appro-priate foreign language materials for use in a group of specially adapted courses. The third objective, and perhaps our principal concern, is to encourage students to continue second language study and use beyond the three semester require¬ment by offering them the opportunity to combine advanced language work with disciplinary study, typically carried out in the context of institution-wide general education and distribution require¬ments, and/or requirements for a major.
The participating faculty members have all had significant international experience. Language faculty have demonstrated their commitment to language teaching and have been involved in teaching and doing research about the culture and history of their areas; non-foreign language disci-pline specialists have studied and conducted research in their language(s) of competence and have interest and expertise in cross-cultural analy-sis and interdisciplinary work. Both groups are committed to integrating language study and

Chapter 16 / Foreign Lansuages Across the Curriculum 151

~T

d Writing Component," is indicated 0l] transcripts. In order to receive this certifi. students must complete the prerequisite s in a particular foreign language (0r strate equivalent proficiency) as well as extend their foreign language compe-two AFLC courses.
Students who wish to combine the study 0f language with the study of the history of the respective language community ogram carried out in that language may (in addition to the AFLC) a "Foreign Area itration," which also is indicated on the ipt. These concentrations require five ' study of the same language plus success-etion of four courses which address the and culture of the particular area. At least courses must offer the AFLC.
the two projects have certain ele-common, they nevertheless differ in sig-ways. They are, therefore, described lly below.
H PROJECT: "FOREIGN LANGUAGES DISCIPLINES"
project seeks, first, to bring together a of faculty, representing the range of the ies to study and discuss, in a structured setting, the central role of language in the ies and its implications for the undergrad-ral arts curriculum. A second goal is the on into courses taught by these faculty of foreign-language texts in order to pro-students a greater understanding of between language and the human-and deeper understanding of for-, and a greater mastery of foreign s.
Languages in the Disciplines" is
to demonstrate to students that the
of foreign language competence is an
of a liberal arts education and that
of the humanities will deepen
mastery of the language increases.
g the use of foreign language materials to

jjjes beyond those offered by foreign language lartments gives students access to important ts not normally studied in language classrooms i often unavailable in English. More importantly, students learn how, in -erse disciplines, language itself influences the ipe as well as the content of the discipline. In a rman history course, for example, students who - d in the original language some of Wilhelm von jmboldt's writings on education learn that
■ Itur und Bildung carries implications much
■ jper than the English translation of "culture and
ication" would suggest. Reading these texts in » original leads students to explore the meaning culture and its emergence in Europe in the eigh-nth and nineteenth centuries as an autonomous egory with ethical connotations. The class then
- imines the enormous importance in German
aety of the Bildungsbiirgertum—another term
ose meaning is partly lost when translated into
glish as the "educated middle class."
The students who elect to apply their foreign .guage proficiency in a particular humanities irse are not the only beneficiaries of this deci-o. The entire class gains from the deeper level of cussion and the increased attention to meaning de possible by allowing at least part of the class work directly with non-English texts in their . ■ xanslated original form. Gradually, the conven-: na! wisdom of students is changing. Students
■ -.o read particular texts in their original language
- seen by their teachers as well as their peers to
re attained far greater depth than those who
"■ ve had to rely on English translations or instruc-
summaries. In time, it will become part of the
udent culture" that it is both advantageous and
eresting to do more than simply "satisfy" the
eign language requirement. Students will come
" recognize the compromises, judgment calls, dis-
" tions, and sheer blunders involved in transla-
" n, and this recognition will make their reading
even the daily newspaper more sophisticated
" i thoughtful.
Student language competencies also are lanced as students read, analyze, interpret, and cuss foreign language materials pertinent to a rticular discipline. Students who have limited :ir language study to the requirement are :ouraged to continue further by the opportunity apply their foreign language proficiency to the

subjects of interest to them, albeit subjects rarely included within the curriculum of most foreign language departments.
Project Design
The NEH project is a two-year effort (June 19S9-June 1991) centered on faculty development, course adaptation, student recruitment, library acquisition, and program evaluation. The core group of fourteen faculty members (seven pairs, each pair composed of a foreign language special¬ist and a humanities specialist) modify seven already-existing humanities courses to include for¬eign language texts. In these courses, which are designated as having an "Applied Foreign Language Component" (AFLC), students who enroll in the AFLC "track" of the course may elect to complete approximately half of the course read¬ings and other assignments in a particular foreign language. The AFLC students, in addition to the regular class meetings, meet in weekly discussion sessions conducted in the foreign language, which are led by a faculty pair consisting of the foreign language teacher and the course instructor.
Faculty Development and Course Adoption
The core group of faculty participates in an ongo¬ing program of faculty development. First, these faculty meet in a week-long introductory seminar in which they explore the relationship between foreign language study and the study of other humanities disciplines, and the curriculum and pedagogical implications of integrating foreign language texts into traditional humanities courses. After the seminar, faculty pairs devote four weeks to modifying courses in history, literature, philoso¬phy, and religion, so as to include an Applied Foreign Language Component (AFLC). Three courses were modified the first year of the project, and four additional ones the second year. These courses, together with the language of their AFLC, are as follows:
Backgrounds to British and American
Literature (Spanish) Modern Germany (German) The Christian Tradition in History (French) Essentials of Christian Theology (German)



*.

Chapter 16 / Foreign Languages Across the Curriculum 153

lete a basic set of (English-language) rt. But those who elect the AFLC will do at le rercent of the total course reading and rela-ch in the foreign language. (Those who ect the AFLC will complete this additi0l ig and research in English.) Written work
students may be submitted in English or >reign language, depending on the writ; iency of the student. In addition to the th
per week of class required of all stude-id in the discipline course, those who el FLC discuss their reading and research r one-hour sessions conducted in the fore .ge and led by the faculty member from ige department together with the cou ctor. Students who elect to enroll in ■ section of these specially designated hum; ourses receive an additional one-quar -
credit for their participation in the requi ■ foreign language discussion sessions.
/ Resources
;r important component of the project i tared program of library acquisition. A p al librarian with expertise in the humani: foreign languages works in close collabc ith the project director and faculty. Th aals determine the strengths and weaknej college's current collections of books, jc ewspapers, government documents, ma eo and film resources in various disciplir 1 or produced in languages other tr , and identify additional foreign langu. 2S appropriate for student use in the spec ted humanities courses offering an AFLC
»SE PROJECT: "FOREIGN LANGUAGES NATURAL BEHAVIORAL SCIENCES"
a three-year project (September 1989-1992) which extends the possibility for foreign language study to the natural and ral sciences, and to mathematics.
Design
SE project contains many of the same fea-the NEH program, but also offers several

ial options for students to complete the Students can earn Applied Foreign >t Component credit in a variety of ways:
ial foreign language immersion (the NEH
lei).
cted readings. Individuals or small groups
tudents work with disciplinary faculty ibers who have foreign language compe-e to complete directed study projects in a ign language.
■}tz\ immersion January term course, ing St. OlaPs January term, courses which nally are taught in English are taught in a ign language by two faculty members, one i the sciences and the other from foreign uages. Courses may be taught on campus uroad.
ience course taken in a foreign language iad Certain of St. Olaf's study-abroad pro-is allow students to enroll in science or lematics courses taught in the language of lost country.
nternship in a foreign language community iad This includes individually arranged rnships as well as group experiences in h a St. Olaf faculty member accompanies )up of students who complete a one-month nship abroad during the January term, ial immersion semester. During certain :sters, two-course or three course interdis-nary seminars are offered in a given for-language. The thematically integrated ;ram of study constitutes half or three ters of the students' academic load for the
SE offerings to date include, in addition to :cted readings courses, the following team-taught disciplinary courses:
American Foreign Policy (Spanish)
Ethical Management: Germany and America
(German) Marriage and the Family (Spanish)
Library Resources
Holdings of foreign language editions of books in the natural and behavioral sciences are quite limit-

ed in the typical undergraduate library. Thus, to support the AFLC courses, a coordinated program of library acquisitions is being carried out. The work is directed by a professional librarian with expertise in the sciences and in foreign languages, working in close collaboration with the faculty participants.
PROJECTED OUTCOMES
These projects will determine whether or not, at a four-year undergraduate liberal arts college, signif¬icant numbers of students who are non-foreign language majors will elect to continue foreign lan-guage study beyond the graduation requirement when given the possibility of attending use of their foreign language to general education require¬ments and/or their major fields of study. The sev¬eral options for content learning in a foreign language will be evaluated and compared with respect to their contributions to subject matter mastery and to enhancement of students' foreign language abilities.
The projects also seek to build a cadre of disci¬plinary faculty members who will become mentors and role models for students who wish to apply their foreign language study to their non-language majors. This is accomplished both through sup¬port of existing faculty who are reactivating and enhancing foreign language competencies, as well as by making command of a second language a factor in the recruitment of new faculty.
The identification, acquisition and develop¬ment of appropriate learning resource materials for content learning in foreign languages is a fur¬ther goal of the project. This includes primarily library materials, but also other media such as films and videotapes.
The long-term "global" outcome which both projects hope to achieve is a gradual erosion of the traditional separation between foreign language study and the study of other disciplines in the minds of both faculty and students. The projects do not propose to make chemists into German teachers or French teachers into historians. Rather, the two projects seek to open a dialogue and develop the basis for interdisciplinary cooperation, a process which will yield enhanced relevance for foreign language study and add a new and exciting

Chapter 16 / Foreisn Lansuases Across the Curriculum 155


DISCUSSION QUESTIONS
' [f one of the objectives of foreign language education is for students to "acquire a broad awareness of Foreign cultures and a deeper understanding of language as a human phenomenon," can such a goal be achieved at all in the absence of a substantial degree of competence in the foreign language? If learning i few words in Spanish and a brief exposure to a few structures in that language will not produce the kind of knowledge and multicultural understanding that foreign language instruction aims to instill, what more will be required?
I. It is argued that substantial competence in a foreign language can enhance comprehension of the broad spectrum of subject matter. How is this possible?
3. Content-based instruction is "bifurcated with one branch focusing on language and the other on con-tent," according to these authors, who also say that it is the connection (or "integration") of target lan-guage with content that is crucial. Discuss the bifurcation and integration relative to the theory of pragmatic mapping (see Chapter 37).
\. If the goal of content-based instruction is to "enrich disciplinary studies," how can it be demonstrat¬ed that this has occurred? What sons of evidences could be marshaled (for instance, course enroll¬ments, faculty participation in the program, student reactions, scores on standardized tests in the disciplines, etc.)?
nal sources of information—would i: t simply their AFLC course, but all of ;. This suggests that the ultimate impa
projects will go well beyond the creatit Is to be used in a relatively limited nui ;es in the humanities and sciences. :erms of the faculty as a whole, the A i has generated more discussion about nguage competence and use than i on the campus for many years. The ny faculty—and not simply language f
talking about languages and the degr hey are currently used on campus by and students is a positive developmer
of 1990, over a dozen non-foreign acuity were enrolled in language cou : intent of using their language in a va 2xts including, in some cases, the A i. These developments constitute pro. one of the major goals of the applied guage projects, namely, attending fo : study and use across the campus, or student response, the goal of aver; i ten students per AFLC course has
indications are that this enrollment . continue. For example, nineteen suit illed for a three-course interdiscipli on seminar offered in the fall of ' e auspices of the FIPSE grant, ent evaluations confirm students' recogni-he value of the program. They particularly ted the opportunity to discuss ideas and : substance made possible by the weekly scussion session. As one student wrote, "I dditional hour per week to concentrate on
and speaking about the material." This lity enabled a number of students to con-n a substantive way to full class discus-■olving both AFLC students and students in the English-language track. AFLC stu-xe able to contribute insights they might
had in the absence of the weekly foreign
discussion sessions.
:nts reported they had been able to main-1 language skills and to apply them to new
As one observed, "I enjoyed being able to
language to another discipline, and it was be able to read firsthand many primary ts." Many students cited their expanded ■oved second language vocabulary and a

ceived strengthening in their reading comprehen-Most recognized that participation in the LC track of the course helped them focus on ar was critical to the course and, in general, »d their preparation for and participation in the ^ar class sessions. As students stated, "Often in AFLC we discussed more elaborately topics that ■e pertinent ro the [regular disciplinary] course" 1 "I gained more insight into the readings." Xhe majority of students conveyed either ver-[y or in writing appreciation for the ability to -k directly with materials written in a second mage. The responses of students enrolled in the ■rtnan AFLC section of a history course on idem Germany were typical: "Some things, ie words, are not easily translated and by read¬me text] in German, I could better understand it was being conveyed," and "Reading about average German's life in German gave it a

meaning that would have been lost in translation." Certainly the student support for this experiment has been more than sufficient to justify its institu-tionalization.
St. Olaf s Applied Foreign Language Compo¬nent program and similar programs at other col¬leges and universities in the U.S. and Canada have received considerable attention within the foreign-language teaching profession and beyond (Moline, 1990; Straight, 1990; Watkins, 1990). These pro¬grams provide multiple ways of encouraging sec-ond-language use across the curriculum and thereby help to overcome the curricular isolation of second languages to which we referred earlier. We envision a future in which the academic use of languages is seen as an integral part of the under¬graduate curriculum of most colleges and universi¬ties—and we and others are beginning to make that vision a reality.

Chapter 16 / Foreign Languages Across the Curriculum 157

PART 4
LITERACY IN ALL ITS GUISES
ecoming literate means much more than just becoming able to read and write.
It involves the acquisition of the skills needed to relate written representations
in the language of literacy to relevant aspects of experience in general. The literacy
connection does not merely relate to this momentary here and now, though that
connection is everyone's starting point, but to the whole world around us with its
past, present, and future, together with all of their diversities and complexities.
Becoming literate in this broader sense means being empowered in such a manner
as to become a participant in the larger world community—a community of many
languages, cultures, political entities, and of increasingly complex multinational
organizations of all sorts. In the twenty-first century, being literate means having
the intellectual tools and skills that give access to all sorts of meanings and possi¬
bilities that would otherwise remain out of reach. Practical literacy means know¬
ing how to negotiate physical space more efficiently, as in using public
transportation systems. It means being able to consider and select from a wider
] range of alternative courses of action than would be possible in a preliterate state.
It means greater freedom, responsibility, and the power to participate in social, economic, educational, political, and other activities of the larger world communi-ties. In a sentence, literacy is the essence of a practical education for modern living. Therefore, Part 4 is about literacy in this broad sense of the term. It is new in this edition of Methods That Work and is predicated on the growing realization among educators everywhere that acquiring a language and learning to read are part and parcel of the same fundamental problem of actively comprehending rep¬resentations. For readers familiar with the writings of Krashen (1981, 1982, 1985b, 1989a, 1991) and those of Frank Smith (1978, 1982, 1988), the close analogy between acquiring another language and learning to read will not come as a surprise. But regardless of whether the relation between these educational objec¬tives seems familiar or surprising, the insight is well-grounded. Acquiring a lan¬guage and becoming literate are fundamentally similar problems. This can be proved through sound theory, empirical research, and methods that work in both literacy and language instruction.
The first chapter of Pan 4, Chapter 17, is aimed at elementary teachers and others who work mostly with children at the early stages of becoming literate. The principles explained there, among them the scaffolding of discourse to enhance its comprehensibility and to establish its social purposes and motivations, are thor-oughly grounded in pragmatic theory. In addition, the methods advocated in Chapter 17 have been tested with populations of subjects where the failure rate in
PART 4 / Literacy in All Its Guises 159

any particular exemplar of a story) is probably universal across languages and cul-tures. In Chapter 37, we will see logical proof that the formal schema underlying narratives, in fact, must be universal. Further evidence supporting the universality of a formal schema for narratives is seen in Chapters 20-22. In Chapter 20 a series of seven different studies, involving narratives, expository texts, and descriptive prose, shows that literate native speakers of English, as well as non-native users of English from a variety of language backgrounds, are able to understand and recon-struct portions of texts more easily when those texts appear in their original schemata-constrained arrangements. As will be argued in Chapter 37, that order is grounded ultimately in the formal schema associated with the episodic structure of experience itself. Empirically, there is confirming evidence that the logical basis underlying narratives is universal across language and cultural backgrounds. By contrast, however, comprehension of the particular content of texts (what Carrell in Chapter 19 has called "content schemata") appears to be dependent on particu¬lar experience and familiarity with certain kinds of subject matter as is shown in Chapter 20. As a result, knowledge of content schemata appears to be language and culture specific to a much greater extent than formal schemata are, and this is exactly what the theory (Chapter 37) predicts.
Chapter 21, by Yukie Horiba, goes some distance toward helping us to see just what a narrative schema might be like. It turns out that an analysis of native and non-native protocols in think-aloud readings of a story in Japanese reveals a linear propositional structure with multiple inferential links between its overt and covert propositional structures. By investigating such complexities, Horiba offers insight into the character of narrative schemata. It appears that such schemata are grounded in nothing other than inferential logic coupled with particular facts of ordinary experience. Horiba's data show that native speakers of Japanese are, as we would expect (and as the research of Chapter 20 also shows), more capable than non-native speakers in sorting out and expressing the relations between the proposi-tions underlying the comprehension of a story in that language. Non-natives, how-ever, are able to compensate for their lack of skill in Japanese by appealing to their knowledge of the world and the scaffolding that it provides in helping them to understand the story (also see Taira, Chapter 28). Horiba's work helps to put a par-ticular shape to the otherwise nebulous concept of a formal schema.
The last chapter in Part 4, by Tudor and Tuffs, extends schema theory to the processing of video materials. The authors show that both formal and content schemata are important to the processing of an eight-minute expository video. The subjects of their research were students of business administration at a university in Belgium. The video in English (a second language for the subjects) concerned the pros and cons of a private road system in Britain. The results show that advance knowledge of the formal schema underlying the video, as well as specific facts contained in the video (the content schema), are helpful in improving both comprehension and recall of the material. It appears that scaffolding, just as prag-matic theory predicts (see Chapter 37), is helpful in video comprehension as in other forms of literacy.
PART 4 / Literacy in All Its Guises 161

CHAPTER 17
Literacy in the Third World for All the Children
Richard F. Walker, Saowalak Rattanavich, and John W. Oiler, Jr.
EDITOR'S INTRODUCTION
T)uildmg literacy is a challenge. Even for children from relatively privileged back-ID grounds, traditional approaches widely produce unacceptably high failure rates.
When literacy must be provided to children who do not yet know the language in which they need to become literate, failure is the norm more often than not. In this chapter, we offer an alternative pioneered by Richard F. Walker of Australia, Saowalak Rattanavich of Thailand, and colleagues from Australia, Thailand, and elsewhere. Results with Aboriginal children at Traeger Park School in central Australia (at Alice Springs in the Northern Territory) led Walker and colleagues to believe that what came to be known as the CLE (Concentrated Language Encounters) approach could work in almost any diffi¬cult context. For his efforts to bring literacy to third-world countries Walker was made a Member of the Order of Australia on June 8, 1992. The fundamental thesis of his work, and that of his colleagues (see Hart, Walker, & Gray, 1977), has been that we ought to consider what children can do in actual discourse contexts rather than to pre-judge what they might not be able to do on the basis of some preconception or other. It turns out that all normal children can do more than many of the theoreticians have stvpposed. They can even acquire a language while becoming literate in it. The proof is found in thou¬sands of cases where children in difficult third-world contexts have achieved literacy in a non-primary language. It is generally assumed that six or more years of formal education will be sufficient to produce persons who can read and write, but this is not always true. Even in developed countries that have had compulsory public education for several gen¬erations, substantial numbers of illiterates can be found, and all over the world, many children leave school each year with inadequate literacy skills. Actual failure rates vary, but figures above 50% are not uncommon. While the skills required vary across situa¬tions, is it unrealistic to expect the great majority of students to complete their school years with essentially all the literacy skills they will need for adult life? This chapter addresses that question and concludes that it is possible to teach virtually all children to read and write, regardless of their language background or prior experience.

HOW CHILDREN LEARN TO READ
Children from actively literate families seem to acquire naturally a basic understanding of what reading and writing are about.1 A surprising num-

ber of them begin to read before they ever set foot in a classroom (Smith &£ Johnson, 1976: 28). Success in reading is essentially a problem of con-necting written language with experience. But to do so, as in solving any other representational

Chapter 17 / Literacy in the Third World for All the Children 163

:erest in seeking out methods that work,
oppressed populations, re convinced that success or failure in rhildren to learn to read has more to do t goes on in classrooms than with what ren bring by way of background. As we prefer to look to cases where success spite the odds, and to see how methods : can be created for the full range of stu-
o come to school—not just for those nily background happens to fit a precon-Dfile for success. The evidence shows, at that privileged children of high socioeco-tus are likely to learn to read even before ; to school and that most of those who » so beforehand will go on to become lit-ty much in spite of what the schools do do. Therefore, it is apparently just as riate for the schools to take full credit for successes in the case of the privileged and D children as it is to blame high failure le less successful groups on socioeconom-, The sad fact is that in a large percentage he schools are less helpful than we might cabling children to become literate. On
hand, if we turn to methods that work, ill news is that we stand to leam a great
can learn from the children who read )ing to school and from those few pro-here even highly oppressed minorities :e nearly 100% literacy achievement at 'here is good reason to suppose that we ed teach all the children to read. The rated Language Encounter (CLE) tech-:scribed here have been proved in practice much wider range of children than tradi-:racy programs. In fact, we will argue that nee shows that such procedures can work ually all the children at school, even in bird-world contexts.
:e proceeding, a few words on the phrase ■ated language encounter (CLE) are in he name for the CLE approach comes zden (1977). She argued that children anguage mainly through occasions where intensely involved in understanding and derstood. We agree with this idea and see amental task of educators as figuring out increase the density of such intense com-

municative successes in the experience of school¬children. Our job is to make high motivation com¬municative successes occur as regularly as possible in classrooms. In this chapter we describe a philos¬ophy of language and literacy teaching which accomplishes this fundamental objective. We begin by discussing some actual samples of discourse in the hope that we will be able to show (rather than just tell) our readers how CLE approaches really ■work in practice. Then, we go on to discuss the overall organization of CLE programs, especially in Thailand, where successes in teaching children to read and write in Thai, even when their first language is not Thai, encourage us to believe that it is possible to teach all normal children to read regardless of their socioeconomic profile or their experiential background before coming to school.
LANGUAGE RESEARCH EVIDENCE: MARY
The fundamental principles of CLE language and literacy teaching were worked out in searching for a solution to the gross failure among Aboriginal students at Traeger Park School, in the Northern Territory of Australia (Gray, 1983). Mary was a beginning student at that school. The transcripts that appear in this section reveal what her lan¬guage use was like in teacher-child interactions Transcript A) and while playing (Transcript B) tad working with other children (Transcript C). ■Inr noint is to show how child-teacher and stu-udent interactions can be enriched through aching (Transcript D). The transcripts were in a language research project soon after :tarced going to school (Walker, 1981). we find her early in the school year, Mary ; a dialect of English in addition to her rinal first language. In English, she can tand what is said to her and make herself tood, but, according to traditional wisdom, ilmost certain to fail in school. Her teachers iribute her failure mainly to limited English ~ ;ncy perpetuated by unfavorable socioeco-factors. The first sample of discourse, ript A, involves Mary and her classroom r. Another child named Jane, a friend of i, is seated side-by-side with her on the iom floor, where they have been making

things out of colored rods and fittings that connect them (what Americans call "tinker toys"), in a free activity period at the beginning of the school day.
Transcript A
Teacher: What are you making, Mary? (Mary does not respond.) What are you mak¬ing? (No response.) Very nice. (The teacher examines Mary's construction.) What's this part? Is that pink? (Mary shakes her head, indicating that it is not pink.) No? What is it? (No response from Mary.) Blue?
Mary: Yeah.
Teacher: Mm. It's blue. What's this one here?
Mary: Blue.
Teacher: Good. That's blue. What's this one? Yellow? (Mary nods her head.) Can you see another yellow one? (Mary points to another yellow one.) Very good. . . . (The same procedure is repeated for other colors.) What are you doing now, Mary? (No response.) What are you doing, Love? (No response.) Are you putting it together? Building something? (No response. Mary goes on working.)
Mary: (As she works, Mary says:) Making chair.
Teacher: Making a chair? Oh? Who's the chair for?
Mary: Sitting down.
Teacher: For sitting down? Oh. Who's going to sit on it? (No response.) Is Jane going to sit on the chair or you? (No response.) Who's going to sit on it?
Mary: Jane.
In this conversation, the teacher works hard to establish communication with Mary. She tries to teach the names for objects and colors, on the basis of Mary's own activity. Simultaneously she seems to be trying to see whether Mary already has those concepts and the English labels for them. The teacher's first try only requires that Mary name the object she has been making, but Mary does not respond even to a second trial of that probe. The teacher tries to establish a better rela¬tionship by admiring the chair that Mary has built.

Chapter 17 / Literacy in the Third World for All the Children 165

Vera:
'. of a playground interaction on the sai sveals that Mary can operate at a level )ar
what the teacher was able to elicit. Mary ■'<; g a double push-pull swing with Sue. The rls are sitting on the same seat, one behind ler, with Mary in front and in control of the 'ull bars. Two other girls are sharing a sirni. ng nearby, and a fifth is dancing around th* awaiting a turn. We notice that here Ma-v ring to actual material events in which sh-2 other children are presently involved. :ript B On again. On here again. Me push. (S'ie says to Sue:) Look, you push me. Yo: push. (To the other pair of girls, Mar. says:) We'll go really fast like you. (To Sue:) Like this, eh? Hey, Marilyn! Fa': down and . . . (Mary tells the fifth girl, who's dancing around, to repeat the trick she's just done). (Then to Sue:) Move back. Sue! (Mary is slipping off the front of t"e seat.) No. You! No! Hey, Sue! (Mary •<; annoyed because Sue has slipped along the seat.) I'll take you really fast. Sue! Yc.i thing! (She's annoyed that Sue is wrigglirg around.) (Now, both girls get off the swing.) Me turn. Jane and me and Jai-.j now. Me and Jane now. (To Jane:) Yor-and me. (To the other girls:) You and Si:e at the back. Jane, other side, other side. his playground episode, Mary is garrulous ninant to the extent that no one else has a speak. More significantly, her dominance ithers is achieved and maintained throug.i nglish. Clearly, she can use English conli md powerfully in this context. At the end lay, in free classroom time before leaving ne, Mary is playing with large colored •n the floor, and we get some additional into her ability to use English. iptC Fhe colors, the colors, colors. (Bill, a class¬mate, moves over to Mary and begins to innoy her. She threatens him quietly:) 1*11 :ell on you. (Mary's classmate Vera says:) Bill Brown! Leave her alone! Mary: (Calling softly for the teacher:) Mrs. Peters? Bill: (Then to Bill:) I'll tell on you, Bill. You're Mary: making troubles. You're making troubles. Teacher: Not me. (Bill retires, and Mary continues Mary: playing. Looking at the colored beads, she Sally: says:) All the colors, all the colors. Teacher: From Transcript C we learn that Mary under- Mary: stands classroom dynamics well enough to mount Teacher: an appeal to the teacher's authority. She also Mary: knows how to attribute blame for a breach of expected behavior. Finally, after Bill's aggression is Teacher: quashed, Mary's returns to her almost poetic use of English as she expresses her appreciation for the Bill: colors of the beads. Teacher: On several other occasions, different ones of Bill: Mary's teachers try to draw her into an instruc- Teacher: rionsl dialogue. In each case, a stereotypical dia- Mary: Iogue of the adult-to-infant kind develops (such as Teacher: in Transcript A)—where the adult ends up provid¬ ing answers sought from Mary. By this method, the teachers obtain almost no idea of what she can Mary: do in English as evidenced in Transcripts B and C. Teacher: In all the transcripts (many not shown here), it is Mary: dear that Mary's teachers are unaware that she Teacher: car. .ise English at a much higher level. They act as Mary: if Mary were a bare beginner in English. For her Teacher: parr. Mary is content to play the game as the Mary: teachers have unwittingly defined it. Why take any Teacher: new risks? Better to play the safe bet of waiting for the adult teachers to supply whatever responses arc squired. This leaves them with the self-perpet- Mary: natmg impression that Mary is a nonparticipant. Teacher: But what would happen if the classroom game Mary: were substantially changed? Transcript D was Sally: recorded on the next morning in a "concentrated Teacher: language encounter." This was one of the first Mary: CLEs at Traeger Park School specifically designed Sally: to enable Aboriginal children to participate fully in Teacher: classroom learning. This session, which had been performed twice on previous days, involved mak¬ ing toast in a corner of the classroom. Mary: Transcript D Teacher: Bill: Can I put some butter? Mary: Teacher: We have to wait first until the toaster is Teacher: ready to pop up. What color do you think the bread will be when it pops up? What color? Brown. Brown. Brown. That's right. It pop up. I can smell it now. (To the other children:) Can you smell it? (Enjoying the smell of the toast.) Hey! Does it make you feel hungry, Mary? (Nods her head.) My mother got peanut butter. Your mother's got peanut butter, has she? We got some peanut butter home. You've got peanut butter at home too? Yeah, and one of these. And a toaster? And one of these. (To Mary:) What's that called? (She's interrupted by the toaster popping.) Oh! It's popped up! Black now. It's black. D'you think it's burnt? Yeah. I think it's probably all right. (Excited:) It's black! Black! It's black! Dinner time. Yeah. It's dinner time. Dinner time. Dinner time. Take the toast out of the toaster now, Bill. (Bill does so.) What are we going to do now, Mary? Clean it. (Surprised:) We're going to clean it? (Demonstrates:) This way. Scrape it. Scrape it. Why do we have to scrape it? It's yucky! With the knife. No, that's all right. (She thinks the toast doesn't need scraping.) Then what do you have to do? What are we going to put on it now? Peanut butter. What are we going to put on it before the peanut butter? Butter. Good girl! Come on, then, you spread this. A nice piece of toast. Bill's going to Chapter 17 / Literacy in tine Third World for All the Children 167 into the toast-making discourse (e.g., that ner time), the activity underway provides >f scaffolding so that the fiction works. The
lants know what is going on and what is
d of them. The fiction is fully comprehensi-
losely relates to the facts at hand, which are
in lots of ways other than just through deci-
spoken English. The objects can be seen
ched, and the toast is about to be spread
and peanut butter and is about to be
supportive social relationships between
and each other and the teacher are coded
1 expressions and gestures that are readily
ble and comprehensible in the context. The
are out in the open, so to speak. The
unlike trying to imagine someone, for
tinker-toy chair might have been made,
sitting in it, or why the teacher would be
ed about such a possibility, exclusively on
of talk about those hypothetical (unreal
I) possibilities—especially when the
upplied only in a non-primary language,
. There are huge differences, as a result,
Transcripts A and D.
reasons for the differences between
A and D become clear to participants in
process. The teacher realizes that because
nd the other children) have shared in toast
on several occasions, they have become
with it. They know the names of things,
to be done with them, and the sequence in
hose events will occur. As a result, Mary
what the role of toast maker requires,
she's seen others take turns in that role.
Mary has realized that all the language
s to be made on her as toast-maker will be
>n procedures with which she is already
and comfortable. In other words, she can
ely relate all that is said in this context to
persons, and events so as to comprehend
everything that goes on. She relates the
of the language used, even unfamil-
to what she sees happening, and to what
is expected of her in the context. She
in bringing about the on-going trans-
of the immediate context of experience
phase of the toast making to the next
activity is completed.
terms of the theory of Chapter 37, she ch basis of pragmatic scaffolding from

hich to advance easily, seemingly without effort, previously inaccessible levels of linguistic per¬formance. For example, notice the dialectal switch \c jnakes during the session shown as Transcript n Early in the session, when the toast pops up she volunteers, "It pop up." We note here that she Joes not use any surface marking on the verb. No ■opula appears as in "It's popping up," nor does -fie third person singular (non-past) morpheme (-s} on the verb pop appear, as in "It pops up." However, as the discourse progresses Mary consis-■entlv switches to forms that overtly show the cop-ila as in "It's black," "It's yucky," and "It's ■nclting." The other transcripts do not show any -uch elaborated surface forms. Why? Apparently, Marv is naturally sensitive (subconsciously, just as ■CrasKen claims, 1985b, 1989a-b, 1990, 1991, .992, and in Parts 2 and 3 of this volume) to J-alcct changes. As a result, without any special :ffort she switches to "It's black" after the teacher _avs "It's popped up," and then she continues in ■he teacher's more elaborate dialect. She does not leeci to be explicitly "taught" such standard struc-urei, and all indications are that such explicit -caching would fail to produce the desired results n any case. However, in a rich CLE-type context, >nce a variant has been modeled for meaning and ■unction, Mary can use it for similar purposes quite naturally.
In the carefully planned context of a CLE, the latural use of the teacher's primary language or Jialect works, precisely because the modelling has u!l contextual support. There is plenty of scaffold-r.g 10 make the meaning transparent from the loir.r of view of the participants in the toast-mak-ng activity. Also, because of the cooperative ispect of the learning experience, there is a predis-";ob:t:on on the part of the children to uncon-ciously assimilate the elaborate language of the cacher. There is no threat and, therefore, no rea-on not to accept and use that language.
Remove the scaffolding of the CLE, however, ind the classroom scene will revert back to the ort of discourse seen in Transcript A, where the most at-risk children are almost certain to remain confused. As Gray (1980: 3) put it, they just don't know "what is required from them in the learning task." This in turn leads to "low self-esteem as learners" (just as Gray observed) and an ever-renewed cycle of failure after failure. The children

are not incapable in any general way, but relative to the specific classroom alternatives that they arc presented with, they just aren't apt to understand what is going on or why. They have little chance of understanding what the teacher is getting at or why. Social relationships and motives are unclear. The child is apt to feel threatened or at-risk because of such uncertainties. Therefore, we must change the classroom context.
There are children like Mary in schools every¬where. They are have sufficient intelligence—no matter what sociopolitical contingencies may have been dealt to them by circumstances beyond their control. The "Mary's" of the world's classrooms are not all Australian Aborigines, and they are not necessarily from any particular language or ethnic minority, but there are many of them out there. We have only to visit the schools, and this is espe¬cially true in developing countries, to realize that exclusion from classroom learning occurs on a gigantic scale. With children like Mary, whose first language is not the language of instruction, a par¬ticularly effective recipe for wholesale failure is to allow only teacher-directed interaction in the class¬room, and to "cover" just those concepts, words, and structures that are ser down in some textbook to be drilled. In all too many cases, across large regions of the world, and even in developed nations in many places, literacy teaching begins with the alphabet and progresses through phonic drills to individual words, later to sentences (most-ly chosen to illustrate particular phonological and syntactic patterns), and often bogs down in a morass of decoding confusion before the children can become readers.2 In such cases, real reading and writing (for useful purposes) are postponed indefinitely—even permanently for many children. As a result, whole generations of students have found themselves engaged month after month in puzzling, laborious, analytical, boring activities that have no apparent purpose. Before long, they give up trying.
Such direct, analytical approaches to the teaching of language also discourage students from any real learning outside the formal lessons on which they are taught to rely. Moreover, the con¬fusion is self-perpetuating as the new crop of teachers in each generation of educators slogs through the same old ineffective methodologies and is taught to believe that they are the best ones

Chapter 17 / Literacy in the Third World for All the Children 169

works in practice, but first it m-, 1 to summarize the order of events it-e Thai-language program was developed and subsequently implemented on a tna th substantial numbers of some of thi .-risk minorities in that country, whic] ead in 1991 to its being adopted for the ation (approximately six million priman hildren) by the Ministry of Education a! ts National Plan for 1992-1996. The pro-presently being implemented throughou-)n. In the meantime, pilot projects for new jgrams are simultaneously underway near India and in several schools near Dhaka sh. It is inreresting that what began as t ESL and literacy program for Aborigine! alia went on to develop into various simi rams in third-world nations, and at lasi full circle to an ESL program in addition '-literacy program in Thailand. Thai-literacy piloting projects began first rought-stricken provinces of northeastern , where 97% of the children speak , or some other minority language, have Thai as their primary language, pilot programs (called "lighthouse" pro-re begun in the southern border , where Malaysian is the predominant . Later the CLE concept would be the extreme northern provinces, where a hill-tribe languages and dialects are spo-ing 1991, CLE demonstration schools in the remaining parts of the nation in for implementation of the 1992-1996 Plan. By the summer of 1992, CLE meth-already being used in more than 16,000 ols.
THE CLE FAMILY OF APPROACHES?
with Aboriginal students at Traeger
, Gray noted that the most successful
sessions were those in which students
)lved in contexts where they were doing
and useful things, but where they had
to challenging language tasks to get
.3 This same principle underlies all
and activities. Connected to the
iple is the matter of "scaffolding." The

that what is said relates to and is supported h°trie context—by what is perceivable, by what is
*o eninS> by the actions students are performing, Ci the accompanying gestures and tone of voice, * j ky the previous experience of the partici--' nts—the easier it will be for students to compre-i- "ltd, participate successfully in what is going on, yd thus to acquire language and literacy skills r-om that context. Scaffolding is important to all icnns of discourse. Just understanding what is
jng on can be a tremendous help in figuring out \ -jat is being said and how to participate in it. In edition, knowing the facts pertaining to any ■ctivity or context of meaning is crucial to break-
■ s the code of written discourse pertaining to
- Dse facts. The facts, of course, include not only :' ; objects and material setting but also the social -.lations between participants, their goals, crtneanors, and manners of communicating with c ch other. The already established connection ; tween utterances and the facts of experience is r icial in enabling the preliterate child to begin to
- come literate.
In keeping with these observations, CLE pro-cvais can be divided into three major "stages"
■ iich in turn can be broken down into several
chases." The latter, in turn, can be further
-„■ iuced to units or activities in the classroom. We
- ilize that any such parsing of time segments is
. bitrary, but in fact, schooling is inevitably
rsed up into distinct time segments. In CLE pro-imming, Stage 1 roughly embraces the first two three years of schooling and typically begins ...th children in kindergarten or first grade. However, it is applicable in many preschool set¬tings. The objective of this first stage is to produce enthusiastic readers and writers of the language used for instruction. Stage 2 aims to broaden the students' repertoire of text genres so as to encom¬pass the full range of texts they will be apt to encounter throughout their schooling and in later life. This stage usually occupies the middle portion (about two years) of the primary grades. Stage 3, then, completes the primary schooling and may extend into the secondary school program. The main objective of this last stage is to teach students to organize their knowledge in terms of the full range of text genres they will need in later life. In this stage, they will learn to create such texts, to learn from the process of creating them, and, in

general, to understand the vast and varied connec¬tions between texts and literate discourses of all types and the contexts of experience to which they relate. Here we concentrate primarily on Stage 1 of CLE programs. Readers who want to know more about Stages 2 and 3 are encouraged to read Rattanavich (1992).
Stage 1
The overall objective of Stage 1 in any CLE pro¬gram is for students to become readers and writ¬ers. At the end of Stage 1, they should be able to read various types of simple texts and to recall and discuss what those texts are about. They should be able to write brief texts of several kinds, observing spelling and other conventions of writ¬ing. In progressing through this stage, they will learn the written forms of hundreds of words, many more phrases, and how to bring effective strategies to bear in recognizing and writing words they have not previously encountered in a written form.
Stage 1 of a CLE program usually covers two or three years, depending on whether or not it is used in preschool or kindergarten classes. The pro¬gram for each school year is organized into units, each of which extends over a number of weeks. The actual number of units that are covered in a year varies from class to class, but most Stage 1 classes in Thailand will do ten to fifteen units in a year. Within each Stage 1 unit, the teacher and srudenrs progress through five "phases." There are two kinds of Stage 1 program units, both of which are illustrated in Figure 17.1. Based on Chapter 37 and, in fact, all of the chapters of Part 3 concern¬ing content-based approaches, we should predict just the two avenues that have materialized in CLE approaches.
On rhe one hand, Figure 17.1 shows "activity-based'" approaches, grounded in the experience side of Figure 37.1 in Chapter 37, and on the other, there are "text-based" approaches which are grounded in the representational side of Figure 37.1. Text-based approaches begin with written material. Language teachers, and especially lin¬guists, might expect that activity-based approaches would be best when working with children whose first language is not the language of the classroom. However, experience teaches that either acriviry-

Chapter 17 / Literacy in the Third World for All the Children 171

E5
activity games
activity games
activities that have been used in programs of Thailand include making a , learning to preserve beans, breeding a garden, and many others.4 Text-' , by contrast, start with shared read-Starter Book, usually based on an narrative. A favorite narrative with of Thailand is a story about a mon-a tiger that are friends when they are lit-et separated and later meet in the jungle s. Since the two are natural enemies, as children can readily appreciate, the is whether the tiger will eat the monkey, of the story, which has an American the two old friends recognize each other well.
programs using published CLE materials to the generous help of Rotary volunteers), To" Starter Book is provided. Some of be used as reference items for certain based units, but more often than not, the -based route will begin with a practical tration by a visiting teacher, trained in

■ .„E procedures.5 As can be seen from Figure
.1, Phases 3, 4, and 5 are similar for both kinds program units. However, in the "text-based" tion, Phases 1 and 2 depend on a story and
■ ike it necessary to determine what the facts are
i dramatization, pictures, etc. By contrast, the
ctivity-based" approach begins with an expen¬
se and then converts it into a discoursal repre-
itan'on in Phase 3. In both cases the discourse
:ms are pragmatically supported by a rich scaf-
.ding of action and dramatization.6 If the scaf-
ding is well developed, there is little danger that
_ ildren will fail to understand what the facts of
■ ; story or activity are. Then, regardless which
- ate was chosen in Phases 1 and 2, by the time
1 ases 3, 4, and 5 roll around, children become
t -ectly involved in the business of creating written . icourse. These last three phases involve negotia-
■ n of a group text (a text created by consensus),
■ ; making of a Big Book (a clean copy of the pre-■. ausly agreed-to group text), and follow-up . mes. This process virtually guarantees total .■ quisition of the material, skills, and social strate-
■ :s connected with all of the negotiations of
■ raning that will have taken place.
Here we come to what is no doubt a critical . :ference between the CLE approaches and more -- iditional methods—namely, that in CLE pro-.-- ims we never give a text or activity a glancing
■ ■■ posure and then pass on to something new and
familiar. Owing to the kinds of games, coopera-
■ e learning (also see Kagan, 1992), and related
. :ivities that always accompany the development
a Big Book, every negotiated text is brought to evel of comprehension that saturates the entire
■ ntext and fully connects it to the experience of
■ :h and every participant. This never occurs (and
'ically cannot occur) in the sort of exchange that
■■ : saw above in Transcript A, but it does 1 ivitably occur with the sorts of repeated cycling
- emplified in the toast-making activity of anscript D. In fact, in a CLE context properly
- veloped, saturation of the entire context with idiiy accessible meanings can hardly be prevent-
- from occurring naturally.
It remains to be seen in more detail what is -" ralved in each of the several phases for the two
.E routes just described for Stage 1. We work " "st through the activity-based route and then
ik at the text-based approaches.

Phase 1 of an Activity-Based Unit
In Phase 1 of an activity-based unit, the teacher demonstrates an activity such as making toast. The demonstration usually involves several steps: (a) talking about what will be happening; (b) showing students the materials and equipment that are needed; (c) showing how the task is carried out; and (d) having students share in the activity with the teacher by performing it for themselves. From the beginning, the teacher and students talk about what they are doing as they do it. This talk is structured, not in accordance with anything laid down in a program text, but in accordance with what is going on. We follow the principle recom¬mended by good writers of showing rather than just telling (Peck, 1980; Swain, 1980). Neither the talk nor the sequence of actions, however, is ran¬domly arranged, but proceeds according to the requirements of the activity itself.
For instance, in making toast, we don't just go willy-nilly through any series of steps, but we work through a fairly determinate sequence of steps owing to the very nature of the activity itself. In consequence, as with any purposeful process, wc end up going through a highly predictable (redun¬dant) series of steps that must occur in a fairly fixed order. As a result, the talk that occurs rela¬tive to the actions being carried out by those who are cooperating in completing the task will similar¬ly be structured around and thus richly scaffolded by an awareness of the steps being taken. In every case, the children and teacher talk about each step in the process, e.g., taking the bread out of the bag, putting it in the toaster, etc. They talk about what is happening while it is going on. They talk about what is needed for each step along the way. They are apt to do this before it happens, while it is hap¬pening, and after it has happened. They talk about who will do, is doing, and has done what, when, and where. They also talk about how they feel about what is happening all along the way. The overall result is lots of comprehended input and lots of negotiated output in a highly redundant, context-rich, affectively positive setting.
Phase 2 of an Activity-Based Unit
Phase 2 in the activity-based route involves absorption of the facts. In this phase we make

Chapter 17 / Literacy in the Third World for AH the Children 173

>wth is assured as the teacher, or studen- 1 lves, direct their attention to differei. f of what is going on or as they bring \n, 'eloping discourse elements of their ow nee, e.g., that Mary's mom puts peani-m toast at her house, and that, when ti¬ts too dark, they just scrape off the blacl rface with a knife, and so on. As a resul an inevitable and natural progression froi going on in the here and now to the richi -■f experience beyond the classroom, soon as the students are confident in tall ut an activity, they will be able to comfor .ake the transition to written form: ;r, since Phases 3, 4, and 5 are common t rivity-based and text-based routes, we no sr Phases 1 and 2 of the text-base :hes.
I of a Text-Based Unit
ie sequence begins with shared reading, th
objective is to get the children to undei
ie Starter Text thoroughly. This mean
e overall structure down to the specifi |
:h as what happened, who did what, an !
iled characteristics of people, objects, an }
Hie reading should be a leisurely, inform; supported by dramatization, pictures, ge; nd realia wherever possible, in order t ! the language comprehension in every ; way. In some cases, where the native lan-': the children differs from the language of and where the teacher knows that primary i, it may be necessary to describe some ele-: the story in the stronger language of the to make certain that the facts of the story understood.
rever, the reading of the text must not merely into a translation. The goal is to hildren to understand the story in the lan-' the text itself. If we want the children to able to read in Thai, translating stories to Malaysian will not get us to our goal. If ren cannot get the gist of the story in Thai :ever the primary language may be), then :ld not be using a text-based unit yet. L who lack sufficient language skills at omprehend the gist of the text will proba-fit most from an activity-based approach

...-il they are somewhat farther along in their lan-
\: ige skills-
However, even in cases where the children i.-vc some significant comprehension of the lan-"." ice of the text, the substance of classroom talk, .■"i to a ^arSe extenl: tne language needed for that r '<, still needs to be supplied to the children. It — st, therefore, be modeled in such a way (chiefly ' rough dramatization) so that at least some of the ^ dents begin to understand and are able to help --,; others to do so. In addition to reading the „-. nr, acting it out, and repeated cycling with new -:•. ps and additional scaffolding, songs and dances - the same theme can also help the children to ■Jerstand and remember the story. Such activi- :. also increase their enjoyment of the whole cess and provide the essential changes of pace - - [ activity that are so important for young chil- . -.n.s Songs and dances that are drawn out of the ; y context can also help the children to under-: id and remember the facts of the story. These . _■ vities can provide a rich form of scaffolding for „ ^prehension. Teachers should be patient with the less-confi- _ t students, giving them praise for non-linguistic - slvements and waiting for their confidence to Id up to the point where they can join in the * :. But first, the children must get the gist of the y in Thai (or whatever language is being used). . - first, this might even be accomplished in their pnmary language, but they must be brought to a point where the language of the story itself is also understood. At any rate, once they become involved as participants, the children will have permanently left their most serious difficulties behind them. Phase 2 of a Text-Based Unit Assuming that students are already fairly compe¬tent in the language of the text, Phase 2 can come almost immediately on the heels of Phase 1, the shared reading of the text. The objective of Phase 2 is to have the students recall what they have heard during the Phase 1 reading. We must make sure they understand what was read and drama-t:zed, etc. This phase usually begins with a warm-'jp period of questions about what happened, who was involved, what song was learned during Phase I. and the like. The teacher elicits at this point a retelling of the story. (Obviously, this step would be too difficult in most cases for students who do not yet have substantial facility in the language the story was told in. In those cases, as noted above, the activity-based approach is preferred.) After the story has been retold once or twice, role play is recommended as a means of involving all of the students in recalling the story and in using the language in which it has been told. The utility of role play will vary as will the point at which it should be introduced. The biggest factor, of course, is whether the language of the story is the primary language of the children. If it is not, role play will probably need to be delayed or sim¬plified. When it can be introduced, role play will give many opportunities for different children to become increasingly familiar with elements of the text. Repeated role plays are excellent devices for instilling language skills in an interesting, familiar, and nonthreatening context. At any rate, by the time Phase 2 is completed, whether the students began with an activity basis or a text basis, they will be ready to proceed to Phases 3, 4, and 5. Phase 3: Negotiating a Group Text By the end of Phase 2, both activity-based and text-based students will be able to draw on shared language and experience to talk about what they have learned. The former group is ready to tell how the activity in question is performed (e.g., what do you do to make toast?), and the latter is prepared to tell what happened in the story (e.g., did the tiger eat the monkey?}. The foundation that has been laid down for Phase 3, therefore, is a whole language experience or text, not some iso¬lated bits and pieces of a possible text or experi¬ence. Therefore, in Phase 3 we do not attend to single isolated words, phonic relations, verb forms, or any other analytic aspect of literacy. Phase 3, like the ones before it, looks to the whole experi¬ence of what you do when you make toast, or what happened when the monkey and the tiger met after they had grown up. The teacher asks the students to tell the story or recount the experience so that the group can write a book for themselves. As in Phase 2, the emphasis is on episodic develop¬ment: "'What happened first?" "Then what hap¬pened?" and so on. The difference is that in Phase 3, she now writes each sentence on a large sheet of Chapter 17 / Literacy in the Third World for All the Children 175 le pages used for writing group texts shoi ;e enough for students to see individual sy -learly, to chorus read from the develop! nd to check the accuracy of reading by otl -.ts. Spaces may be left at appropriate pla, t work to be added later. To decide whi -spaces should be, and how large they y o be, students need to understand the c< id decide how to illustrate it. Since the ne; of a group text is almost never finished ssion, certainly not in the early developmi *es, it is left in place between sessions, wh: : students to discuss it at leisure. Befc ling another session as a group, everyc iie text in chorus up to the point where tf : previously. Here, a student can be asked le part of the teacher, pointing to the woi text as the others read it. soon as the text is completed, students a r read it aloud, considering whether there rt that might be improved. If the teacher 1 ded patiently and carefully in negotiati ct with the students, most of them will ) read the whole text and to identify a )y "reading up to it." That is, the cognit: ntum of knowing how things begin a :omes after each event or description alo y enables the students to benefit from t mcies that episodic organization engenders, vill be able to "recognize" words in their context (in the developing text) before they die point that they can recognize them in :her (and less-familiar) context. Words, >, and other elements presented out of con-ill need to be related back to the familiar text at first in order for them to be easily ized.
4: Making the Big Book
'hase 3 is completed, then a "fair copy" (or ' one, as Americans would say) is made of otiated text. This is the Big Book. Now that ts have learned to work in small groups, in be as many Big Books as there are activi-ps. Additional books will be made, normal¬ise 5. A book that is to be used with the class needs to be as large as a full-sized 3r chart. Books for use with smaller groups only one-fourth that size.

In some parts of Thailand, the "Big Book"
•i is just a few large sheets of paper stapled togeth-
- At Traeger Park School, by contrast, elaborate
- jd-covered Big Books of intriguing shapes were
j. ade. What is done will vary according to local
j-nditions. The procedures used in making books
i Jl also change as students become more experi-
- Ced and become able to work with less supervi-
- )n. In any case, the first task is to decide on page trout and illustrations. This process involves
-: ading the text and discussing what should be in
- e illustrations and where they should be placed
'■i the pages.
As students work in small groups to print the
iok for themselves, the more-able students are
,-lled on to help those who are struggling to read
■■e text9 so that all can share in the decision
- iking. The teacher will have prepared for this,
■ iile the students were still working as a single
r-oup, by modeling helping behavior and then
ving prospective leaders "be the teacher" from
- ae to time. Watching students at work in small
oups making their Big Books reveals much
■ out their ability to read and write. It is particu-rly important to monitor how the groups go out editing their book so that they develop bits of systematic and effective proofreading
■ d self-evaluation.
Phase 5: Games and Other Group Activities
By the time a group of learners have produced their own book, they will be able to chorus read it with ease. Most will be able to read it individually, with minimal prompting. As a group exercise, they will be able to identify any word within its context by chorus reading up to it. They will also be able to do this individually. Some will already be able to identify some of the words in isolation. Most importantly, all will have a firm grasp of the meaning of the text, from its overall structure down to sentences and individual words. The book now becomes a resource for language activi¬ties in which students can focus on the smaller ele¬ments of the written language, such as sentences, words, letters, and even the time-honored phonic correspondences (sound-letter relations). Even spelling and parsing exercises such as sentence dia-gramming may be undertaken on the basis of the Big Book.

What makes all of this vastly more meaningful than it would be in a traditional analytically ori¬ented classroom is that in the background of all these analytic activities is the whole textual basis for the Big Book, which has been, by now, thor¬oughly grasped by the students. The rule for CLE programs is that until such understanding has been established to a high degree of saturation, analytic activities pertaining to such things as phonics (sounding-out-words) and the like should not be undertaken at all. Language games offer many ways to practice language skills and to draw atten¬tion to particular items and features without inducing boredom. As a new activity or game is introduced, the students are first trained to take part in the activity and then to manage it by them-selves, in small groups.
In determining when to move on to Phase 5 activities, teachers generally take a number of fac¬tors into account. Enjoyment and responsiveness of students to the activity are important indicators. Students get more from tasks they like to do. Also, the activities must cover the full range of objec¬tives for each given unit of study. Table 17.1 gives some ideas of the range of games and activities Thai teachers employ in Phase 5 of Stage 1 CLE programs.
CONCLUSION
Of course, determining just when the transition should be made from Stage 1 to Stage 2, where a greater variety of texts will be encountered, or determining when to go from there to Stage 3, where students will learn to produce essentially any kind of text and to learn from the process of doing so, is a Judgment call. However, program experience suggests that two to three years will be needed, as a rule, for each stage. We do not describe those other stages here, as they are dealt with in detail elsewhere (see Walker, Rattanavich, & Oiler, 1992). Instead we will conclude by sum¬ming up our central message.
There is no need for so many children to fail to learn to read either in industrialized nations or in third-world countries. It is simply not true that "ILLITERACY MAY BE INHERITED" as an article in The Australian (a national newspaper) has suggested (June 29-30, 1991). The article

Chapter 17 / Literacy in the Third World for All the Children 177

latching a word or word dents and teacher can col-ohrases of the Bis Book.) m the Bis Book or filling in
ar taking dictation.
the Bis Book to go with a
: competitively by giving out
This may include reading
■etitively by having different tences related to the Big n come up with the latest d by the other team wins, •cutting up elements from sling them in the right order. or poetry, in which the task is
to role play an interview with
newscasters, or masters of
ver popular with children. Tieone's birthday or some
•pie reasoning in such a manner would have recommend special "treatment" of Mary editary illiteracy. But the evidence shows ary and all of her Aboriginal classmates : enthusiastic and capable readers. CLE ches have by now succeeded in enabling iousands of low socioeconomic status (at-lildren to become fully literate. The fact j has been done in contexts where there has long history of gross failure (where only a rcentage of the children became literate) ; that it can be done with any children in any socioeconomic context. : case of the little Aboriginal girl named s not an isolated one. She was able to literate in a non-primary language. When , normal children such as Mary fail, it is

. ■ their failure, but rather that of school language 1 literacy programs. The key to teaching virtual-
". j]j children to read and write in whatever the
...guage of instruction may be is to present lan-ige and literacy tasks in heavily scaffolded con-
... ts where the children can see how the language
- whatever form it may appear fits with whatever ■> ; is going on. If they are provided with such rich
- itexts of experience, the children will not only ,, the connections between linguistic forms and
facts of experience, but they will also readily
- able to determine how texts and discourses of
types fit into the larger power structures of
:ial organization. The children who can read
;■ 1 write in the languages that are used to shape 1 control social structures will themselves be
; -iched and empowered. As this occurs, they will i great extent be released from their "disenfran-
; ;ed" condition.
MOTES
1 The authors gratefully acknowledge funding of roximately $700,000 from the Rotary Foundation of ary International and approximately $100,000 from notary clubs in Australia for the projects in Thailand that are mentioned in this article. We are also grateful to the Srinakharinwirot University, the Thai Ministry of Ecucation, and a number of volunteers from other co.ntries who helped to make CLE literacy programs a reality for children in Thailand and elsewhere in Asia. It has indeed been an inspiration to see people with vision froTi education, from government, and from Rotary, cooperating in an all-out sustained effort that has changed the lives of literally millions of students and many thousands of teachers. Even before the conclusion of [he five-year Rotary-funded project in Thailand, these workers had successfully introduced the CLE approach to Leaching literacy into more than 16,000 schools scat¬tered throughout most of the 71 provinces of that king¬dom. We cannot name all those courageous and dedicated people, but we can say, sincerely, that our .".carts go out to the wonderfully diverse people of Thailand, "The Land of Smiles," who joined in the quest to teach all the children of Thailand to read. 'Alio, Rattanavich and Oiler congratulate their senior colleague and coauthor, Dick Walker, on his being made a Member of the Order of Australia as well as bcs-ig awarded an honorary doctorate in curriculum and

instruction from Srinakharinwirot University for his work on third-world literacy projects.)
2 In fact, as was noted in Chapter 16 by Allen, Anderson, and Narvaez, the same sort of analytical teaching also occurs in many language classrooms with the same depressing results. Building language up from meaningless bits and pieces just doesn't work as a cur¬riculum. It doesn't work for getting children to read, and it doesn't work for getting people to acquire non-primary languages.
3 We are reminded here of Krashen's "input hypothesis" (1985b, 1991, 1992), which says that advances occur when acquirers are pressed just beyond the limits of the language with which they are already comfortable—to what Vygotsky (1934) called their "proximal zone of development" or what Krashen calls "i + I" (the stage just beyond the itb level which they have already achieved).
4 It should be pointed out that some of the pilot programs in the drought-stricken areas had to be placed in schools where the children were suffering from insuf¬ficient and unsanitary drinking water. This problem, therefore, had to be solved before any real progress toward literacy could be made. With volunteer help and funding from Rotary International, it was determined that a reservoir could be dug to a sufficient depth and size to fill the water supply needs. The tank would fill during the monsoon season, and absorption and evapo¬ration were determined carefully by qualified engineers so that there would be sufficient water to carry through the drought season in subsequent years. Drinking water was separated into a distinct reservoir while the main tank was also stocked with fish. It was further deter-mined that a poultry farm could economically be posi¬tioned over the fish farm with the poultry droppings feeding the fish. (It may not sound too appetizing, but it is ecologically efficient.) The children and teachers at the school would manage the fish and poultry farms, thus gaining valuable skills while also meeting the need for an adequate supply of water and sufficient protein in the diet of the children and much of the larger community. In addition, it was possible to cultivate vegetable gar¬dens by using water from the newly created reservoir. So far, over 100 such projects have been completed, and about 1,000 are planned. In communities where this work has been done, the water problem has been solved so that children and teachers are able to devote suffi¬cient attention to their educational objectives. In these communities, a great many literacy activities are built around the management of the fish and poultry farms

Chapter 17 / Literacy in the Third World for All the Children 179


Much has been written over the past sixty years about the place of modern foreign language (FL) study in the humanities and about the contribution that FL study makes to a liberal education.1 A tecurring theme in the post-WWII literature is that modern FL departments can strengthen their posi--ion in the humanities and enhance their contribu¬tion to liberal education by creating area studies programs. Kelly (1969) traces this interest in area studies to attempts made in Army Special Training Programs (ASTP) during World War II to "prevent American soldiers from offending the sensibilities of those whose language they were learning" (p.
going research efforts.
Vll of Parr 1, which discusses ways of creatm ing, is relevant here.
ncidentally, as shown in Chapter 37, these activi . all others like them can only be reported aCCu irough representations of the true narrative type iong and dance are both activities that man ichers are ready to supply at the drop of a hat :e not only willing to invent catchy tunes am horeography, but they are very capable of doini ugh some teachers would no doubt be dauntn suggestion that they should compose a simp|f create a dance to go with it, the response of chil such activities seems to be universally positive re is little doubt that all elementary classroom enriched by task-related, richly redundant activi his sort. It may also be worth noting that ever ;cond language learners seem to be highh ve to songs, games, and related activities in thi nguage (see Part 6). n peer teaching, see Assinder's chapter in Part 5
e a function of what happens in classrooms .elate the discussion to the general tendency ities," and the like to explain high failure
ible to Mary in the contexts of Transcripts lotivations are present in B, C, and D that is greatest among the four transcripts?
the question "Who's the chair for"? How .? By contrast, suppose she encounters the f cues will make the problem solvable? Or and why. Or take a content word such as ould not know it in advance of the toast->Iem in that context?
me for children whose primary language is

CHAPTER 18
Immersion in First-Year Foreign Language Instruction for Adults
Steven Sternfeld
EDITOR'S INTRODUCTION
T
he previous chapter showed that children of low socioeconomic status (i.e., in situa¬tions with a traditional failure rate of nearly 100%) can in fact acquire literacy in a second language provided they are given sufficient scaffolding for the oral and written discourses of the classroom. In this chapter, Sternfeld reports on an approach to first-year foreign language instruction for adults at the University of Utah. He recommends an immersion-type experience which puts the students directly in touch with literature of the target language and culture. Although other authors, notably Brinton, Snow, and Wesche (Chapter 14), Leaver and Stryker (1989), and Allen, Anderson, and Narvaez (Chapter 16), had advocated immersion-type or content-based approaches with adults (following the models employed so successfully with children by the Canadians), extensions of such concepts into the teaching of foreign languages to adults were usually limited to applica-tions with students at the advanced intermediate stage or beyond. However, Sternfeld shows that foreign language students in their first year of foreign language study can also benefit from immersion-type exposure. As in the case of children in difficult socioeco¬nomic settings, the evidence shows that students are often capable of more than the tradi¬tional theories suggest.
INTRODUCTION
379}. On the heels of these ASTP courses, universi¬ties began to develop "language and area pro¬grams," a form of interdisciplinary study drawing on the fields of foreign language, history, and geography.2 Kelly hastens to add that while ASTP courses were the catalyst for the creation of inter¬disciplinary programs in the modern languages, the model for these "new" programs came from the classics:
The traditional classics course includes a thorough treatment of both the history and geography of the Roman Empire and an account of daily life in classical times, in other words, a full area pro¬gram subordinated to the literary and linguistic content (p. 379).

Chapter 18/ Immersion in First-Year Foreign Language Instruction for Adults 181

iency than do students in conventional 1 language programs. Not only do irnrm udents outperform students in traditional ms, but they equal native French speakers -sceptive skills. While communicatively co in speaking and writing, they do not gem each native-like levels of grammatic cy. Finally, immersion students show p0 :titudes towards learning French and -t in continuing to study French; this is st to students in conventional SB coursi Kpress relatively negative attitudes towar French programs and towards learnii in general.-5
Le impressive results of the Canadi; sion Program have already led to the ci )f similar programs in half a dozen scho ts across the United States.6 College I :tion, however, has yet to embrace tl sion model. There has, of course, bei ed interest in incorporating content into ; .ms in the form of authentic material ■er, the rationale for "content-mediatec is, 1987) instruction is that authentic mat cilitate the development of language ski dding learners with more highly contextu; d motivating input.7 Thus, while conter ed instruction and immersion progran i common the use of authentic material ;oals tend to differ: content-mediate tion uses subject matter as a vehicle fi g a second language while immersion pn ise a second language as a vehicle for teaci ject matter.
rnfeld (1988) identifies two notions which xount for the fact that the immersion di itself has failed to make its way into col-programs: 1) the notion that immersion is fective, or indeed only effective, with very earners, and 2) the notion that immersion inly when learners have multiple hours of :ontact over a period of many years, ng to Stemfeld, current research does not t the "too little, too late" argument, risons of proficiency levels attained by . late immersion students have shown that { cases children who begin immersion in ; are able to catch up with children who n kindergarten (Genesee, 19S5, 1987; 1978; Lapkin et al., 1983). Moreover, the

■earch on the effectiveness of short-term (e.g.,
-.semester), non-intensive subject-matter teach-
• z has shown that students in subject-matter
urses make linguistic gains equal to those of stu-
"i.Cts m SB courses and also learn their subject
fitter (Edwards et al., 1984, Lafayette and
■ iscagliaj 1985). Stemfeld concludes that subject-
i-jtter learning could conceivably be "pushed
. -.^rr (Krashen, 1982: 173) to the beginning level
■ college FL instruction provided that two condi-
- ms are met: 1) that expectations concerning ini-
- J production and comprehension are adjusted . -wnwards, and 2) that compensatory pedagogi-; [ strategies are employed. Such strategies include
- rusmg initially on comprehension, not forcing
- oduction, and modifying the format of written
- jignments and tests.
THE UNIVERSITY OF UTAH'S PILOT IM/ML PROGRAM
te goal of the pilot IM/ML Program at the
. liversity of Utah was for Anglophone students to
: Spanish in an exploration of some facet(s) of
. tin American studies. In these "sheltered" class-
- ims,s Spanish was the language of instruction
■ lile Latin American history, geography and cur-
- it events constituted the content of instruction.
,ther than learning to survive in a Spanish-
;aking country, IM/ML students had to learn to
rvive in a Spanish-speaking classroom.
In 1985-86 two first-year IM/ML courses in evanish were piloted by a professor in the Department of Languages and Literature. One of these courses met one hour a day, five days a week for three quarters; the other was offered as a sum-mer intensive which met three and one-half hours a day for seven and one-half weeks. Eighteen stu¬dents completed the year-long course; 25 complet¬ed the summer intensive course. To avoid self-selection and thereby enhance the generaliz-abtlity of the results of this pilot program, neither pilot course was previously advertised as being part of the IM/ML Program. No students chose to withdraw upon being informed of the nature of the course, and attrition for the first-year IM/ML ar.d SB programs was comparable.
A questionnaire was administered to all stu¬dents completing the IM/ML Program and the con-

ventional SB program at the end of the first year. Students were asked to describe previous language instruction (school, year, and length of instruction) and informal contact (both in the United States and abroad) for all languages ro which they had been exposed. These data indicate that absolute beginners were actually in the minority in both the IM/ML and SB Programs. In the case of false beginners, the fact that these students1 initial expo¬sure to Spanish had usually been brief (one or two years) and had not immediately preceded their enrollment in college Spanish would explain their presence in these first-year courses. Many students also indicated some informal contact with Spanish through friends, relatives, work, and/or rravel. Over half the students had had some exposure to another foreign language besides Spanish.
DESIGN OF INSTRUCTION
In keeping with the principles of the Canadian immersion approach, none of the four language skills was explicitly taught initially. Thus, there was no underlying linguistic syllabus (grammati¬cal, notional/functional, or situational), nor were there any vocabulary, structure, or pronunciation exercises. Following Krashen (1982, 1985b), it was assumed that the "comprehensible input" which would result from students interacting with the spoken and written language would lead to their acquiring Spanish.
Four activities were designed to promote stu¬dent interaction with oral and written discourse: 1) daily reading assignments, 2) pre-reading lec¬tures, 3) classroom discussions, and 4) subject-matter quizzes.
1 Daily Reading Assignments
Extensive daily reading assignments constituted the core of the IM/ML Program. These assign¬ments were "purposely longer. . . than students can study crypro-analytically" in order to encour¬age rapid sketchy reading (Newmark, 1971: 16). For rhe study of history and geography, we adopt¬ed Spanish-language texts typically used in upper-division introductory civilization courses.9
For current events, students received five issues a week of La Opinion, a Spanish-language



Chapter 18 / Immersion in Fii

:-Year Foreign Language Instruction for Adults 183




:course to a dictionary. As students' linm, etence and background knowledge increaj. ;ed for such comprehensive pre-reading ac*r' vas considerably reduced. By the end of - '* e, pre-reading lectures were presented o-'" the subject of the reading assignment v--o the students.
;cussions of Reading Assignments
ler to compensate for students' minimal cc -. nsion skills in the early stages of the p -. most of the class period was given over ading lectures; relatively little time was ■■ iscussion of the reading assignmer-over, in order not to place too grea-. id on student production, early discussi -the form of a question/answer period -students could ask the instructor for clai of words and phrases in the previous nig ig assignment. Students were able to !e -y the simple formulaic language necess s type of interaction. ; the need for comprehensive preparation ading assignments diminished, more t -be devoted to discussion. Furthermore .ts* linguistic competence and subject-ma -edge increased, and as more abstract, is.c -:ed readings were assigned, discussii [ beyond teacher-centered textual clarif ' highly interactive group discussions invi nmentary and analysis.
>ject-Matter Quizzes
ts were quizzed daily on the reading ass: . The purpose of these quizzes was age students to keep up with the extern eading and to provide a means for teac idents to monitor progress in reading ci sion and subject-matter knowledge. Lree types of quizzes were used in of the one-year program. Initially, stu-were asked to write an interpretation in h of the contents of several paragraphs d from the previous night's reading, 60-90 Is being allotted for each paragraph, laries were not allowed, and the only grad-iteria were coherence and plausibility.


*
our instructional strategies—the designa¬te quiz as an "interpretation," the impo-f a rather severe time limit, the exclusion , paries, and the rather lenient grading cri-ivere intended to encourage students to luickly a global meaning to the text, even personally assigned meaning was at vari-ith the actual meaning of the text. Of early interpretations were very impres-, as students relied almost exclusively on identifiable cognates, context, and what-ckground knowledge they brought with the course.

:r about 75 contact hours, students' lin-competence and background knowledge veloped sufficiently to allow greater s to be placed on accuracy and complete-was at this point that the interpretations :lassified as translations. It should be kept 1 that the ability to produce accurate ■to-English translations was not an explic-of the IM/ML Program. As previously led, the purpose of the interpretation was to encourage students to keep up with ing assignments and to monitor the devel-of students' reading comprehension skills kground knowledge. In order to maintain le focus with the translation quizzes, it mtial that students not be allowed to use ties.
mt two-thirds of the way through the short-answer quiz format was introduced, pen-book quizzes required students to ort responses in Spanish to simple ques-the material in the newspaper and the civ-textbooks. At first these questions were on the board; later they were dictated to ;nts.
PROGRAM EVALUATION
i.wu questions were of primary concern in our evaluation of the pilot IM/ML Program:
• What was the program's impact on students' Spanish-language skill development?
• What was the program's contribution to stu¬dents' liberal education?

Spanish Language Skill Development
Students' Spanish language skills were assessed for the purposes of ongoing evaluation of the IM/ML Program and articulation with the SB Program.11 Thus, the same battery of tests was administered to all first-year students, i.e., those in the two sec¬tions of the pilot IM/ML Program and those in the four sections of the conventional SB program taught by graduate students under the supervision of a Spanish faculty member.
The considerable differences in the goals, curriculum, instructional strategies, and internal evaluation procedures of the IM/ML and SB Programs posed particular problems for test development. Given the problems inherent in "program-fair evaluation" (Beretta, 1986) and constrained as we were by practical considera¬tions in the development, administration, and scoring of language tests, we initially limited ourselves to developing tests to measure just three skills: reading comprehension, listening comprehension, and extemporaneous writing. We adopted a combination of strategies to mini¬mize test bias, including the use of pragmatic tests (Oiler, 1979), the careful manipulation of test format and/or content both within and across tests, and the development of a dual scor¬ing system.
Two reading tests were devised to measure extensive and intensive reading skills. The Reading Summary (RS) required students to sum¬marize in English an essay written in Spanish, while the Reading Translation (RT) involved the translation of a short prose passage from Spanish into English.12 In an analogous fashion, two tests of listening comprehension were developed. The Lecture Summary (LS) required students to sum¬marize in English the contents of a videotaped lecture given by a professor of Spanish on the his¬tory of Spain during the reign of Philip II; the Dictation (DT) was based on a passage taken from the videotaped lecture. Writing was assessed with a single measure, a Writing Sample (WS). Students were given a prompt in Spanish on eating habits in Latin America and then asked to write, in Spanish and without the aid of any reference materials, about their own eating habits.

Chapter 18 / Immersion in First-Year Foreign Language Instruction for Adults 185


8,1 Summary of Mean Scores
Test
Group

ore
rtert iicy
■ on only one of the five tests, the W >le.
7he question remains, then, whether thi t reflects actual differences in the dev^,,, of writing skills or whether it is an artifaci lesign. Indeed, the design of the writing i rather problematic, given the very differ which writing plays in the two prograi
the primary function of writing in the ■am is to provide students with control ice in the use of the vocabulary and str presented in their text, writing for SB s
tends to invoke heavy monitor use ( en. 1982, 1985b, for a discussion of moni [n contrast, given the focus on subject ma le lack of formal language instruction in L Program, IM/ML students were essenti; ible of monitoring their writing for grj il accuracy.
ie decision to test extemporaneous writ was made in an attempt to minimize mon uereby ensuring a more equitable basis irison of the two groups. Yet this decis '. SB students at a disadvantage, since t . experience with unmonitored writing, msate, a guided composition format \ i using a topic discussed in some detail in dents' textbook (eating habits). In the e ay have placed IM/ML students at an e ■ disadvantage. Since the subject of perse habits was not part of the content of 1 course, many IM/ML students had 'ior experience in handling this subject in h. As a result, IM/ML students were less o have the appropriate lexical items in their 'ocabulary. Finally, the overall greater accu-: SB students may reflect the fact that SB ts, unlike IM/ML students, could rely in
memorized routines, the accuracy of which ly surpassed that of their developing inter-


RT
RS
DT
LS
SB (n = 46)
IM/ML (n = 36)
SB (/i = 46)
IM/ML (n=36)
WS

.16 -.23 -.17 -.10 -.14
.03 .33 .19 .16 .12
.02 .08 -.05 .12 .23
.11 .01 .06 -.16 -.33
TA3-E 8.2 Multivariate Test
df
.0167* .1057
Wilks-Umbc/a Rao's R (Form 2)
2.968205 1.886825
5,76 5,76
.8366264 .8895743
" iBLE 18.3 Univariate Tests of Content Scores
df
F-Ratio

Translation .3343 1,80 .5717
Summary .0930 1,80 .7551
.2418 1,80 .6299
■ummary 1.5886 1,80 .2086
6.5588 1,80 .0119*

ie aim of the pilot IM/ML Program was to
: the contribution of introductory FL study
il education, we were particularly interested
assessing the extent to which students had been
:ellectually stimulated and culturally broadened
their study of Latin America. An open-ended
;ay format was chosen as most appropriate for
r purposes. Open-ended questions are a data
thering tool frequently used in qualitative evalu-
on (Patton, 1980).
Since students' views on the meaning of a Hb-il education differ considerably, we did not empt to query students directly on their percep-ms of the contribution which the IM/ML
Program had made to their general education.13 Rather, we chose to ask students to describe what they felt they had learned in the course; from this data we hoped to draw our own conclusions con¬cerning the program's contribution to students' liberal education.
The essays were written in English during the last week of the program. Students were asked 1) to trace the development of their understanding of and attitudes towards Latin America over the course of the pilot IM/ML Ptograra, and 2) to evaluate each component of the course (newspa¬per, civilization texts, lectures, discussions) in terms of its contribution to these changes. This second evaluation was included to provide pro¬gram developers with feedback on the perceived utility of the various instructional activities.

Chapter 18 / Immersion in First-Year Foreisn Language Instruction for Adults 187

irning Outcome 4: Students developed awareness of the complexity of issues with hey were already familiar:
thought that I understood the problem of 1 immigration in this country, but reading tides on the "paramilitaries" from Alabama :he citizens' arrest made in Arizona really d my view. I started to look at illegal immi->n not as the problem, but as a result of the em, which is poverty.
raing Outcome 5: Students, expressed an in further developing their understanding =tnic culture:
;reat thing about this class is that I can now lot of questions. I have a subscription to La ion and I will continue to look for these ;rs because I am now very interested in what ng on in Latin America. I will also be search-e city and talking to Latins here in Salt Lake : their viewpoint and how they fit into the ■e.
iSION
hat it had been our intent to enhance the ition that introductory FL study makes to lents' liberal education by promoting mul-y, which we defined as the pursuit of intei-y challenging and culturally broadening :s in more than one language. Assuming se self-reports are not totally unfounded, irs that the IM/ML Program has indeed I important contribution to undergraduate >n.
efinitive response to this question is, of not possible on at least two grounds. First, isence of pre- and post-tests, we are unable rm whether more objective measutements ing outcomes would substantiate students' ions. Second, the above-mentioned out¬lay be criticized as failing to fulfill the role raditional SB foreign language study has ;n the undergraduate curriculum, namely i principled examination of the structure of ;. Nevertheless, those who support an area approach to FL instruction will consider utcomes central to the aims of liberal m.

CONCLUSION
e initial results of the Immersion/ Multiliteracy )gram indicate that subject-matter teaching can |eed be "pushed down" to first-year language arses. A comparison of the reading, listening, d writing skills of IM/ML and SB students jws that IM/ML students are, for the most part, iping up with their counterparts in the SB pro-un. Moreover, data gathered from IM/ML stu¬nts' English-language essays can be interpreted lending support to the claim that an area studies proach can enhance the contribution that begin-ig FL study makes to liberal education.
These preliminary results encouraged us to )ceed with the development of the IM/ML jgram. In 1987 we received a grant from the partment of Education to improve the design, plementation, and evaluation of the IM/ML agram. We now have IM/ML Programs in inish, German, French and Chinese and will be dnning a course in Japanese this fall. Subject tter varies from course to course, for the most t the result of the availability and/or accessibili-of reading material. The Spanish course, in iition to La Opinion, now uses the Pequeno rousse Ilustrado (1976) instead of the civiliza-i textbooks to present background information Latin America. The French course has adopted tmal Historique de la France (Billard, et al., 36), a history of France written in journalistic mat. The German course has been built around -volume world history text, Unsere Gescbichte jg, 1985), which is used in secondary schools in rmany. For the Chinese course, special materi-
have been developed under the OED grant,
luding an outline of Chinese history, an :rview of lifestyle in modem China, and stories ;ed on classic Chinese proverbs.
Additional university funding was received in JS to aid in the development of a master cur-ilum for the languages taught in the IM/ML >gram. Scholars from the fields of history, fine s, political science, sociology, etc. will be asked iraw up a program of study that will 1) provide
Department of Languages and Literature with delines for the training of teaching assistants in " MAT (Master of Arts in Teaching) Program o serve as IM/ML instructors, and 2) provide 'ML instructors in turn with a framework for

designing new curricula for IM/ML courses. Considerable work lies ahead; nonetheless, we are convinced that an area studies approach to first-year college FL instruction holds promise for enhancing the contribution that FL study makes to liberal education.
NOTES
1 An initial search has produced nearly 100 articles written over the past sixty years on this topic. The vast majority of these articles have appeared in The Modern Language Journal, the ADFL Bulletin, and the French Review. A review of this literature is in progress.
2 There is considerable hesitation as to the appro¬priate label for such programs, the three most common ones being "area studies," "interdisciplinary studies," and "multidisciplmary studies." Although not widely used, the term nondisctplinary is perhaps the most appropriate. Coined by classics professor Harald Reiche, it was intended to describe tests which "predat-ed—and even called into question—the very intellectual division represented by our [modern day] disciplines" (Elbow, 1986: 5). Such variations notwithstanding, these programs may be intradepartmental or interde¬partmental (cross-curricular) in conception. A number of perspectives serve as a focal point of area studies pro-grams: ethno-linguistic [e.g., Slavic Studies, Germanic Studies), socio-cukural (Women's Studies, Afro-American Studies), geographic (e.g., Latin American Studies, Asian Studies), historica (e.g., Renaissance Studies, Medieval Studies), and sociopolitical (e.g.. International Studies, Multicultural Studies). For a gen-eral discussion of area studies programs and depart¬ments of language and literature, see Champagne (1978), Dease (1982), Giamarti (1979), Jorden (1982), Jurasek (1988), Lee (1983), Nelson (1972), and Tonkin (1985). For a discussion of foreign languages and International Studies, see Burnett and Robinson (1985), Dirks (1976), Haenicke (1979,1976), and Knoll (1976).
3 The term multiliteracy is borrowed from Hirsch (1987), who contrasts it with multilingualism. As Hirsch points out, one can be multilingual without being multiliterate.
4 In this sense, multiliteracy is closely tied to multi-culturalism, although the latter does not necessarily imply a background of shared literate knowledge. In the field of ESL, EAP (English for Academic purposes) Programs have as their aim multiliteracy, since their

Chapter 18 / Immersion in First-Year Foreign Language Instruction for Adults 189


edge of these conventions aids readers2 in com-prehending the text as well as in recalling it later (Kintsch & van Dijk, 1975, 1978; Meyer, 1975; Thorndyke, 1977). Following Bartlett (1932), this knowledge has been called a schema, or more specifically, following Carrell (1983a), a formal schema.
'e course, the year-long section used only one his-Dgraphy text while the summer intensive cour^ ree.
The fact that more than one student likened the or's early pre-reading lectures to a game of "cha-conveys a sense of the role played by nonverbal nication.
Indeed, the long-term viability of the IM/ML a depended in part on IM/ML students develop-[ levels comparable to those of students in firsr-< courses. Since it was not clear how the two would co-articulate at the beginning of the sec-ir, one section of second-year Spanish was set r continuing EM/ML students. This two-quarter e was taught by a professor familiar with the d methodology of the IM/ML Program. The essay "El Conflicto entre los sexos" was ■om Perspectivas (Kiddle & Wegman, 1988), rt prose passage was taken from the opening )h of Macctrio by Bruno Traven (1971). Sfevertheless, as Morello (1988) points out in a tide, we should not overlook the importance of ionship between students' perceptions of their and their attitudes toward foreign language ee also Rivers (1985), who admonishes FL to "look with our students' eyes at the way our m appears to them" (p. 42). 7or example, one student, whose husband was ;h/Spanish bilingual who only used English at uted that she was finally able to speak to her i-law on the telephone. ;" argument against teaching foreign lan-claim that children before a certain age or stories? What kinds of representations e? What other sources of information will elp students relate the news to their own d's study over the immersion/multiliteracy the other tests. What other sorts of testing ison between two such diverse groups of al bases contribute to a liberal education? ents in Sternfeld's study? CHAPTER 19 Evidence of a Formal Schema in Second Language Comprehension Patricia L. Carrell EDITOR'S INTRODUCTION E xpectations concerning what a given text or discourse may represent are apparently of two kinds. The first relates to particular facts and can be directly associated with what we would call "content" (the left side of Figure 37.1), and the second kind pertains to representations themselves and what we call "form" (the right side of Figure 37.1). Both kinds of expectations are called by many names, including schema (by Piaget, 1947; and Carrell, for instance), models (Johnson-Laird, 19S3), episodic structure (Glenn, 1978), scripts (Schank, 1975, 1980; Schank & Abelson, 1977), plans (Miller, Galanter, & Pribram, 1960), expectancies (Oiler, 1974, 1983b), and so forth. Depending partly on the vantage point we assume, we may prefer the term plan (from the producer's side) or expectancy (from the receiver's). Regardless, at their basis, the kinds of structures and processes involved seem to be very similar if not the same. Therefore, perhaps an abstract term (which is biased neither toward producer or receiver, nor toward content or form) such as schema ought to be preferred. Carrell chooses such an approach. However, as she shows in this chapter, it is useful to distinguish between content-oriented ("content") schemata and form-oriented ("formal") schemata. For reasons given in Chapter 37, the most basic "formal" schema is of the narrative type and happens to be the kind Carrell looks at in this chapter. The essence of a narrative schema (a kind of grammar of expectancy and planning) is some person experiencing events unfolding in time and space. An example in this chapter of such a temporal unfolding is one of the unlucky twins losing her dollar bill and suffering the consequences, or it is some other character struggling through some other connected series of events. The key (Dewey, 1916) is development over time. Here, Carrell shows that the formal schema for narratives is just as important to non-native speakers of English as it is to native speakers. Subsequent chapters of Part 4 continue to bear out this finding. Research on discourse or text comprehension has shown that comprehension is determined not only by the local effects of sentences or paragraphs, but also by the overall suprasentential or rhetori¬cal organization of a text.1 Each type of text— e.g., stories, fables, expository and scientific texts—has its own conventional structure; knowl- Chapter 19 / Evidence of a Formal Schema in Second Languase Comprehension 191 rURc 19-1 - Underlying Structural Schema for Stories Used in This Study Story inning 2 Development Ending 2 7 /\ 11 31 Development Ending 1 6 Reaction 2 Goal path Attempt 2 9 Attempt 1 4 rs under each story node = input order of nodes in standard versions; adapted from Mandler [1978a] d with permission.) prehension. Within the context of the sti¬ll schemata3 (background knowledge ( ■rical structures of different types of :1I, 1983a), narrative, story schemata ai nly rhetorical schemata of interest and : to ESL reading comprehension. The t Cerent types of expository rhetorical o 1, in the sense of Meyer (1975, 1977, r, Brandt, &C Bluth, 1980; Meyer & Meyer &c Freedle, 19S4), are another ii rea of investigation. A number of studi ning to emerge to demonstrate the rele L reading and writing of the investigate,, d, rhetorical schemata—for example, i ch of Connor and McCagg (19S3a, 1983 aring the performance of Japanese a ih readers of ESL on an expository text w n rhetorical pattern; the research of Hir , 1983a, 1983b), comparing Japanese a h readers, reading in their respective nat iges, on texts with a typical Japanese rhetc ucture; the research of Burtoff (1983), cc ; the different rhetorical patterns of Engl tory prose produced by native speakers ise, Arabic, and English; and the research II (1983c), comparing the written rea lifferent types of expository prose by difl rive language/cultural groups of ESL readt these studies illustrate the importance gating formal, rhetorical schemata in E % comprehension. cause narrative schemata have been claiir developmentally prior (Freedle &c Hz Glenn, 1978; Stein & Glenn, 1979) :ory schemata, and have also been clain note cross-culturally universal and less c ecific than expository schemata (Mand =r, Cole & DeForest, 1980)—although t claim is not without debate (Kintsch , 1978)—it would seem important to t : effects of simple narrative schemata :hension in English as a foreign or secc ?e. To my knowledge, no previous resea. type has been conducted in ESL. SCHEMATA DURING ENCODING schema may be thought of as the rhetori-cture or the grammar of a story. It has even xr. characterized as a grammar which generates tree structure consisting of labeled nodes, the >nstiruents of a story (see Figure 19.1).
As can be seen in Figure 19.1, each story con-;ts of a common Setting plus an Event Structure, h:ch consists of a series of causally or temporally jnnccted Episodes. (The stories used in this search consisted of two temporally connected •isoccs, Episode 1 and Episode 2, as illustrated in g'.'.re 19.1.) Each Episode consists of a Beginning, Development, and an Ending. The Development insists of a Reaction and a Goal Path, the latter msi.sting of an Attempt and an Outcome, andler (1978a: 16) defines these simple story ■mponents or nodes as follows: A Setting usually ■nsibts of stative information about one or more the characters, usually their location in time and ice; :he Beginning is usually some action or event mat begins the story; the Reaction is an internal reaction of a character caused by the beginning event—the reacting character becomes the protago¬nist of the episode; in the Goal Path, the protago¬nist formulates a plan to deal with the problem created by the beginning; the Attempt comprises

the efforts and actions of the protagonist to try to reach that goal; the Outcome tells us whether the attempt was successful or not; the Ending usually consists of some further consequences or emphatic resolution to a series of events, or may also include a reaction on the part of another character. Only the terminal nodes in the diagram are actually real¬ized overtly as segments of the text. The other, nonterminal nodes, as in a sentence diagram, repre¬sent the hierarchical organization of the story.
The story schema may also be thought of as a set of expectations about stories, about the units of which they are composed, the way those units fit together and are sequenced.
This cognitive schema guides both the encod¬ing of a story during input (i.e., as it's being read) and during retrieval (i.e., as it's being recalled). During encoding, the schema provides a frame¬work within which the incoming material may be structured; it helps the reader know which aspects of the material are more or less important or rele¬vant, and it also lets the reader know when some part of the story is complete and can be stored, or that some proposition is still incomplete and must

Chapter 19 /Evidence of a Formal Schema in Second Lansuage Comprehension 193


3THESIS
nain question in this study was whether V :ts are influenced by a simple story schetn rehending stories read in their second :, English. The question was addressee aring recalls of two types of stories— tructured and the other deliberately vio|a -" :quence of events prescribed by the si ■ ia. It was hypothesized that if ESL subj fluenced by a story schema, then when re read which violate the expected si a, quantity of recall and temporal sequ recall will be affected.
BLE19-1 An Example of a Standard and Interleaved Story
Each standard story was then rearranged to create an interleaved version. In these versions, fol¬lowing the Setting, the five basic nodes of each episode were presented in interleaved fashion: two Beginnings (Beginning 1, followed by Beginning 2), two Reactions, two Attempts, two Outcomes,
OD
Cts
ESL subjects participated in the experimt were intermediate-level (Level 3) ESL i -enrolled in the Center for English a I Language (CESL), an intensive progr ^matriculated foreign students at South > University at Carbondale. Their ove: 1 proficiency was in the range of 50 to Michigan Test of English. Their native I backgrounds included Arabic (n = 1 i {n = 8), Malaysian/Indonesian (n = 8),; se (« = 5), as well as others (African, Gre i, Chinese, and Korean).
als
;imple two-episode stories were construct ling to the story grammar outlined :r and Johnson (1977). In fact, one of i was one of four used by Mandler (197! irst language research with both adults and n. Figure 19.1 illustrates the analysis of Dry, Each story has a common Setting, fol->y two temporally connected episodes. This ■e is a loose one, in which the protagonist ■nts of the second episode bear only a tern-Nation to the first. However, in all the sto-
events made a sensible whole, and the two nists—one per episode—were closely relat-
were united by the common Setting in ley were both introduced.

STANDARD VERSION
Setting: Once there were twins, Tom and Jennifer, who had so much trouble their parents called them the
unlucky twins. Beginning 1: One day, Jennifer's parents gave her a dollar bill to buy the turtle she wanted, but on the way to the pet
store she lost it. Reaction 1; Jennifer was worried that her parents would be angry with her so she decided to search every bit of the
sidewalk where she had walked. Attempt 1: She looked in all the cracks and in the grass along the way. Outcome 1: She finally found the dollar bill in the grass.
Ending 1: But when Jennifer got to the store, the pet store man told her that someone else had just bought the last turtle, and he didn't have any more. =■ Beginning 9.-. The same day, Tom fell off a swing and broke his leg.
Reaction 2: He wanted to run and play with the other kids. ~ Attempt 2: So he got the kids to pull him around in his wason. C Outcome 2: While they were playing, Tom fell out of the wagon and broke his arm. " Ending 2; Tom's parents said he was even unluckier than Jennifer and made him stay in bed until he got well.
INTERLEAVED VERSION
Setting: Once there were twins, Tom and Jennifer, who had so much trouble their parents called them the
unlucky twins. Beginning 1: One day, Jennifer's parents gave her a dollar bill to buy the turtle she wanted, but on the way to the pet
store she lost it. Beginning 2: The same day, Tom fell off a swing and broke his leg. Reaction 1: Jennifer was worried that her parents would be angry with her, so she decided to search every bit of the
sidewalk where she had walked. Reaction 2: Tom wanted to run and play with the other kids. Attempt 1: Jennifer looked in all the cracks and in the grass along the way. Attempt 2: Tom got the kids to pull him around in his wagon. Outcome 1: Jennifer finally found the dollar bill in the grass.
Outcome 2: While the kids were playing, Tom fell out of the wagon and broke his arm. Ending 1; But when Jennifer got to the store, the pet store man told her that someone had just bought the last turtle,
and he didn't have any more. Ending 2; Tom's parents said he was even unluckier than Jennifer and made him stay in bed until he got well.
(from Mandler [1978a], used with permission)
and two Endings. In these sets of paired nodes, the node from Episode 1 was always presented first. An example of a standard and an interleaved story and its division into the basic nodes is shown in Table 19.1. Appendix 19.1 contains the complete text of all three stories, in both their standard and

Chapter 19 / Evidence of a Formal Schema in Second Language Comprehension 195

sages corresponded to the order of the origj Ties for each subject. Subjects were asked t, heir recall of each story, writing as much a Duld remember of each story, as exactly a Duld remember having read it, in the sam in which they read the stories. They wee • write as well as they could, but not to b concerned with their writing. They wer at not only would they not be given a grad ir writing, but that, in fact, their writin :s wouldn't even see what they wrote. W sized that we were interested in how mud >w well they could write it down. As wit! ; time, writing time was subject-control lee fer, as a group, they spent approximate! JO minutes writing their recalls of all thre
g Procedure
; criterion4 of recall was used. Two rater; agree that the essential meaning of a storj :ad been reproduced, allowing possible distortions or additions in meaning. :er reliability exceeded .90. Disagreements tiled by a third rater. Distortions included owing: (1) recall of only a subsidiary part ie, possibly even leaving out the main event described; (2) repetitions of previously 1 nodes, in which other nodes were pro-before the repetition; (3) character and infusions, consisting of importations from sode into another, or from one node into
within the same episode. The most typical ;r confusions were a single character being
the protagonist for both episodes or a :r from one episode appearing in the other, as were of the following types: (1) redun-
i.e., saying the same thing twice, often in y different way; (2) reasonable presupposi-
which obviously inferable material was o the original node—for example, in the ■ Twins story (see Table 19.1), Attempt 1 iclude something like "She retraced her lefore continuing with the original node; gerations, in which a statement was inten-
some way; (4) irrelevant or wrong infor-added to an otherwise correctly recalled lis last kind of information was in contrast nable presuppositions and exaggerations.

■ iBLE19-2 Quantity °f Recall
Intermediate ESL N = 40

Standard Interleaved
n = 20 n = W
90% 87%
(54/60) (52/60)
n.s.
isodes recalled
90% 83%
(97/108) (86/104)
n.s.
des Recalled
75% 68%
(402/539) (330/482)
■ 5<.05 Although these were frequently related to the story line, they could not logically be assumed to be the case. For example, in the Unlucky Twins story the subject might have written for Outcome 2 "Tom broke his arm and had to go to the hospital." The only type of additions which were not included were those in which irrelevant or wrong material was substituted for the essential meaning of an entire node. For example, in the Unlucky Twins story it might have been recalled that the girl found her dollar, and not being able to retrieve the Ending, the subj'ect might have written "She felt glad and went home." Thus, if there was virtually any recognizable piece of the meaning of the node present in the recall, the node was scored as being present. Otherwise, it was scored as missing. RESULTS Quantity of Recall The results indicating the quantity of recall are given in Table 19.2. L1 adults and children N=96 (Mandler, 1978a) Standard Interleaved n = 48 n = 48 89% 89% 96% 67% 72% In addition to reporting the results for the ESL subjects of this study, I will also indicate how they compare to the first language results obtained by Mandler (1978a). However, I will make no statis-tical comparisons outside the ESL group. Furthermore, it should be noted that Mandler's first language subjects operated in the oral medi¬um (listening to and retelling the stories); these ESL subjects operated in the written medium (reading and writing the stories). It was noted ear¬lier that Mandler's subjects also had a 24-hour recall interval. The same number of stories was recalled in both the standard and the interleaved story condi¬tions. A story was counted as having been recalled if any part of it was recalled. The ESL subjects in this study recalled 54 out of 60 (or 90%) of the standard and 52 out of 60 (or 87%) of the inter¬leaved stories. (N.s. in Table 19.2 indicates that the pair-wise comparisons are not statistically sig¬nificant according to the test of significant differ¬ences in proportions from independent samples. If the difference is statistically significant at the level p < .05, a double asterisk is shown.) Thus, there Chapter 19/ Evidence of a Formal Schema in Second Language Comprehension 197 19.3 Mean Number of Modes Recalled per Episode qgUREl9.2 Types of Nodes Recalled 1978a, 1978b) for native English speakers. In act, Figure 19.2 illustrates a remarkable degree of onsistency in the pattern of type of nodes recalled letween Mandler's native English subjects and iSL subjects in this study. The only difference in he patterns between the ESL subjects and "handler's native subjects was in the Ending nodes. "he native English subjects tended to recall the mdings less well than all other nodes with the xception of Reactions; the ESL subjects tended to ecall these somewhat better than Outcomes and ven Attempts. Temporal Sequence of Recall )f central interest is the order in which the story lodes were recalled. As Mandler (1978a) found for native English speakers, the ESL subjects' recall of the well-structured stories followed input order, almost without exception. Although there were nodes missing from episodes, there was no inter- d Interleaved t-Test independent samples 2 M = 3.21 r=2.06 >n SD = 1.493 df= 182
1,5 Range = 1,5 p<.05 pisodes N = 86 episodes Dn. After a 24-hour recall interval, a greater ■ of nodes were retrieved for the standard (M = 3.62) than for the interleaved stories 21). Mandler (1978a) reports a similar sig- difference for her native English speaking tnd children taken as a single group; in her more nodes per episode were recalled from d stories (M = 3.5} than from interleaved A = 3.2). However, her native English g children behaved quite differently from .Its. Her second-graders recalled relatively either type of story (M = 3.0 for both ver-idults showed equally good recall of both 4 = 3.8 for both versions); and her fourth-:h-graders were both quite good at recall-dard stories (M = 3.6) but had more trou-lling interleaved stories (M = 3.1). On this :, the ESL subjects in this study are most ndler's fourth- and sixth-graders. final analysis carried out on quality of 'as on the types of nodes recalled. Figure )ws that some types of nodes were recalled 2tter than others. The abscissa represents 1, schematic order. In the case of the stan-ries, this also represents the input order; in of the interleaved stories, this does not t the input order. More will be said about 'oral sequence of recall in the next section. obvious in Figure 19.2 that Reactions— is the protagonist's internal reaction to the ng (a thought or emotional response, ere the least well recalled; action nodes better recalled. As Figure 19.2 illustrates, ing also corresponds to Handler's finding 1.00 .90 .80 .70--.60--.50 .40--.30--.20 -- .-o -- 0 Attempt Outcome Ending Setting Beginning Reaction Standard ESL (.94, .90, .54, .66, .73, .79) A Interleaved ESLC92, .91, .38, .53, .69, .76) • Standards Interleaved L1 Adults(.98, .85, .51, .85, .90, .64) (Mandler, 1978b) approximate proportions leaving of the episodes; nodes from each episode were recalled contiguously. Further, Episode 1 was routinely recalled before Episode 2; from all 20 ESL subjects recalling stories, with a total of 54 stories recalled, there were no recalls in which Episode 2 was recalled before Episode 1. Within an episode, if the nodes were recalled at all, they were recalled in the standard order, with rare exceptions. The only exceptions were that four times an Attempt preceded a Reaction and two times a Reaction preceded a Beginning. Of particular interest is the extent to which subjects in the interleaved condition followed the input order or the ideal, schematic order in sequencing their output. If a story schema affects second language retrieval, then the evidence of that effect is the extent to which the interleaved versions were recalled according to the ideal order rather than to the input order. One measure of this evidence was calculated by taking those recalls in which at least one node Chapter 19 / Evidence of a Formal Schema in Second Language Comprehension 199 adjacent Nodes and iy , E1 ■2,E2 78 54 ,02 E1,E2 2, E1 0 .42 5 .41 ed. To determine whether certain pairs of t nodes in output were more likely to occur :hers, two measures were computed (see 9.4). The first measure was the frequency structurally or ideally adjacent nodes from e episode (e.g., Reaction 1 and Attempt 1, :ome 2 and Ending 2) recalled together in sequential order, given that both were i. (This is the top part of Table 19.4, "Structurally Adjacent Nodes.") The sec-isure was the frequency of two nodes adja-input recalled together in correct input gain given that both were recalled. (This is torn part of Table 19.4, labeled "Input t Nodes.") :h higher percentages of structurally adja-des than of input adjacent nodes were 1. The recall frequencies for the structural-em nodes are all 58% or higher. The fre-s for the input adjacent nodes are all rably less than 50%. Mandler (1978a) -•oorts similar findings for her native speakers, -kus, when the unstable, interleaved input order is ,-ored in memory for a relatively long period of ' me (24 hours), it tends to be converted to the deal, story-schematic order and tends to be -called according to that schema. DISCUSSION "he differences found in quantity of and temporal :quence of recall between the standard and inter-aved stories in this study demonstrate the effects f a simple story schema on the comprehension of mple English stories by ESL readers. Quantity of :call was enhanced when the story was structured ith a rhetorical organization that conformed to ie reader's schema for simple stories—i.e., one •ell-structured episode followed by another— hen the story was structured with a rhetorical rganization that did not conform to the reader's :hema. Further, the recalls produced by the read¬'s of those stories which violated the story :hema showed a strong schema effect in the ten-:ncy of temporal sequencing to reflect the story •hematic order rather than the input order. In Dth of these findings regarding quantity of and imporal sequencing of recall, the ESL readers shaved similarly to native English speakers, espe-ally to fourth- and sixth-grade children. The finding that the ESL subjects in this study re most comparable to Mandler's fourth- and sixth-grade native English-speaking children and not to younger children or adult native speakers may reflect their developing linguistic systems as well as the effects of the story schema. The greater difficulty of retrieving story information from interleaved stories may reflect less flexibility on the part of ESL learners' retrieval processes when compared to Mandler's native-speaking adults. In general, the ESL learners had relatively more diffi¬culty in retrieving information from the inter¬leaved stories. This may also be due to confusions occurring at the time of encoding material pre¬sented in an unfamiliar format. Native-speaking adults don't have to devote as much effort to lin¬guistic encoding and can therefore devote more effort to encoding the incoming interleaved stories in their ideal schematic form as two separate episodes, tagged with an interleaving algorithm. Non-native readers who must devote more effort to linguistic encoding are less able to devote atten¬tion to sorting interleaved input into an ideal schematic form. The fact that Ending nodes were better recalled by the ESL readers in this study than by the native speakers in Mandler's research (1978a) is an interesting anomaly which must await future explanation. I have no satisfactory explanation for the fact that these non-native readers apparently found material at both the beginning and the end of episodes particularly salient and memorable. The reasons for the particular salience of Endings over Outcomes for the ESL readers, compared to the reverse for native speakers, will require further investigation. In addition to the further investigation of story schemata of the kind represented in this study, research into the effects on ESL reading compre¬hension of other kinds of rhetorical organization is needed. In particular, as mentioned at the begin¬ning of this article, more research is needed on the effects on ESL reading comprehension of different types of expository rhetorical organization, in the sense of Meyer (1975, 1977,1979; Meyer, Brandt, & Bluth, 1980; Meyer & Rice, 1982; Meyer St Freedle, 1984). The research of Connor and McCagg (1983a, 1983b), of Hinds (1979, 1983a, 1983b), of Burtoff (1983), and of Carrell (1983c), previously discussed in this article, all demonstrate the relevance to ESL reading and writing of the investigation of a wide variety of formal, rhetori¬cal schemata. Pursuing the investigation of such formal schemata cross-linguistically and cross-cul¬turally—in the spirit of contrastive rhetoric (Kaplan, 1966, 1972; Kaplan 6c Ostler, 1982; Houghton, 1980; Houghton & Hoey, 1983)—will enable us to determine which rhetorical patterns appear to be cross-cultural, or even universal, and might therefore require minimal emphasis in ESL reading and writing pedagogy (e.g., the simple story schema of this study), and which rhetorical patterns tend to be culture-specific and therefore require greater teaching emphasis for different groups of learners (cf. Carrell, 1983c). In addition to investigating the interaction of texts of different types of rhetorical structures and the formal schemata of different groups of ESL readers, research in schema theory has shown the effects on ESL reading comprehension of content Chapter 19 / Evidence of a Formal Schema in Second Language Comprehension 201 ^r ly supported by an internal grant from SIU-C'. of Research Development and Administration. The research reported in this study will be limiter written medium; therefore, the focus will be ot der. However, much of this research is genera]^ ■ the oral medium, and would be applicable to lis. as well as to readers. Mandler's research has itly employed the oral medium (Mandler, 197sa. er& Johnson, 1977). For the effects on ESL reading of content ita (background knowledge of the content area «, Carrell, 19S3a), see, for example, the research fensen, Joag-dev, and Anderson (1979), Johnson 1982), Hudson (1982), Alderson and Urquhart , Carrell (1981, 1983b), Carrell and Wallace , and Carrell and Eisterhold (1983). These stud-reviewed briefly in the discussion section at the the article, where formal and content schemata led in a unified schema-theoretical view of ESL Mthough Mandler (1978a) scored her recall data )Oth a loose and a strict criterion, disallowing istortions and additions, only the loose criterion appropriate for scoring the recall protocols of itermediate-level ESL readers. Due to their inex-ntrol of the vocabulary and grammar of English, d have been virtually impossible to tell whether ppeared as slight distortions or as reasonable ns were really such and hence due to schema or whether they were due to inexpert control of grammar and vocabulary. For example, in one f "The Studious Roommates," a subject wrote, 7hen the examination started, suddenly he felt ie didn't finish and did properly so he didn't ; anything." Is the word properly a distortion of jnal text or a problem of vocabulary control in :he student intended the word poorly? It was to give these ESL readers the benefit of any ind not to penalize them for vocabulary and itical shortcomings. Therefore, only the loose l was used in scoring their recall protocols, and ;rary decision had to be made as to whether the rion recalled was distorted or added to or not. If rmation met the loose criteria, it was counted as seen recalled, period. Therefore, distortions and .s are not separable from recall of the essential tortions and additions are not available for sep-lantitative analysis by standard versus inter-:onditions as they were in Mandler's study AppENPDC19.1 Three standard and Three Interleaved Stories ,-gE UNLUCKY TWINS standard version) Once there were twins, Tom and Jennifer, who had so much trouble that their parents called them -je unlucky twins. One day, Jennifer's parents ■ave her a dollar bill to buy the turtle she wanted, -Hi on the way to the pet store she lost it. Jennifer •rzs worried that her parents would be angry with er, so she decided to search every bit of the side- /alk where she had walked. She looked in all the racks and in the grass along the way. She finally .-Dund the dollar bill in the grass. But when ;nmfer got to the store, the pet store man told her : iat someone else had just bought the last turtle, -rid he didn't have any more. The same day, Tom ■ :ll off his bicycle and broke his leg. He wanted to jn and play with the other children, so he got the hildren to pull him around in his wagon. While - icy were playing, Tom fell out of the wagon and roke his arm. Tom's parents said he was even nluckier than Jennifer and made him stay in bed ntil he got well. HE UNLUCKY TWINS nterleaved version) )nce there were twins, Tom. and Jennifer, who ad so much trouble that their parents called them ie unlucky twins. One day, Jennifer's parents ave her a dollar bill to buy the turtle she wanted, ut on the way to the pet store she lost it. The ime day, Tom fell off his bicycle and broke his :g. Jennifer was worried that her parents would e angry with her, so she decided to search every it of the sidewalk where she had walked. Tom ■'anted to run and play with the other children. Jennifer looked in all the cracks and in the grass along the way. Tom got the children to pull him around in his wagon. Jennifer finally found the dollar bill in the grass. While the children were playing, Tom fell out of the wagon and broke his arm. But when Jennifer got to the store, the pet man rold her that someone else had just bought the last turtle, and he didn't have any more. Tom's parents said he was even unluckier than Jennifer and made him stay in bed until he got well. THE PLAYFUL PETS (standard version) Once there were two playful pets, a dog and a cat, who lived together with their master in a big city apartment. One day the master gave the cat a new toy mouse to play with, but as soon as her master left for work, the cat chewed big holes in the toy. The cat didn't want her master to take away her toy, so she looked for a place to hide it. She looked under all the furniture and in all the closets for a hiding place. She finally found a good hiding place in the kitchen. When her master came home and found the chewed up toy, he scolded the cat and threw her toy away. The same day, the dog ran away from the apartment. He wanted to run and play outside with other dogs. So he ran around the neighborhood barking and barking with other dogs. While the dogs were running loose, the dog-catcher came and took them all away to the dog-shelter. The master was so angry when he had to go get his dog and pay a fine thai-he kept the dog tied up after that. THE PLAYFUL PETS (interleaved version) Once there were two playful pets, a dog and a cat, who lived together with their master in a big city apartment. One day, the master gave the cat a new toy mouse to play with, but as soon as her master left for work, the cat chewed big holes in the toy. The same day, the dog ran away from the apart¬ment. The cat didn't want her master to take away her toy, so she looked for a place to hide it. The dog wanted to run and play outside with the other dogs. The cat looked under all the furniture and in all the closets for a hiding place. The dog ran around the neighborhood barking and barking with the other dogs. The cat finally found a good hiding place in the kitchen. While the dogs were running loose, the dog-catcher came and took them all away to the dog-shelter. When the master came home and found the chewed up toy, he scolded the cat and threw her toy away. The master was so angry when he had to go get his dog and pay a fine that he kept the dog tied up after that. Chapter 19 / Evidence of a Formal Schema in Second Language Comprehension 203 NDIX19.2 >le Recall Protocols; Interleaved Versj, Standard Order Recalled
UNLUCKY TWINS
1 produced by an Arabic speaker)
was two twins John and his sister Mar .ster asked her father to give her a dollar rurtul from the pets store and he gave he bill. She went to the turtle store, but < that she lost the bill. So she was afraid fr( ther. Then she decided to look for it wr as back she found it on the grass. Later < I it up and went to the pets store to buy turtul. But for sorry he told her that t have any he just sold the last one. This v happened with her. ow we'll now what happened with 1 r. While he was driving his bicycle he j and he broke his leg. Then he was play lis friend and broke his arms. So for sun bad day for the two twins. >LAYFUL PETS
produced by a Malaysian speaker)
ay a master bought a mouse toy and gav< cat. The played and chewed a big hole a a it tried to hide this toy from its mastc he found in the kitchen. When the mast lack he found the toy was chewed with b ie scolded the cat and threw away the toy. ten the master left home, the dog also war nn away and played with the other dogs, g at the neighbor house, calling it's frien same time the dog catcher came and « en to the shelter. The master was very ang d to pay for the dog. So he tied up the dog TUDIOUS ROOMMATES produced by a Spanish speaker) ime there were 2 students living in the sar. One of them was very bad in his class ai 1 pass the course this semester. So, he stu hard for about 3 days, in the morning, jfternoon, and in the night. But when he went ie school he was very tired. Then he can't pass The other student received a letter from his girlfriend, sayind him tha her will come the next week. So, he studied all yours next class to can leave all time with her. But when he want to the train station she don't come in it. DISCUSSION QUESTIONS ": there were no such thing as a "story schema," what effects on recall could be expected to be caused -y scrambling the sentences of a story? Or coming at the question from the other direction, assuming - iat there is a formal story schema, what kinds of effects would scrambling the sentences of a narrative e expected to produce? "Thy should narrative schemata be "developmentally prior" and "cross-culturally universal"? (Also see ;hapters 20, 28, and 37.) liscuss the formulaic openings of stories that begin "Once upon a time ..." (see Chapter 33) and include "... and they all lived happily ever after." What other cues are typically used in the surface )rms of text or discourse to inform us what sort of schema (factual, fictional, narrative, expository, ortatory, etc.) ought to be employed? That evidence does Carrell offer to show that non-native speakers as well as native speakers seem to ;ly on the same underlying type of formal schema? Also, what explanations can be given in this con-xtion for the lack of contrast in some cases between the sequential and interleaved episode structures iat Carrell used? Chapter 19/ Evidence of a Formal Schema In Second Languase Comprehension 205 20 se ie Chavez-Oiler, Grover K.H. Yi Hurtado de Vivas AUCTION experimental studies. The idea that •pears in a meaningful sequence, i.e., tual form, is certainly not new (see r, attempts to measure the impact of ig the term of Carrell, Chapter 19) Id pass before any of the authors of r & Qbrecht, 1969), and it would •fa century (Oiler, 1975a; Chihara, Yu, 1986; Chavez-Oiler, Chihara, urtado de Vivas, in press; and Oiler •kthroughs on the experimental side, ical evidence for the universality of a dramatic learning benefit attribut-hema is presumably only one aspect the schema itself remains nebulous, in structures built up from logical y are built, from Chapter 21 by >bowing why children and adults in from episodically organized verbal Oiler, 1983b; and especially Taira, discourse dropped out of the blue, n is the kind we see in a chronolog-. The results, summed up here, tell •■ sort recommended in Chapter 17 ^search reported is known as "cloze
ng across time. The events of the experi-)ry, it turns out, however, are highly con-1. They don't just occur in any old order, experience, after all, is not altogether fie-It is constrained by physics—by what can


r"ou-
mot happen in the material world. It is also , [jjgd by physiology—by what we can per-" what happens in the world. This doesn't keep people from making mistakes or hav-;ions, but it does enable us, in many cases, — ctly distinguish actual experience from fan-id fictions of many sorts. a result, materials to be studied in class-will be easier to understand at the start, *-cier to recall, and easier to learn from, if they * Worn to the kinds of experience that people are n ill to understand. If the material is arranged in a ^■arJngful sequence, that is, if it is episodically rcanized in the manner of ordinary experience, h"-jdcnts will naturally have a better chance of ■ demanding it. In fact, this is partly because the ,■ ider.ts know more about episodically organized -item's in advance. They have an a priori famil--itv sbout them. We know what kinds of things i-j !:!Iy to happen and in what ways they are
■ rmally constrained in experience. If young
'. argic's recently acquired balloon at the fair
nvs away and strikes a tree and bursts, we
dcrsrand her crying (see Rumelhart's discussion,
.-75:. She is disappointed about the loss of the
lioon. In the past she was happy to get it, and
w ir. the present she is sad about losing it.
The fact is that we know a great deal about
■■v episodically organized materials before we
:n gsc around to studying them. All of this is
- :rrv obvious to anyone who thinks about it, but
■ .i also often disputed by theoreticians and/or
lored in practice. Instead of-studying meaning-
■ , episodically organized materials at school,
; Idren learning to read are often exposed to rela-
" sly unorganized and unmotivated materials such
the Dick and Jane readers, or exercises just like ' m that make no more sense than the kinds of
ngs Dick and Jane did and said. ("Jump! Jump!
n, Spot, run! See Spot. See Spot run. Look! . ok! Jump, Dick! Jump! Run! Run! Run!" Etc. It 1 /er, apparently, occurred to the writers of the uick and jane type readers to ask where Spot was going or why he was running or what Jane and Dick were jumping about.)
In language classes, isolated phonemes, words, ="d sentences are not commonly studied anymore, but conversations (textual and video-type social '■lunettes) typically parachuted from the blue sky r.re probably the most common denominator of

most current foreign and second language pro¬grams. It may or may not occur to students to ask why Ms. Jones is asking permission to leave work early, or why Mr. Smith refuses to let her go, but in many cases, the vignette, having been dropped from the blue sky, actually has no past and no future. It is exceedingly unlike ordinary experience in this way because ordinary experience absolutely cannot exist without a past and a future. The brain-damaged patient on a recent PBS program who thought he was just waking up or becoming conscious for the first time every two or three minutes was a decided¬ly abnormal (tragic) case and could not be said to have any normal experience at all. He could not, for instance, recognize his wife from one two-minute segment of his life to the next. He had the illusion that he was meeting her anew during each new seg¬ment of time. In this respect, his experience closely resembled that of the fictional characters in many basal readers and language programs—characters who just suddenly come into existence, without any known past in one lesson and disappear forever before the next lesson, where students find a whole new cast that has recently come into existence, and so on and on it goes through an unending series of unconnected vignettes.
In Chapter 28, Taira shows that such a dis¬jointed series of vignettes that appear, more or less randomly, in many language programs are less conducive to successful comprehension and lan¬guage acquisition than exactly similar materials arranged in the form of a sequentially developing story line. But, to mention that, we are getting ahead of our own story. Here we are concerned with a series of experiments that lay the ground¬work for Taira's important breakthrough. In this chapter, we want to show only that sequentially arranged text is easier to understand and to learn from than the same text that has been disrupted by a scrambling procedure. The technical details of our story are contained in Oiler and Jonz (in press), so here we will try to stick just to the main outlines of the research projects so that teachers and other users of the second edition of Methods That Work will be able to appreciate the powerful empirical evidence that has been amassed support¬ing the central thesis of this whole book (especially as spelled out in detail in Chapter 37)—namely, that heavily scaffolded, multi-modality, episodical¬ly organized experience in the classroom will give

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 207


ilir.g). Since it is true that grammar-based 'xpectancies ranging beyond five to ten words on ■ithsr side of a cloze blank can help subjects to ;ucss missing words, then cloze procedure is sensi-ive to formal constraints. In fact, this is now well stab'.ished both on theoretical and empirical .rounds (see Bachrnan, 1990; also the concluding hapter of Oiler &c Jonz, in press). In addition, it is " ow known that the point of diminishing returns or the effect of discourse constraints ranges across cmcrxe boundaries and much larger segments of tex- than was previously supposed by some researchers such as MacGtnitie (1961), Carroll (1972), Alderson (1979a, 1979b, 1980), Porter (1978, 1983), Shanahan, Kamil, and Tobin (1982), and Shanahan and Kamil (1983).
Lghaus, 1897). Taylor (1953, 1954, 1 -, also improved the procedure by provid a much richer research basis than it *. id prior to his work.
fowadays, researchers distinguish ber . tests (or exercises) that involve deleting i -ord (where n is usually a number betwi ') and ones where the researcher care 5 the words or other elements to be de -achman, 1985, 1990). The first type of ." Ises are called "fixed-ratio" tests. The ] illed "rational-deletion'" exercises anc he broad class of "variable-ratio'4 appn this present chapter, all of the cloze >yed were of a kind of hybrid "fixed-r; rational-deletion" type. In general, in cres;. ie cloze exercises in the research reported (see Appendices 20.1 and 20.2), every nt'a was deleted, but in a few instances, if de.ct. e word at the nth position would have result-an impossibly difficult item (say, it wr.s a >r name used only once in the text), that item : be skipped in favor of the word just after it t before it.
here are many well-known applications of research to educational tasks, but especially lects of literacy. For instance, cloze exercists be used to determine, roughly, whether a text is suitable in terms of its difficulty led given group of students. Table 20.1, from and Jonz (in press), gives some rough guids-for interpreting cloze scores over a given text re to the reading abilities of a typical group ejects.
E 20.1 Rough Estimates of Readability
Based on Cloze Scores Between 1 /5 and 1/7 Deletion Ratios
ily Exact-Word Contextual-
id Levels Scores Appropriatness
idability Scores
sndent 53% and up 85% and up
tionai 44% to 52% 66% to 84%
:rate)
tional 43% and below 65% and belc.
Lilt)

^e term independent readability—including
-word scores (roughly) of 53% or above
cr _;ontextual-appropriateness scores of 85% or
^ *". r_-has been used to refer to that level at
* " ti a written discourse can be understood
, • mt special study helps (such as a dictionary
n. tor); instructional readability—applied
*-- hly) ro exact scores between 44% and 52%
, . ^textual scores between 66% and 84%—
-1 een used to indicate the level at which learn-
--. -nerally understand enough of a discourse to
n it from it in a classroom or other study set-
and frustrational readability—applied
hly) to exact scores of 43% or lower or con-
._ ■ il scores at or below 66%—means the text
s too difficult to be of much benefit. (Again, it
■nust be stressed that these guidelines are rough.
A good deal of elaboration is given, however, in
Oiler &C Jonz, in press.)
THE IMPACT OF FORMAL CONSTRAINTS
.c this chapter, we are interested in more than just -he readability of any given text or the literacy of my given group of subjects. Here we are con¬cerned with the impact of formal constraints including the special case of formal schemata) on :hs relative redundancy of any given text as mea-•urcd by cloze items inserted in that text. More pcci fically, we want to test the hypothesis that a equentially arranged text will be easier to under-rand and to learn from than one that is disrupted jy iome degree or other of randomization (scram¬bling). Since it is true that grammar-based 'xpectancies ranging beyond five to ten words on •ithsr side of a cloze blank can help subjects to ;ucss missing words, then cloze procedure is sensi-ive to formal constraints. In fact, this is now well stab'.ished both on theoretical and empirical .rounds (see Bachrnan, 1990; also the concluding hapter of Oiler &c Jonz, in press). In addition, it is " ow known that the point of diminishing returns or the effect of discourse constraints ranges across cmcrxe boundaries and much larger segments of tex- than was previously supposed by some researchers such as MacGtnitie (1961), Carroll (1972), Alderson (1979a, 1979b, 1980), Porter (1978, 1983), Shanahan, Kamil, and Tobin (1982), and Shanahan and Kamil (1983).

In addition, the point of diminishing returns, if it exists, is not a limitation on formal constraints per se, but on the capacities of human producers and interpreters to use them. While a constraint on human ability is, in a way, a constraint on the rep¬resentations we are capable of producing and/or understanding, this does not reduce the logically independent theoretical status of the constraints themselves. A text in its material existence is as different from an act of using it as a baseball is from an act of being thrown or caught. Though it remains true that the size, weight, shape, etc. of the ball depend greatly on the size, shape, and strength of a human hand, the shape of the human hand does not make it logically impossible for a baseball to be constructed so as to be either too large or too small for any pitcher to ever throw it. In the same way, the characteristics of texts are limited by the capabilities of our minds, but nonetheless have a separate, real existence that is independent of ourselves. In any case, the results discussed in the following experiments show con¬clusively, we believe, that contexts beyond the five-to-ten word level are generally helpful in enabling subjects to infer intended meanings and in enabling them to supply missing material in cloze items. This is true for rexts of the narrative kind as well as at least some texts of a descriptive or expository sort, as we will see. More important¬ly, we know now that a great deal more is learned from working through cloze exercises over texrs where the underlying formal constraints are undis¬turbed than in the same exercises over the same texts where the formal arrangement has been deliberately disrupted (by some degree or other of scrambling).
In what follows, we sum up the results of seven studies in which one or more of us partici¬pated. The first two deal with the degree to which formal constraints impact scores on cloze exercises over descriptive, narrative, and expository texts by native and non-native subjects while the last four experiments concern the relative amount of learn¬ing (or information gain) that occurs when natives or non-natives work through narrative texts arranged according to a typical formal schema of rhe narrative type versus the same texts where the sequence has been disrupted by a scrambling pro¬cedure. As we progress, we will discuss only as much background and analysis as we think is



*

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 209



>f the 5 texts). Each of the five 100-item H vas structured so that a text of 100+ word°' al prose appeared first and the remain iJ ■ divided into scrambled segments of 5Q ; id 5 words, appeared next in a random Q-A . : segments of varying lengths, which fo-g. ng degrees of less than sensible prose, Wh ~ ed on the page immediately followin» ^ words of text as if they all formed n SL*I„ ;raph. On each of the 5 cloze tests, therefo* of the texts were represented but each one -rent order of approximation owing to i r.L- ~j. scramble procedure applied. Each clo?c tc • represented all 5 orders of approximation --text appeared in each order of approxi r aric-xactly one of the 5 cloze tests. The desit;- Wi% :matically counterbalanced to distribute vari, s in passage difficulties equally over :'-c * xs of approximation to English prose. Aisc, tr. s of subjects were assured to vary only ran^on y the way the 5 cloze tests were handed out • i participating group of subjects. The tests we ked from number 1 to number 5 and !-, man. so that nothing but the luck of the drav., r.r_ :re any individual was seated, could detc-r.' it subgroup of subjects might work on Teii 1, ■, or 5.
The intent of the experiment was to shov di, 100+ segments, arranged in the orde1 ev¬icted by their original authors, would be CJM( process (as shown by higher cloze scores -r.a same texts cut into 50-word segments pLtb.-i in a random order. Similarly, the 50-wo-d ^ ■ nts presented in random arrangements v, o.-l easier to process (again, shown by higher c.o/-ires) than the randomly arranged 25-worJ oci ■ :nts, and so forth. The 5-word segments won' the most difficult to process.
iterials and Tests of Study 1
>ur of the 5 passages used for Study 1 wei cerpted from selections in a reader edited b arrell, Laird, and Freeman- (1970) intended fc •liege readers of English. The fifth was a passag ken from a novel by Victoria Holt (1973). The xts ranged from (I) a description of a carry-out stauxant of the 1950s, to three expositions cov¬ing such diverse topics as (H) how to get an edu-ition, (III) how to construct a paragraph, (IV)

- a sentence is, and (V) a fictional narrative
*r,' t the death of Sir Edward Travers. Each pas-
,'"\. was at least 100 words in length plus a lead-
sl1"" ntence and a lead-out sequence of at least 5
B .*»ls beyond the 100th word.
** \y deleting the 3rd word of the 2nd sentence
■ich passage and every 5th word thereafter
V exactly 20 cloze items were attained, 5 cloze
of 20 items each were constructed over each
. - e 5 passages. These cloze tests yielded 5 tests
.. - e l00th+ approximation to the actual texts
segments were each taken from. Each 100-
* i-1 segment contained exactly 20 cloze items,
-„ of which, except for the first and last, was
" - -ated from other items before and after it by
„ tiy 4 words. To achieve the 50th order of
- -aximation, each of these initial texts of 100+
. is was duplicated and cropped at the ends so
- it was exactly 100 words long. Then it was
„ ed m the middle to produce two 50-word seg-
s. To get the 25th, 10th, and 5th orders of
•- nximation, the 50-word segments were dupli-
: - i 3 more times and cut 3 more times into 25-
i, 10-word, and 5-word segments. This
led a pool of 100 cloze items, each item
- aring in all 5 orders of approximation to nor-
irose (100th+, 50th, 25th, 10th, and 5th).
For example, following the lead-in sentence in
passage I, in the 100th+ condition, there were 20
doze items of the form:
For those who hang out there, the Carry-out offers a wide array of sounds, sights, smells, tastes, and tactile experience which titillate and sometimes assault the five senses. The air {U warmed by smells from {21 coffee urns and grill •J-l thickened with fat from {41 deep-fry basket. The jukebox {51 up a wide variety {£1 frenetic :'pd lazy rhythms. (7) ■. ■
The 5th order approximations consisted of the sanit. items, but in 5 word segments, such as
fat from {4_1 deep-fry basket, wide variety {£1 frenetic and The air (U warmed by ...
and so forth. The 10th order approximations con-sisred of the same items but arranged in random¬ized 10-word segments as in
T!ie jukebox {11 up a wide variety i£X frenetic .mdTheair {U warmed by smells from {21 cof-Eie urns ...

Therefore, 100 distinct items times the 5 condi-tions of arrangement (i.e., the 5 orders of approxi¬mation to normal prose) gave a grand total of 500 items from which to construct 5 distinct 100-item cloze tests. Each of these tests would contain one of the 5 passages in the 100th+ condition, another in the 50th word condition, another in the 25th, another in the 10th, and the remaining one in the 5th. All passages would appear in all 5 tests, and each passage would appear exactly once in each of the 5 orders of approximation to English prose. It is crucial to note that every single item appeared exactly once in each of the five orders of approxi¬mation. The items in each approximation were identical except for the scrambling procedure applied in each case.
Subjects of Study 1
Ninety-three adult native speakers of English each completed one of the 5 cloze tests. Subjects were drawn from introductory linguistics and English classes at the University of New Mexico and from a group of participants at the Sixth Annual Michigan Conference on Applied Linguistics in February 1975. Subjects ranged in age from the late teens to over 50. There was a nearly equal bal¬ance of males and females.
Results of Study 1
Tests were scored by two methods. First, the exact-word scoring method was used, and, second, words other than the exact word which fit the total context of the original passage were counted correct. The latter method is referred to here as the "contextually acceptable" (A) criterion and is con¬trasted with the "exact-word" (E) criterion. If for¬mal discourse-level constraints make a difference, and assuming cloze is sensitive to such constraints, scores should be highest on the 100th+ order of approximation to normal prose and should be ordered in descending rank from 50th, to 25th, to 10th, to 5th. Figure 20.1 shows that this predic¬tion was sustained for all conditions and for both the exact-word and contextual scoring methods averaged over the five texts. The means on which Figure 20.1 is based are the column means of Table 20.2. An appropriate analysis-of-variance (see Oiler &c Jonz, in press) showed that the

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 211



10th 5th Row Means
9.7 5.8 12.1
4.3 3.0 5.5
8.2 4.2 10.3
3.6 1.3 3.7
6.8 6.0 10.5
4.9 4.3 6.2
5.8 5.3 10.3
4.2 3.9 7.6
10.5 6.3 12.6
6.3 4.6 7.8
8.2 5.5 11.6
4.7 3.4 6.2
RE 20.1 Contrasts Over All Passages in Five Orders of Approximation to English Prose
5 10 25 50 T
Orders of Approximation
:e that formal constraints on textual materia'. ly do make a difference to its comprehensibili-The only surprise here, based on a reasonable >ry (see Chapter 37), is that such an outcome uld be a surprise to anyone (and we must sup-e that it is surprising to people who have jed that cloze procedure cannot be sensitive to :ourse constraints beyond the five-to-ten word :1). The cloze procedure is sensitive to discourse straints, and the possible point of diminishing

T

■ irtis for that sensitivity must be well beyond 50 " " 'ds cf context, not at 5 or 6 words, as previous-TW-d by Miller and Selfridge (1950). Neither .. cio/-e items generally limited in sensitivity to '. K-ai within the immediate sentence context, as r,,ucd by MacGinitie (1961), Carroll (1972), "Odsrson (1979a, 1979b, 1980), Porter (1978, ■ -831.. and Shanahan, Kamil, and Tobin (1982; \. o Shanahan & Kamil, 1983).
STUDY 2: NORMAL NATIVES AND
HQH-NATIVES
Background of Study 2
'1 na^auskas (1972) had shown that cloze items .-bedded in whole sentences presented in a ran-
7i order were more difficult than the same items ; -heeded in the same sentences presented in their
p-.r.al order as part of a narrative. The result, ■ vvc-cr, was achieved with educable mentally -.■ irded subjects. Therefore, the experiment by _ jnara et al. (1977), which is summarized here as ;- d> 2, set Out to see if the Ramanauskas result . ild be achieved with normal natives and non-
:ives. Another purpose of the Chihara et al.
dy was to rule out the possibility that the proce-. re used in Study 1 above (Oiler, 1975a) might
re biased results against a negative outcome by . ting and scrambling segments of prose irrespec-
j of sentence boundaries. Although Study 1 demonstrated that greater
ounts of meaningful surrounding context (other ' igs being equal) Increased the ease with which
ive subjects supplied missing words in a variety
cloze exercises (narrative, expository, and
criptive; see Appendix 20.1), it had not been
>wn explicitly for normal natives and non-natives that cloze procedure would be sensitive to constraints ranging across sentence boundaries. Nor had the degree to which such constraints were available to language users at different levels of ptoficiency been investigated. Our first hypothesis for Study 2, then, was that cloze items would be sensitive to constraints ranging beyond sentence boundaries, and our second hypothesis was that increasingly proficient language users would be better able to capitalize on this fact. In other words, the abstract formal schema and other dis-

course-level constraints underlying narrative texts would be more accessible to more proficient English speakers than to less proficient ones. Our first hypothesis had already been supported in part by Carroll et al. (1959), Ramanauskas (1972), and by Study 1, but our second hypothesis had not been tested prior to our own Study 2.
Materials and Tests of Study 2
Two passages of prose were selected from texts written for non-native speakers of English. Text A was a narrative from Praninskas (1959: 217) about Joe going off to college and all the things his mother made him do before he could leave. Text B, from Wright, Barrett, and Van Syoc (1968: 327-328), was a similar narrative about a Greek boy named Nicholas going off to Athens to work for his uncle in a mechanic's shop. Two cloze tests were constructed over each text. Roughly every 7th word was deleted from A and every 6th word from B to get two sequential cloze tests (sequential A and sequential B). No lead-in or lead-out (unmutilated text at the beginning or end) was allowed for either test. Then the tests were divided roughly into thirds. Divisions were always at sen¬tence boundaries. In this way each sentence that appeared in the sequential version would also appear in the very same form, but not in the same order, in the scrambled version. The last sentence of the last third of A in the sequential form became the first sentence of the scrambled version of A. The last sentence of the second third of A in the sequential form became the second sentence of A in the scrambled version, and so on, until all sentences were exhausted. Then the same proce¬dure was applied to passage B. There were 56 cloze items in A in each of its forms (sequential and scrambled) and 48 items in B in each of its forms. The four tests appear in Appendix 20.2.
Subjects of Study 2
Adult students of English as a foreign language at the Osaka YMCA, 201 in all, and 41 native speak¬ers of English at the University of New Mexico served as subjects. The non-natives, all native speakers of Japanese, were enrolled in basic, inter¬mediate, or advanced EFL classes at the Osaka YMCA in Japan. Four groups were distinguished

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 213


71 17.93 6.88 16.92 7.85 1.01 9.82
66 31.79 7.69 26.36 8.56 5.43 11.94
64 41.13 8.82 33.19 8.89 7.94 13.29
41 71.57 8.75 60.29 9.69 11.27 8.87
242 36.93 19.61 31.15 16.90 5.79 11.84
TAB ' E 20-3 Mean PcrcentaScs °f Exact-Word Scores Between Groups by Condition
Sequential (Q) Scrambled (R) Difference
SD
Mean
SD
Mean
SD
Mean
plished by random assignment of tests ts in each test setting. First the tests v.C d alternating between the four combir-.' Later, when they were passed out in eCi >om context, they were handed out in *k irrangement. This resulted in random sc'c-f the four blocks of subjects (i-iv, define i with approximately equal numbers in e^j, . Moreover, the four blocks were aboiy y distributed over the four different levels o1 .ency defined earlier. All of this is importan1 ;e our results depend not on the equivalent randomly created blocks of subjects but or fference between the four proficiency level; ning, intermediate, advanced, and native], le design, speaking somewhat technical!) or a moment, therefore, involved two com-• crossed between-subjects variables (groups nd blocks 1-4), and three within-subjects les each nested within the other two: order >r second) was nested within text (A and B), in turn was nested in the main variable of st, condition (sequential or scrambled). In words, approximately half of the subjects in the sequential version and B in the scram-n = 120), while the remainder did B in the ntial version and A in the scrambled (w = Of the half that did A in the sequential vet-approximately half did it first while the ider did it second, and so forth.
ts of Study 2
20.3 gives mean scores and mean differences i sequential scores minus mean scrambled ) across conditions (averaging over texts and
0 for each of the four groups.1 From Table
it is apparent that the scrambled texts were,
:dicted in our first hypothesis, more difficult
: whole than the sequential texts. Further, we
>m Table 20.3 that the relative effect of long-
constraints as shown in the contrast between
ntial and scrambled texts (see the columns
d "Difference") increases as the proficiency
>jects increases. This finding conforms to the
rtion of hypothesis two. In addition, taking a
an Cziko (197S, and also personal communi-
1 from him), we followed up in our recent re-
sis of these data with the appropriate sort of
ical "trend analysis" and were able to show

C--P
jinners jmedi3tes -i/anced ives rials
the tendency for more proficient subjects to ■ ime increasingly sensitive to discourse con-,~. nts—as shown in the increasing difference scores from beginning to intermediate to advanced to native subjects in our data—is a linear trend. The growth pattern is completely consistent with the original prediction. Further, it is clear that both native speakers of English and non-native spcr.kers [Japanese-speaking EFL learners, in this case1' are sensitive to discourse constraints. Exactly as semiotic theory predicts (see Chapter 37 and O'h-: et al., 1991), non-natives become increasing¬ly more sensitive to the impact of discourse con¬straints as they improve in English proficiency. Beginners are hardly able to benefit from discourse constraints at all, intermediates are benefited somewhat more, advanced subjects more than intermediates, and native speakers even more than advanced EFL students.
Appropriate analyses of variance showed that, as intended by the experimental design, there was no difference among the randomly assigned blocks of subjects but that there was a highly significant contrast {p .001) across the various levels of pro-ric.ency. Moreover, separate analyses of variance fot each contrast (see the "Difference" scores in T::Me 20.3) across groups—beginners with inter-mec'iates, intermediates with advanced, and advanced subjects with native speakers—showed that each of them was also significant. Indeed, the contrasts became more and more dramatic with "r.c.-) successive comparison.
Another interesting result, one that was not predicted, but that is investigated in detail by Ckihara et al. in the updated version of the 1977 project in Oiler &c Jonz (Chapter 8, in press) was

that the non-native subjects (speakers of Japanese) apparently had more difficulty understanding the text about Joe going off to college than the one about Nicholas going to Greece, while native speakers of English found the Joe text (A) a little easier than the one about Nicholas (B). Further, this difference appeared to be somewhat indepen-dent of the level of English proficiency of the sub¬jects. It seems that this contrast has to do with culturally engendered expectations—the sort of schemata that Carrel! (Chapter 19) distinguishes as "content schemata." The text about Nicholas traveling to Athens is apparently more consistent with the experience of the Japanese adults study¬ing EFL than the one about Joe's mother sending him here and there on various errands before he gets to go off to college. The latter experience is apparently unfamiliar and unexpected by Japanese subjects.2 Beginning, intermediate, and advanced EFL students (all native speakers of Japanese) con¬sistently found text B about going to Athens easi¬er. This result has also been replicated relatively recently (in a study with an entirely different focus) with a different group of Japanese students at Osaka Jogakuin Junior College by Chihara, Sakurai, and Oiler (1989). However, owing to the way texts and conditions were counterbalanced, the contrasts between the two texts could not unduly influence the contrast over conditions. The latter sort of contrast has to do with the "formal schema" (see Carrell, Chapter 19) underlying the two narratives that is quite independent of their specific content. The contrast in subject perfor¬mance on the sequential versus the scrambled vari¬ant of each narrative stands independent of the difference in performance on the two narratives.



it

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 215



. GURE 20.2 Patterns of Difference Scores Over Conditions and Texts by Groups
ig through cloze items over a text arn ?rmai order than in a disrupted order ement in our experimental design, then -
involve a method to measure inform he learning effect).
evious studies, showing the sensitivi items to discourse constraints, had i yed a design with two or more texts Lmanauskas, 1972; Oiler, 1975a, Stu :a et al, 1977, Study 2; and Cziko, l tly to exclude or control any learning e chers involved in those studies assurnec ng through a single text more than produce a learning effect. Taylor {ljib imanauskas (1972) had also already de.-p.on-l that working through a passage in its no:-■rder would impact performance on ,i [uent cloze test over that same passage "-jro-only that subjects were able to under.si.nrd a").
) avoid the possibility of such a learning which would have contaminated (and sor¬did contaminate!) the very differences c: st in many previous studies (e.g., Porter. , two options have generally been employed, ossibility, preferred by Marshall (1970) and me other researchers {Alderson, 19~9a, ■, 1980; Porter, 1983; and Shanahan er al.. , was to test different groups of subjects over rious conditions that a given text migh: be [to (e.g., a sequential order versus soir.e : of scrambling or other distortion). This I, a between-subjects design, however. has idesirable effect of mixing whatever di ffcr-there may be between the groups with ar\ :nces owed to arrangements of the tex! Vcr that design option has generally not bticn ved.
nother option, used independent!} b\ nauskas (1972) and by Oiler, Bowen, D:e-., lason (1972), is, technically speaking, a in-subjects, randomized-block design" w;ie:e differences are distributed over more than :xt to avoid a learning effect. In the case of 'O-text option (in the design of Oiler, Bcvc., 8c Mason, 1972; and also Study 2 abovi . f the subject pool takes a certain version of :xt first and a different version of a differ-'r.: ext. By this procedure, recommended to us Donald Bowen, it is possible to averse

across texts and across conditions so as to distrib¬ute group differences in a manner that effectively eliminates them. (This was explicitly demonstrated for Studies 1 and 2 above, in each case by the lack of contrast across randomly created blocks of sub¬jects. It was also demonstrated independently for Studies 3-7 below; see Oiler Sc Jonz, in press, for details.) This latter option is preferable to using different intact groups of subjects, because in gen¬eral the differences across individuals (and groups) are large relative to any effects owed to contextual constraints in experimental treatments. The reason is easy to understand: differences in levels of profi-cicrcy across individual subjects and/or groups are acquired over months and years of experience whereas differences produced in an experimental context are usually (but see Taira, Chapter 28, this volume, for an exception) due to experience that takes, only a matter of minutes or an hour or two at the most.
In previous studies (especially, Studies 1 and 2 above; also in Cziko, 1978) one of the main goals was to see whether cloze items were sensitive to different levels of constraints within texts. It was desirable in those studies to avoid any learning effect, because it might have obscured the con-tras:> owed to different degrees of disruption of coherence experimentally imposed on the texts. Here, however, in addition to replicating the con¬tends owed to sequential versus scrambled arrangements of texts, we needed to measure the

learning effect itself. We wanted to know just how strong that effect might be and whether we would find it in advanced non-native users of English as a second language as well as native speakers of English and whether such groups would differ in their degree of sensitivity to coherence (as was observed in Study 2).
We predicted that the learning effect (or infor¬mation gain) owed to the experience of having worked through a passage in its normal order would be greater than any learning attributable to working through the same cloze items in a text whose sentences had been scrambled. The reason for this prediction is straightforward: scrambled text will (usually) contain less information and be less coherent than the unscrambled sequential basis from which the scrambled version was derived. The reason is that some of the formal and content constraints (e.g., the formal schema and the content schemata underlying the text) will be violated by the scrambling procedure.
It is important to note early on, however, that the hypothesis that information gain will be greater from episodically organized material does not derive from cloze procedure per se nor from any empirical research with cloze or any other testing method. In fact, it derives from the semiotic theory spelled out in Chapter 37 of this volume (also in Oiler, 1990; Oiler et al„ 1991; and Oiler &c Damico, 1991). More specifically, the result predicted here derives from what is called (see

Chapter 20 / The Impact of Discourse Constraints on Processing and Leamins 217

n as a foreign language in Thailand (see ts of Study 3 section below). According to s of Study 2, the texts in question i adix 20.2) fell at the lower edge 0f uctional" range (compare Tables 20.1 for advanced EFL students at the YMCj i and were easily within the "independt for native speakers (again compare Ta ind 20.2). The results of the Oiler and in Japan showed that college teachers of an scored about 19 percentage points hit i average than the most advanced studeni saka YMCA and only about 10 points be ative speakers of Study 2. Therefore, t I and college EEL teachers in Thailand shthe tests too difficult or too easy.
rets of Study 3
ipants were teachers of English as a for' iage (total N = 90) who attended tr :ial seminars for English language read sored by the United States Informal zy in collaboration with the Thai Ministr ation and the Teachers Colleges at K {n = 30), Chiang Mai [n = 31), and Sonj 29). About two thirds of the subjects v .es and ranged from their early twentie 3. All were native speakers of Thai and ;me of testing, were practicing teacher; sh as a foreign language at a college or :y school in Thailand.
jn of Study 3
subject completed all four cloze tests in ;cutive 50-minute testing sessions with

Occasion Two
Sequential A Scrambled
Scrambled B Sequential
Sequential B Scrambled
Scrambled A Sequential
tD
13)

r, .jvening break of about 20 minutes. In the first -'. od of testing, subjects were given a pair of tests 1* jisting of either text A or B in the sequential " - iition and the remaining one in the scrambled ",- dition- The pattern of test distribution (a ran-*i, lized block design) is displayed in Table 20.4. P .*ing the second testing period, depending on ■i" :h pair of tests had been completed by the sub-„■ in the previous period, subjects were allowed
- idditional 50 minutes to complete the comple-
itary pair of tests consisting of either B or A in
[■. scrambled condition and the other text in the
i.. lential condition.
The design allowed us to measure directly the ning effect of working through a story versus king through the same sentences that made up story but in a random order. There were a er large number of hypotheses to be tested, but
-!u' one of greatest interest here was that subjects ild make substantial gains owed to working
- - ugh the sequential material and essentially no
at all from working through the same materi-a scrambled arrangement.
Results of Study 3
. ti-le 20.5 shows the mean scores of interest, and re 20.3 graphs the results. To test the hypoth-:■. of greatest interest here, we must compare the ; ;m scores on occasions one and two on the two i'lential cloze exercises with the mean scores ;"ss the two occasions on the two scrambled ->. The reason that these mean scores are the i; ones to compare can be understood by a little ■ if reflection. First, we suppose that the expect-:nean score on either of the sequential texts I there been no prior experience with either of "■e texts) would be exactly the mean across both ; that was actually obtained on the first occa-. This assumption is sound because subjects . in fact, obtain that mean and they did do so out any prior experience with either of the cloze tests. Second, we make the same :mption for the mean on the corresponding of scrambled tests. Again, the assumption is id. The mean attained on occasion one by the .ects under study was attained without any prior study of either of the two texts. Therefore, we assume that the expected mean for these same subjects on the second occasion when they


SD
TABLE 20.5 Mean Percentage Scores on Cloze Tests by Occasion and Condition for Thai EFL Teachers (N = 90)
Mean
Occasion
Condition
11.58 9.78
10.52
11.10
56.03 50.37
55.86 58.13
One
Sequential Scrambled
Sequential Scrambled
Two
encounter the corresponding pair of sequential texts will be the one they obtained on the first occasion. Any difference observed would be owed to learning from the prior experience with the scrambled variants. The same holds for the scram¬bled texts encountered on the second occasion. Any difference between the mean attained on that occasion and the mean for the scrambled texts on occasion one must be owed directly to a learning effect. The reason that these assumptions are logi¬cally valid is because, in fact, no experience inter-vened between occasion one and two that might produce any observed difference.
So what do we see? The results are shown most clearly in Figure 20.3. Compare the mean over the two sequential texts on occasion one,
FIGURE 20.3 The Interaction Between Condition (Sequential vs. Scrambled) and Occasion (One vs. Two) for Thai EFL Teachers (N= 90)
60 +
Occasion Two
50 +
Occasion One



Chapter 20 / The Impact of Discourse Constraints on Processing and Learning

219


TABLE 20,7 Mean Percentage Scores on Cloze
Tests by Occasion and Condition for Broward County Teachers (n = 89)
SD
Occasion Condition Mean
One Two
69.34 61.67
69.73 69.42
Sequential Scrambled
Sequential Scrambled
9.11
8.75 9.45
Results of Study 5
Table 20.7 gives the relevant means. In fact, as predicted, the contrast between performance on the sequential texts on occasion one (with a mean of 69.34) versus the sequential texts on occasion two (69.73)—a difference of .39 percentage points favoring the second occasion—was not significant, but the contrast between the mean for the scram¬bled texts on the first occasion (61.67) versus the second occasion (69.42)—a difference of 7.75 per¬centage points—was highly significant. Again, such a large contrast would rarely be expected to occur by chance (p .001). (Because the results are so similar to those of the two preceding stud¬ies, we will not present a graph.)
STUDY 6: FLORIDA NON-NATIVE SPEAKERS
OF ENGLISH
The same texts and procedures were followed here as in Study 5. This replication was also conducted as part of the series of Florida in-service training seminars conducted in the summer of 1991 for teachers in Broward County Schools.
Subjects of Study 6
In this case, there were 38 participants who were either teachers of English as a second language, content area specialists (math, English, social stud-ies, etc.) working with children identified as limit¬ed in English proficiency, or teachers in bilingual programs in Florida. All indicated some language other than English as their primary language. The majority were of Haitian background (about
20.6 Mean Percentage Scores on Clozc Tests by Occasion and Condition fi Carolina Teachers (n = 63)
Mean
SD
icasion Condition
73.28 62.35
74.44 73.75
7.44 8.Q9
6.91 7.37
One
Sequential Scrambled
Sequential Scrambled
Two
in of Study 4
subject completed all four cloze tests in i cutive 40-minute testing sessions with sning break of about 10 minutes, as she in Table 20.4. Otherwise, except for all 5s time for these native speakers, the de: ;st administration was as in the preced in Thailand.
ts of Study 4
20.6 shows essentially the same patterr -he Thai data, except that the contrasts v :-speakers as subjects are more pronounc fore, the contrasts of interest can be seen i
(Figure 20.4). The mean over the sequent
Occasion Two
Occasion One
IE 20.4 The Interaction Between Condition (Sequential vs. Scrambled) and Occasion {One vs. Two) for Carolin EFLTeachers(/7=63)


on the first occasion was 73.28 percentage ■ its, as compared against the mean of 74.44 for ! sequent'a' texts on the second occasion. On the ' '"..age, the native speakers gained 1.16 percentage 1 „" lts from the first to the second occasion, owing ; -Jje experience of working through one of the --jnbled texts on occasion one. However, as in .r- That study (Study 3), this contrast was not sta-• , cally significant. Again, subjects benefited very -. e, if at all, from the experience of working . )Ugh one of the two texts first in its scrambled . - Q before encountering it in its normal narrative ;:- ience. On the other hand, the contrast between - p mean over the scrambled texts on the first occa-s - , 6135, and the mean attained on the second ,:cision, 73.75, shows a substantial gain of 11.40 : entage points that can be reliably attributed to information gained from having worked —; .ugh the sequential variant of the corresponding ./e exercise on occasion one. Again, a difference ■ . large would be expected to occur less than one * : in 10,000 tries.
STUDY 5: FLORIDA NATIVE SPEAKERS
OF ENGLISH
iur next study was conducted as part of a series E Florida in-service training seminars for teachers . the Broward County Schools in the summer of !991.s
Subjects of Study 5
hi this study, the 89 participants were either teach-
s of English as a second language, content area '.pecialists (math, English, social studies, etc.)
orking with children identified as of limited I nglish proficiency, or teachers in bilingual pro-
"ams in Florida. All 89 were native speakers of I nglish. In this case there were, again, about twice
; many women as men (again, see Oiler &c Jonz, in press, for details).
Design of Study 5
Each subject completed all four cloze tests in two consecutive 40-minute testing sessions with an intervening break of about 10 minutes. The design was similar to that of Studies 3 and 4.



Chapter 20 / The Impact of Discourse Constraints on Processing and Uearnins

221


70.97 9.17
61.95 9.07
71.40 8.45
71.50 8.85
58.27 11.35
51.29 9.95
57.91 12.89
59.55 11.38
- (LE 20.9 Mean Percentage Scores on Cloze Tests by Occasion and Condition for Natives (N= 152) and Non-Natives (N= 128)
SD
>ups 1-4) in Study 2 is whether a higher proficiency will result in a greater sensiti :ourse constraints for subjects from S , as was observed in Study 2. Based on t hypothesis of Study 2, we predicted iter degree of proficiency would in fact rid to a greater sensitivity to discours ints for the native speakers of Studies 4 zmtrasred with the non-natives of Studies
o test this hypothesis, the data from tf
eding studies were combined.
jects of Study 7
subjects for Study 7 were merely the i non-natives and natives of Studies 3-native group included the 90 Thai EFL nd the 38 Broward County teachers wh native speakers of English (N = 12S, ad' ■natives) to be compared against t ilina teachers and the 89 Broward C iers who were all native speakers of I 152, native speakers of English). In all, the two groups combined resulted in .
jn of Study 7
:est scores employed in the analyses described in Studies 3-6.
Its of Study 7
20.9 gives the relevant means of grou; ted, owing to the second hypothesis of natives outperformed non-native speak h and were also significantly more ser
formal schema underlying the narr; Lted in their sequential form as conn he same materials arranged in a serai In both cases—that is, for the native e non-natives—the contrast betwee sequential scores on occasions one ani 1 (i.e., not statistically significant), bi st between the two corresponding p> > on the scrambled texts was highly si >r both groups. Moreover, the contra i-narives was significantly less (a differ ........
5 percentage points) than for the natives
Again, both of these last differences were


Condition
Occasion
up
Mean
One
jes
Sequential Scrambled
Two
Sequential
Scrambled
One
Natives
Sequential Scrambled
Sequential Scrambled
Two
:ficant at p .0001, and the contrast between ±s pair of differences was also significant at p ,0G 1 - It does appear that native speakers on the A'hoie are indeed more sensitive to discourse con-;t:a:nts than even advanced non-natives. The ■nc.i:i difference between the scrambled text on Decision one versus the sequential text on the iarr.j occasion for native speakers was 9.02, as Hjajnst a mean difference of 6.98 for advanced ion-natives. Natives benefit more than non-:.itKes on the first occasion from the sequential ext. This is proved by the fact that the natives had ngher scores on both types of tests and yet, never-helcss, had a greater spread between the sequen-ia! ::nd the scrambled text on the first occasion nan ::he non-natives did.
However, it should be noted that the advanced ion-natives in the data pool for Study 7 benefited jropurtionately more than die native speakers did >n occasion two from the prior experience of ■-oriong through a text in its sequential form. This a:i be seen by noting that the contrast between hi? sequential versus the scrambled performance cr natives on occasion one was already 9.02%, as gamst the corresponding 6.98% for non-natives. -lov/Liver, the non-natives gained proportionately -ore (from 6.98% to 8.26%) from working t.-.rough the sequential exercises than the natives did (from 9.02% to 9.55%). This, we believe, is further evidence in favor of the universality of episodic organization as a factor underlying the interpretation of narratives.

The fact that non-natives gained almost as much (in total percentage points, 8.86%) across the two occasions as the natives did (9.55%) from prior acquaintance with the respective story line contrasts markedly with the noteworthy differ¬ences between their observed levels of overall pro¬ficiency. For instance, averaging across the various cloze scores in Table 20.S, we get a mean of 6S.96 for the 152 natives and 56.76 for the 128 non-natives—an average difference of 12.21%. In view of this difference, the non-natives gained more from the narrative schema than they would be expected to do solely on the basis of their profi¬ciency in English, just as Horiba (see Chapter 21) argues, proficiency in the target language is clearly not all that is at work here. It seems that a formal schema, just as Carrell supposed (Chapter 19), really is at work.
CONCLUSIONS
This series of studies confirms previous findings and extends them. First, it confirms that cloze tests are highly sensitive to discourse constraints in at least some fairly typical narrative, descriptive, and expository texts. Second, it shows that the amount of learning owed to working through a narrative text in a sequential form is substantial whereas the learning owed to working through the same text in a scrambled version is next to nothing. For class¬room applications this finding is far-reaching. It

Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 223

llture groups suggesting its universality (See er 37). Also, the greater sensitivity of more-ient subjects to discourse constraints has p regardless of primary language or cultural round. This suggests that the contrasts ed across the two texts used in Studies 2-7 e due to experiential and cultural factors (ari y of what Carrell has called specific "con-:hemata") while contrasts in the impact of Ve discourse constraints (a "formal schema" i by Carrell and others) are probably owed ioric universals. The formal schema of expe-based narratives, it would seem at any rate expected to play a central role in the work-intelligence in general, as well as in the pro-g and language acquisition related to ves. These, it is argued in Chapter 37, form mdation for the discourse processing that in normal language acquisition and literacy.
■ay of sounds, sights, smells, tastes, and tac-;enses. II The air (1) TS warmed by I smells . I fat from (41 THE deep-fry basket. I The
lazy rhythms. (7) THE pinball machine I is
or ability I to will (10) THE ball into t one \.ND buzzers report I progress or (131 i.RS to drink I Royal Crown Cola and (15)
a long-legged (17) BLOND in shorts I and table delight I at her Chesterfield (201 CIG-
ed obscenities. In the background, a sleek irrell, Laird, 8c Freemen, 1970: 29).
BEEN done; not I often, but (21 THE fact :s (4) DO get a I start in 15) INTERESTED, rsonai (71 EXPERIENCES I have I to offer I faculty, the (101 OTHER students, and I 1 METHOD might lose ! a boy (131 HIS e capacity I and the (16) DRIVE to learn, I 1 BACCALAUREATE is one I of the U21 ;e II students some day will refuse to take do not themselves accept, titles (Gorrell,

Passage III
fhe statement of the main idea, discussed in the preceding chapter, helps the writer organize the subdivi-ionsofhispaper.il As he (1) WRITES, he also I recognizes and (2) UNIFIES material within I the sub-livisions, (3) USING the paragraph I as his (41 UNIT of composition. I Paragraphs can (51 BE written in 1
.cany ways, (61 BUT one basic I type is (71 SO useful for I expository writing (81 THAT we are I calling t {9J-1HE standard paragraph. I It includes (10) THE following: (1) topical I material to (111 INTRO-jTjCE the subject; 1 (2) development to (121 ILLUSTRATE or support I or extend (13) THE subject, naterial I that mav (141 BREAK into divisions I or even (151 SUBDIVISIONS: and (31 sometime I a
.onclusion. (16) USING this pattern, I the most (17) OBVIOUS and most I useful way (1.8) TO construct
j. | paragraph is (19) TO write a I topic sentence (201 AND add specification II to support it (Gorrell,
Laird, &C Freemen, 1970: 26).
Passage IV
On some occasions, a single word—"Fire!" or "Murder!" for example—is a complete message; context and tone of voice supply the unspoken information. II Usually, however, (11 COMMUNICATION requires more, I that we (21 NOT only name I a topic (31 BUT say something I about it— "(4) THE fire is I spreading" or " (5) THE fire was I caused by (6) FAULTY wiring." A I simple sentence, (71 THE, horse is I eating, gets (8) UTTERED not because I we think (91 OF the topic I horse and (101 THEN search for I something to (111 SAY about it I but because (121 "WE want to I report on (131 WHAT we have I observed, want (141 TO say something ! about the (15) HORSE. If we I have a (161 GREAT deal more I to say (17) ABOUT a horse, 1 the result (18) MAY be, not I a simple (19) SENTENCE but an I article or (20) A book. Writing II a simple sentence or a longer composition includes at least a topic and a comment about it, although the line between the two may not be precise (Gorrell, Laird, & Freemen, 1970:1).
Passage V
When Sir Edward Travers died suddenly and mysteriously there was consternation and speculation, not only in our neighborhood but throughout the country. II One newspaper (1) HEADLINE ran: SIR I EDWARD TRAVERS VICTIM (21 OF CURSE? Another I reported: SUDDEN (3) DEATH OF EMI¬NENT I ARCHAEOLOGIST BRINGS (4) ABRUPT END TO 1 EXPEDITION. A 15) PARAGRAPH in our I local paper (61 STATED. "The death I of Sir Edward Travers, (71 WHO recently left I this country (8) TO carry out I excavations among (91 THE tombs of I the pharaohs, (10) HAS caused us I to wonder (11) IF there is I any truth (1.21 IN the ancient I belief that (131 HE who meddles I with the (141 REST¬ING place of I the dead (151 INVITES their enmity." I Sir Ralph Bodrean (161 AT Keverall Court, our I local squire (17) AND Sir Edward's I closest friend, (18) HAD given financial I aid to (19) THE expedi¬tion, and I when, a (201 FEW days after II the announcement of Sir Edward's death, Sir Ralph had a btroke, it was hinted that his misfortune was the result of the same curse (Holt, 1973: 325).
APPENDIX 20.2
Form A Sequential
Joe is a freshman and he (1) IS7 having all the problems that most (21 FRESHMEN have. As a matter of -act, his (31 PROBLEMS started before he even left home. (41 HE had to do a lot of (5) THINGS that he didn't like to do (6) JUST because he was going to go (7) AWAY to college. He had his eyes (8) EXAM¬INED and he had his cavities filled, (9) ALTHOUGH he hates to go to a (101 DENTIST, and he got his

Chapter 20 / The Impact of Discourse Constraints on Processing and Leamins 225

■>- She made him GO to see his cousins in Bellevue AND his Uncle Ned in Plaintown and HTS Great-Aunt Lizzie who lives in THE southern part of the state. ■. You're growing §Q fast that you'd outgrow a suit TN_ no time.
Form B Sequential
\ .holas Rizos was not a tourist; (1.) HE was in Athens to (2) WORK. He had arrived from America
- JJ4E day before on a Greek cargo (4) SHIP. During his last year in (5) HIGH school, his uncle had
- ted (6) HTM to spend a year in (7) GREECE and to help him in (81 HTS garage. Nicholas accepted the xation (9) BECAUSE he wanted to become a (10) MECHANIC: he thought that the (11) WORK
.. 4ld be good experience for (12) HTM. He would also have an (131 OPPORTUNTTY to learn more
- ,ut the (14) COUNTRY where his parents were born. (1.5) EOR several months he had studied
LT-HE Greek language at night school (171 TN his home town. He wanted (18) TO speak it as well as
\_ POSSIBLE and to be able to (201 READ signs, at least, when he (211 ARRTVED: but now he wished
■_11\ COULD have practiced it more with the (23) SATLORS on the ship. That first (24) MORNTNG.
' holas woke up and looked (25) AROUND at the unfamiliar room. Everything (26) WAS strange to
,. From his (27) WINDOW, he could see the Acropolis (28) AGAINST the bright blue sky. Then
* ) HE remembered; he was in Athens! (30) HOW happy he was to be (311 THERE. He got dressed
:kly and (32) TOINED his aunt and uncle in (33) THE kitchen. They seemed pleased when (34) HE
[ "Good morning" in their (35) OWN language, but it was difficult (36) EOR him to continue. They
. w (371 VERY little English, and at first (38) HE was afraid to try his (39) GREEK. Before long, how-
■ r, they were (40) ALL talking and laughing together. After (41) BREAKFAST. Nicholas and his uncle
(421 TO go to work. They walked (431 RATHER fast because his uncle was (44) TN a hurry; in fact,
45) WAS later than usual that morning. (461 NICHOLAS would have liked to walk (471 MORE slow-
I order to enjoy (48) THE unfamiliar sights along the way.
?orm B Scrambled
NICHOLAS would have liked to walk MORE slowly in order to enjoy THE unfamiliar sights along the way.
HOW happy he was to be THERE¬FOR several months he had studied THE Greek language at night school TN his home town. They walked RATHER fast because his uncle was IN a hurry; in fact, he WAS later than usual that morning.
Then HE remembered; he was in Athens! ~. He would also have an OPPORTUNTTY to learn more about the COUNTRY where his parents were born.
7. After BREAKFAST. Nicholas and his uncle left TO go to work.
8. From his WINDOW, he could see the Acropolis AGAINST the bright blue sky.
9. Nicholas accepted the invitation BECAUSE he wanted to become a MECHANIC: he thought the WORK would be good experience for HTM.

10. Before long, however, they were ALL talking and laughing together.
11. Everything WAS strange to him.
12. During his last year in HIGH school, his uncle had invited HTM to spend a year in GREECE and to help him in HTS garage.
13. They knew VERY little English, and at first HE was afraid to try his GREEK.
14. That first MORNING. Nicholas woke up and looked AROUND at the unfamiliar room.
15. He had arrived from America THE day before on a Greek cargo SHIP.
16. They seemed pleased when HE said "Good morning" in their OWN language, but it was difficult FOR him to continue.
Chapter 20 / The Impact of Discourse Constraints on Processing and Learning 227

pTSCUSSION QUESTIONS
i 'or the texts discussed in this chapter (see Appendix 20.1), what high-level constraints or formal chemata can be discerned? Why is there no point of diminishing returns between 50 and 100 words of ■ext, as many theoreticians predicted?
■ "jPork through one of the cloze tests in Appendix 20.2 in the scrambled version and the other in the equential arrangement. Then do the alternate versions over each text. Reflect on the kinds of discourse ■onstraints that are violated in the scrambled variant. What happens between the sequential text and he scrambled variant of that same text? Did you remember the story? What about between the scram-iled variant of the other text and its sequential version?
lonsider the finding that more-proficient English speakers are better able to make use of formal schemata and discourse constraints in general. Why must this be so? Why are beginners relatively unable to use such high level constraints?
1. For Japanese EFL students, the text about Joe was more difficult than the one about Nicholas {see ' Appendix 20.2), while for Americans, Thais, and non-native speakers of English tested in Florida, the reverse appeared to be true. The Joe text was easier. What sorts of factors might account for this out¬come? By contrast, why would the benefit gained from formal discourse factors {especially from episodic organization) be, as it apparently is, consistent in its effects across all the groups examined?
Chapter 20 / The Impact of Discourse Constraints on Processes and Learning 229

:.. ., round knowledge or schemata must interact ''. the text itself. The more links the reader .'. s between new and previously acquired ■ ■ ledge, the greater the depth of processing is, he easier the retrieval of information from ,ry {Anderson &c Pearson, 1984). In other ^ - ;, the more familiar a text is, the more likely ,. , '/ be remembered. As in the case of LI ,4 ■ ch, it has been demonstrated that L2 students etter comprehend and recall when they have , to background knowledge or schema rele-o the content and/or the formal structure of a Those who do not have such access will have ilty in comprehension and recall, regardless ir understanding of the language.2 >me researchers have reported the effects of ata or top-down processing on the compre-in of specific linguistic items. Adams (1982) jcmonstrated that activated "script" knowledge of -he topic of a text aided recognition of unfamiliar ■'ocabulary which appeared in it, both in English .is LI and French as L2. Lee (1987) found that stu¬dents who had never been instructed in the Spanish subjunctive mood were able to compre¬hend the meaning of sentences which contained -his verb form, implying that the readers give turning to the text through interaction between -heir knowledge and the content o£ the passage.
Although L2 reading research has favored chcma-theory based studies during the last decade, nosr of these studies are vague in the sense that the omprehension process is inferred from what is eproduced after reading has been completed. Most if :hf studies investigate the effects of schemata by i[il:/.ing free recall and/or comprehension questions s ,i measure of comprehension; none of them xair.ines the ongoing comprehension process. Lciror's note: Cloze procedure research is, of oursc, an exception.] Furthermore, many studies reat ESL readers as a group whose linguistic and ultural backgrounds are diverse. This heterogene-y in readers' variables seems to increase variabili¬ty in the way information is encoded in memory and later retrieved. Therefore, it is difficult to infer what takes place in these individual readers' heads. Our understanding of the effects of schematic knowledge on L2 reading comprehension will be enhanced if the processes are investigated on line as well as off line, with the readers' cultural and lin-guistic backgrounds held constant.

Language Competence and Reading Comprehension
Some theories have suggested that L2 readers' inadequate command of vocabulary and grammar may interfere with conceptualizing the text content and can inhibit reading comprehension. With the limited processing capacity of the human mind, unfamiliar language and schemata necessitate "controlled processing" in which component processes are evoked by stimuli only with con-scious efforts (Shiffrin &c Schneider, 1977). The LaBerge & Samuels model (LaBerge &c Samuels, 1974; Samuels, 1977) shows that both decoding and comprehension require attention. One can infer that in LI skilled reading, the relatively low-level processes such as feature and word recogni¬tion and the language-specific processes such as parsing and segmentation are most likely to be automatic, while unfamiliar features, such as words and grammar in L2, need to be allocated to the reader's cognitive attention, resulting in insuf¬ficient opportunity for conceptualizing and resul¬tant comprehension to take place.
A number of studies on the processes of L2 reading comprehension conclude that a reader's limited command of language increases reliance on graphic information and inhibits use of effective reading strategies. McLeod and McLaughlin (1986) investigated strategies used by ESL readers and native readers who were taking oral cloze and oral reading tests. They found that, although advanced ESL readers scored better in both tests, their error patterns on the oral reading test were the same as those of beginning ESL readers. The proportion of the meaningful errors committed by advanced ESL readers was significantiy lower than in the case of native speakers. McLeod and McLaughlin speculated that the advanced students were aiming at decoding rather than comprehen¬sion. The reader is cautioned to recognize that oral reading may impede the L2 reader's comprehen¬sion because the extra task of pronouncing words results in insufficient attention paid to conceptual¬ization (Bernhardt, 1983b).
Comparing eye movements of LI and L2 German readers, Bernhardt (1983a) found that inexperienced L2 readers spent more time overall and spent longer sampling information from the text than experienced L2 readers and LI readers.




Methods That Work (ตอนที่ 1)
Methods That Work (ตอนที่ 2)

(ตอนที่ 2)