Language Proficiency and First year. University
Achievement:A case Study at the University of nar es Salaam
Sarah North. Lecturer                                   Utqfitl Vol. VII No.1, 198$, Jo_lof
Communication Skills Unit,                              ~!:e/l;:/ilor&u~.z-~sa.-,
University of Dar es Salaam
1. Introduetloll
       The University of Dar es Salaam, like tertiary institutions in many other'
Afri.can countries, uses EDiIish as the medium of instruction, thus RlQuiring
students to learn through vlhat is at best a second IllDlWlle, and for' some a
third or even fourth language; This situation makes great demands on a
student's language proficiency, not only in terms of the command of Jenera!
English, but also the ability to apply it in the exercise of such sophisticated-
skills as lecture comprehension,        notetaking, writing academic essays and
reading advanced textbooks in specialist. subjects. For lower levels of
education, there is considerable evidence of the close relationship betWeen
language ability and intellectual development.1 If such a relationship holds
also for tertiary education, then a student's conunand of English and ability to
use it effectively as a tool for academic study may be a crucial (actor
contributing to his succe~ at the university.
       As yet there is, however, no conclusive evidence that languaae ability does
greatly affect performance at university. Various studies have looked at the
association between language test scores and academic results. Research in the .
United States on the predictive validity of language tests for overseas students' '.
generally reveals that althougttthere are correlations with academic arades, the'
relationship is relatively weatt.. Success does not appear to be dependent on .
 Ian,guafte proficiency. Similarly, a survey of students receiving American aid .
 grants".'    showed that English language' tests. were not in themselves a
 satisfactory predictor of academic achievement, although they did correlate
 significlmtly with. performance,' particqlarly if the quantity as well as the
 quality of work was .co~idered.In_Britain,          A. Sen (1970) investipted the
  academic results of 2300 overseas students~ who had taken the EPTB (Davies)
Test, and found, the correla~ion so low that it "seems to indicate that the extent'
of the use and fariilliarity with the Eng1Wl!language has little relevance to final
performance.'                                                  ..
 Iil all these studies, the results have been reported in conelationalterms,              thus
 indicating the strength of associatidi;1 through the whole raDge ot tanauaaC'
 ability. It has. been suggested, however, that such an approach is.
 inappropriate, and that for practical purposes, it is more important for a test'
 to distinguis.h between stuW;nts whose language proficiency is adequate and
 inadequate for .their st\ldies;'    Using banding criteria of this kind, E.          Insram
 (1973) and K. James (1~80) have claimed success in identifying students 'at
 risk' of failure through lfuguistic deficiencies;'.
      , From the research reported above, it appear$ that althoUlhIanauale               ~ys
 some part in deternrlning the success of an overseas student, it is not a crucial
 factor. However, the findings of A. Davies (1%7), obtainerl at different
 institutions in Britain and els~here, indicate that the usociatioD between
                                          27


    lctnguageability and academic perfOl,nal1c; may show great vlwltion from
   plac\: w place." These results should :'ilen us to the danger of
   overgeneraiisation. It ishighly likely that the rehtion~rjp is not ,;onsistem, but
   differs according to the circumstances of the palikular educational situation.
   Thus results obtained under one set of circumstances cannot be assumed to
   apply under another set of circumstances.
        The studies mentioned above pIlmarHy concerned overseas students in
  English speaking countries, and we ;nay therefore question their applicability
  to English medium institutions in other parts of the world, since the situation
  will vary in several important respects. Firstly, students in a foreign country
  will race very different problems of social and cultural adjustment which may
  affect their performance. Moreover, the selection process is likely to ensure
  that such 'overseas' students are not representative of the student population
  in their home country, and language ability may be one important point
  difference. Finally, the composition of staff and ,student bodies 'at home' and
  'overseas' will differ considf;rably, so that there will be great variation..in the
  attitudes and expectations of lecturers towards their students, with consequent
  effect upon teaching styles and criteria for evaluation. These differences will
 probably affect the extent to whichlanguage prOficiencyis a sign~ficantvariable
  for 'home' and 'overseas' students; and therefore make it necessary for each
 ~O\lntr"to assess the importance of the language factor in its o\vn educational
  institutions ..
        Data fwm non-English speaking c0untt:ies shows a more consistent
 relationship between language test scores and' academic resUlts. A. Heron
  (1970) found that mid first year grades at the University of ZambIa were
  strongly correlated with English proficiency in the Humanities and Social
,Science,s. but not in the Natural ScienceE (with the exception of
  mathematics). 10 At Haile Selassie I University, English langu:1gc tests were
  found to be the best predictors of first year grades, while most other
 educational and social factors failed-to show any signifiCant correlatioti1
 Similar results have been obtained at the University of Khartoum, where
 School Certificate English and the preliminary year examination in English
 correlated most highly with examination results in academic subjects.'12
       Unlike the research on overseas students in E,nglish speaking ~ountries.
 these studie,sindicate a clear association between language ability and academic
achievement, at least in the tirst year of university, ~uch a relationship has not,
 ho~ever, been demonstrated at the uruversityofDares Salaam. In 1%7, E.L.
 Khngelhofer conducted a survey of first year students involving a
  language-free intelligencetest and a questionnaire. He found that neither the
intelligence test nor Higher School Certificate results were accurate predictors
of examination performance. and concluded that:
         achievement in the University Cof/ege is Probably more importantly             a
         Function of Proficiency in English than any other single factor13
This cla~m,h0'Yev~r,was not b~sed on emp~ri:'?Je~dence, and later attempts
to corcoomate It failed to establish such c; rehmonshlp.
        L If~;!S\,W;H. Whitl~!e;tte~t':JdaU tlni p:ar education students in listening
skjJ!~f':.a i:omprehension if! English.,a,rl;7 CODQ:i';ted a survey of their linguistic
Ql1d, : ;;'~'l  ,Wf';;; backgrom.a . Thi:>st:.<dy 'k;t" not reported because the results
',;-th _,c.,!!'ly r:e~atl~'e.The h;HJ1hli'S, briefly ~i'ieh in ~, Anderson (1975)•.14were
;.1'" 'x"~,i.'~ tJ.nal re:nllts showed 1\0 corrdation With the, language tests, with


   attitl;1des to E~sh,       wi~h self--assessment of language skills, or with School
   Certificate English (written paper). Anderson took a different approach
   investigatilll features of students' written work in relation to their first yea:.
   results. Again the. findings were largely negative; the only significant
   corr~latioR was that students who wrote with greater grammatical simplicity
   tended to perform better •.                                     '                .
          Th* findings are Sllrprising in View.of the positive results obtained in
   Zamb~, 15 Ethiopia,II!i•. and the Sudan."            One problem may have been the
   failure to control for different subject combinations contributing to the overall
   fll'st year results for different.students.      J.A. Upshur (1967) has shown that
  'predicti~ by language profICiency tests may vary from one acade01ic course to
   another,:I'      and the results of Heron .support this observation}~         If a final
   aca4emic result is derived from different course combinations for the
   individuals in the sample, then these variations n' ay obscure the relationship
   between language ability and academic achievement in each co.urse.
          In discussing the more positive results of earlier studies, Anderson himself
   susgests that they may have been affected by the test format used, claiming
   that:
             0';' English test will predict'the result~ of another test in English if it is
             similar in form when the individual's language performance ..... is meas-
             ured in a way which is nf}' related to the examinattl!ns o~ tests use.d t~
            .measure educational achievement then the results Will be mconcluslv;
   Although his study avoided the possible distortions cauSed by similarity in test
   format, this was achieved by concentrating exclusively on writing. Since details
   of Whiteley's tests in listening and comprehension have not been given, it is
   impossible to determine the t:e$SQJl£. fOL,the .Jt\ilu.re .~o find a significant
 . asS?ciation between language PfO.~cien~ ~d acadeptic: perf0l'D!ance at the
   Unaversity of Dar es Salaam. However, in VIew of the general hebef that such
   an association does exist, and the consequent concern about the lanauage
   standards of University students?l'           it is appropriate to look again at the.
   effect of language factors on students'academic achievement.
    2. Method
          This study investigates the extent to which language proficiency is related
    to academic performance among fll'St year students of medicine and dentistry.
    The advantage df limiting the scope of the research in this way is that these
    students follow the same courses and are ass~sed by the same criteria, thus
    avoiding the problem of comparing non-equivalent results. Excluding those
    students who left the uni~ty           during the first two terms, the sample of 57
    students comprises 78.. of the total fll'st year intake in the two courses. (The
   mnainins: 16 students misse4oi1e or both of the language tests due to late
   arrival. and thus could not be included in the sample) •.
.         Academic achievement is measured by ,tJ?c"!Jlid-y~ coursew~rk resul~
   (CWA), which are based on tests in anatOOlY, blOchemistry. phYSiology and
   ~vioural         scieo£e, administered at reauJar intervals throuahout the f~S! t,wo
   terms of study. For each course. an average score is given outor 100. and, thC:se
   SCO£eS Were avergedto{)l'ovide an o'{erall meuure of ~onnance.               The use 9f
   coursework tests as the criterion has the advantapthat,            'sJnce. th~ COnsIst
   aItJ10st eatirely of multipl&cboice itesns. they require, ,.iri themse~, less


 'linpistic ability than mOlt other fOl1lll 01 ....                      mcnt. In an essay, for
  example, it is often dlffic:ult to determine whether a student has done badly
  because of poor expression, or b«aUle 01 inadequate understanding of the
  subject,» In a multi"l~le$t;however,~r                            pcrfoonance ismorl.' clWly
  related to ,ftwlequaie ,knpwled..ae. Thu. a co~on,                     with lllDAP.Ulle.ablli~ i~
  unlikely tf' m~ca\C ,merely .kiJf in      'test.~               m ~'.        and may proVIde
  evidence thattUnauistic:deficiendOi contribute to ditrIculties in actual learning,
  ratber than jUit expiellina'what has been 1eItnt.                                          ,   .
        Courseworlc results were contpartd With sceres in two language tests, both
  BdtPiniilteredin the first week' 01 term 1. Tbo University ~                          Test (USn
. is a telt ofBnaliah Ifammlf, detianed to retlect: the student's.ability to use the
  grammatical system in academic diSCOUfIC, rather than theoretical knowledge
  of the lanauaae. This test is applied throuahout the ~versity,                        and is not
  oriented towards any particll1ar dilcip1ine. the Study S~$ Proficiency Test
  (SSPT), on the other band, it dcsiped specifically for the Faculty of Medicine,
  and me&sU(es the student', IlQll in visuaHaing written information, taking
  D.Qtesfront an aural iaput, rcadina COIPprehenaien and writing, as applied to
  specia1i1t IUbjcctmatter. 80th theIc lanauage tests differ substantially in
  format from the courteWwlc tats (and frolll each other), thus eliminating the
 pouiDle effects of test ahnilirity mentioned in the previous section. To provide
  Q overall measure of lanauqe       ability, UST and SSPT scores; each of which is
 Biven as a ~taae,          were standardised and summed to produce a total' score
 out of 200, iDdicated below .. LA(1anauaae ability).                                              '
        The study also cansiden the intluci1ce of tyPe of entry (mature or direct),
 and &hePorm VI-results offered "'qualifications for entry. Most students had
 offer,,ed biol0IY, ',chtmistry an,d ph)'lics at,                  'pal Level in the Form, VI,
              11'-
                                                     Princi, ", •
 ~tions.           awl JUathematk:s III SUbsidiary 4vel. A score w.as allotted to
 eaCh ~                as,tplIP'WI;~-6,~5~p-:4'D-3,                     B-2, S-l, F...".().and the
 tOtal (out ~f a tlUIOrctica.l maximum of 19J was l'ecorded for each "student.
        Meana anAl staDdIriI doviatiool of the maiD variables are shown in Table 1
 below. It should'" noteel tUt th... tipres are not directly comparable;as the
 Hfi.t11 __ 4ifrtr.                                                                '
                          CWA         t1ST         SSPT                 LA        VI
             w.&:'         st.1       7t.7        51.'                99.5        8.5
             'It:           '.0        13.7        13.6                 18.1        4.6
                                                                                     "


      3. RESULTS
       <.Usl ;~tlatioli        ma~rix for coursework results (CW A), language test scores
             ,    .   and LA) and Form VI results '(VI) is given in Table 2 below.
         Table 2: Correlation matrixjor the$. main variables (n = 57)
                         CWA               tJST'      SSPT                   LA               VI"
        VI-             .304 •            ;252        •258                 .286*
     :, LA             .669**              .902**     .901**'                              .286*
        SSPT .         •622**              .626**                        .90'1,**           .258
       UST              .57S* •                      •626.*              .'902."'.'       .•252
      CWA                                 .S78 ••     .622*'"             .669 ••          .304.
                    * Sigriificant at 5'" level           *.signjficant at 1'Ie level
     These'h:sultsindicate that all three language measures are strongly related to
    .mid~first year colirsework results in medicine and dentistry, with less than lOJ.
     probability that this association has occurred by, chance. Form VI results
     snowed only a low correlation with coursework, though this was still
     significant at 5OJ. level. The language tests correlated highly with each other,
     but not with results in Form VI,suggesting either that performance in Form VI
     examinations is not related, to language ability, or possibly that changes in
     language proficiency have obscured such a relationship by the time a student
     enters university. Despite the direct relevance of subjects taken 3t Form VI to
     first year courses in medicine and dentistry, it appears that these examinations
     are poor predictOrs of academic success, and language proficiency is a more
     important factor.
           Further in~tigation of Form VI perlorniance involved correlation of the
   scores in individual subjects. The results. presented in Table 3. indicate tIurt
   Form VI physics was. tbe. best. prcdi~oL .of,mid.~first year aqhievement.
   followed by chemistry. The correlations. however. remahied below the level
   obtained for language proficiency~ ..
, "SlirprisingIy. there was no significant correlation with Form VI biology
    Mathematics, however. correlates with scores in biochemistry. which was'in
   .general the subject for which Form VI.results were the best predit;tor. In every
   .other subject, language ability proved to predid: success more accurately tban
   Form VI results.               ~                                                   , '
           Mature and ,difect eutrants show considerable variation in the
                                                                                  .
   qualificattom offered at entry. and this Jnayhave affected the results discussed
                                                  31


   TaIM J; Correlation mfltrL~/or Form fli results (n :: 51)
                   CWA          .physiololY       Anatomy            Biochem .      Deb.Se.
    VI             .304-               .086            .302-            .'73- •          .117
   Biology
   Chemistry
                    .062
                    .27.-
                                       .021
                                        .061
                                                        .014
                                                        .281-
                                                                         .~_.
                                                                          .1'4         -.0'9
                                                                                        .147
   Physics         .3S2--             .130           .35S"               .'84--         .141
                                     -.12S             .053            .302••          -.047
    Moths            .040
                   .669.-           .413"             •610- •          .572- •       •589••
    LA
    - significant at  s.,. level              .. significant at I" level
  above. To check for such an effect, correlations were also calculated for each
  group of students separately. It was found that Form VI results were a more
  accurate predic:tor of success with direct entrants, correlating at the ''''level of
  significance. In c:ontrast, there was no sianificant correlation for mature
  entrants.      Sin«        some mature         entrants     do not offer         Form      VI
  qualifications, mulll at Form IV were also considered in these cases. Again,
  the correlation with coursework performance was not sipificant; in fact the
  figure obtained was fleaative. As the mature entry scheme is designed to free
  university admissions policy from dependence on examination results, this lack
  of assodation is to be expedt4.
  Tjpe.l....,
        In view of the coneem about the academic performance of, mature'
entrants in the Faculty of Medicine,2I it stemS appropriate to consider the
rolative ac:hin~nt            of both aroupa of students. CO\U'SeWork results were 'in
,cnerallower for mature than for direct entrants, but the correlation between
type of entry and cou~work                   was .269, which aniy just reaches ,.,.
si,nificance level. However, mature entrants also ~ored lower on the language
tesll, so at least part of the variance may be attributed to language proficiency.
To check for this relationship, the Hest was used to measure whether the
difference between the two aroups was significant, when the language factor
was controUcd. The aver .. e CW A SQOrel are shown in FlJUl'e 1 below,
IlCllOrdinato whether students J*Md or failed the SSPT 14
        ~t wi!l be _n tbat for both types of entry, ttlere was a consistent
 relati?"~blp"ween lanpqc ability and courscwork results. The difference
 wa SIJIU~t           at the 1'1t level for diroet entrants (t - 2.83, 14 df), and at the
 S'1t}evelfor ~               entrants(t .. 4.31, 39df). Althouahdircct entry students
 performed better than mature entrants at both levels of language 'proflCiency
 the differelWe was not sianificant either for thOle who passed SSPT (t = 1.3:
 30 df) Qf fer these w~ failed (t .. 1.0, 23 df). In short, the lower results of
 mat ..... _traIlts 4lOIIlpand to dirtc:t entrantl ....           to be largely a function of'


                     64
                     62
                     60
                                                            direct entfllnts
                     58
             !        56
                      54
                                         mature entrants
             ~        52
                      SO
                                         SSPTPASS              SSPTFAIL
                                                                               ---
                     Fig. 1: The relationship between CWA results, ty.PJeof entry,
                     and. SSPT scores                     .
  their language ability. The explanation may be that mature entrants have
  gener'll!y :;P':nt sOme time a,way from an En3lish speaking environment, and in
  particular. have not been required to use their language skills for academic
• study. It is therefQre likely that both their linguistic competence and their study
  skills in English .have declined in the period between leaving school and
  entering the University.
        Both lanlU81e tests were stro~gly associated with e(Jursework .... utts. and
  when Nmbined. showed a correlation of .669, which is well above thet.,. level
  of significance. Languale .ahility appears, there.for.." .te be. a fairly .leCurate
  J!odictor of aeademic success in first year studios. ThiJ Nlatioftllhip is
  illWJtrated in Figure 2. in which langua" a~d coul'IOwork scores are plotted
  ~        each  other.


   In addition, the language tests clearly serve the purpose for which they were
  :designed, in successfully identifying students 'at risk'. If we compare those.
   scoriDi above and below half marks on the tests, we find that only 3070of the
   first group had a failing average in coursework, whereas for the second group,
   the corresponding proportion was 32 0J0 •.
         ABwould be expected, the Study Skills Proficiency Test, whiCh is directly
   related to the comm~nicative d~ands of courses in the Faculty .of Medicine, is
   a more accurate predictor of achievement than the University Screening Test, '
   which measures gram',uatical ability in general acadelJlic English: Of the
   various components of the SSFT, reading comprehensiQll (sedion C) was ~ost
,highly related to cOl,lrsework results, followed by ndtetaking from an aural
   input (section B) and paragraph writing (Section D). SCction A, :on the
   visualisadon ofwrittcn information, had relatively low correlation with CWA;
   and with the other measures of lanauaae proficiency •.                         ~ i"
         Table 4 below shows the cOrre18tion • .of the different components oft)fe'
, lanauaae tests. 'Ibis data indicates that the SSFT ec>rrelates hiahly within itself
  and, with thi UST,.upportbla        ~e 'usumptlon th.tdlese        tests do   in  fact ,
  m~ur~the '.arDI ..  thiDl, nune1.Yoverall. 1....    '. abUitY~ ...BnaUsh~ Th,only
   excepti~~ ilHCtlon !L, 'w~~ 11 not '1anl~t1y
                                                   .
                                                          co~,
                                                                , ,
                                                                       ~th  ',scc;»reson


ei~ SSPT or UST. It &,pptUI tlIattlais lNtie ...                        y have iavelved • visual'
skill not highly associated with lanauaae alMlity.
                                               T.w4
       Tobie '4: Correlation mtltrix f"r COnqlOMnts of the lonpa,e
      'tests.(n ='57),
                    UST          SSPT              A                J!l              C              D
 I   ,CWA
        UST
       SSPT
                    •51S"'!
                  .626 ••
                                .622"''''
                              .6U ••
                                                .373 ••
                                               •199
                                              .,~    .
                                                                .463••
                                                              .495.'"
                                                              •844 ••
                                                                              . ~.
                                                                               •544••
                                                                               •73' ••
                                                                                              •455 ••
                                                                                              .527 ••
                                                                                              •785 ••
       A          '.199        .530*.                            .208           •303 •          .304 •
   "B            .495 ••       •844 ••          •208
                                                                  -           •400**          .634 ••
                                                                                             .4"'••
                                                                                   ,...
       C         .570**       .73'••            .303 •         •400* •
       D ..        527 •••      78$ ••       " .304 ••          634 ••         ....      "
     .• significant at 5~ level .•• siJniftClflt at 1~ lev;l
           Considering the results for oaeh uadoqde ceurse IOpItatolY. weftad thet
    inanatomy, bioeh~                 ud bebavioural science. sectiona B, C and D of the
    SSPT were strongly correlated with eoursework perf01'8WJCe. while the
   ,correl~on for section A was couklerably lower. POl physiolOlY. however,
    the pattern was reversed. and sedioil A was the 110st accurate predictor of
   .perfoJ1!18;Dce••Physiol,?SY also showed the lowest correlation with the JaDauage
    tests. suggesting that.lt may be Jes& dependent on 1801uIP ability than the
    other courses. It should be IItrosaed that tanaua&o was a lriIbly aianifteant
    variable in the coursework reaulta for all .ubjcea. but ~                                        this
    variation indicates that its irnportaDClllaY differ from Oo~to course.
    4. DISCUSSION
           The results. of this study indicate that Iauuale ability is a JOOd predictor
    of success in first year Studies of medicine alUr dentistry, as measured by
   mid-year comsework results. The tests of Jll1Q1UUltieal ability and study skills
   in English each correlated with CWA at the' 1.,. level of sipifiamee. By
   CQntrast. the association between' CW A                and     Fonn,~VI resulta. in subjects
   closely relat. ed to. the first :rear.syUaltua.wu rdativ1.::;.                   and m tM ease of
   mature students, not statJltiee$Y .~.                    'DtM             ,latd IClIMIUppert
   ~~:=:in~~lr~::~
           We would D04. of cdurse.' • up,eet to tiad      ... that~
                                                                                           ..t...  ~
                                                                                       ad1ievement is
   wholly predicted by 1atJauap abUitJ; if tbis were 80. no nadv~
   student would ever fail at ~.                     Clearly other facton mustpJay a part.


 and a more extensive study woUld need to consider ~ .contributionof
 variables such as motivation and reasoning'ability. In addition, it would be
 desirable to investisate the tole of basic scientific knowlcdse; it seems unlikely
 that its importance is as low as suggested by the findiop on Form VI
 exaDlinatiom, which are perhapS not a valid measure of sfudeots' subj«t,
 knowledge at the time of university entrY. AIIowina for the effect of other
 factors, it nonetheless appears that language prof'JeiehCYis a hisblY important
 factor ii1 university succeSs.                        '.'
       The purpose of the language tests described here is to identify students 'at
 risk' of failure due to linsuistic <J,eficiencies. This purpose, itself presupposes
 that,.there i~ a .'threshold level' - a DlinimUDl standard of E"8lisb prof'lCiency
 necessary for satisfactory academic perfonnance. We would not DecessarilY
 expect that every student whose languaae ability is aboVe this threshold level
 will perform wdl in his studies, since other faCtors will a1ao play their part •
.SiDlilarly, a student with a weak command of the English ~.-y                   be able
 to compensate by, for example, putting in more work than his fellow students.
 However. we would expect that students who show weaknesses in language
 proficiency will have a far greater failure rate than those whose languaac
 ability is adequate ••.                                                     '
       TWs hypothesis is supported by the results of the present study. As FJIW'e
 3 shows, students who obtained less than half'markson language ability (LA)
 were far more likely to perform poorly in their coursework.
                                        Figure 3                    '.
 Fig. 3: Relative performance on CW A for those scorin. above           n..ll.at_w   balf
 marks on LA                                           '   -      .     -~
     12
                                                                       LA:
                                                              above halftimrb      0
                                                             below half marks      "tU,
                                                                                   . , 11'.


        It would be raSh, however, to try to draw any fum conclusions from this
  data as to the precise score corresponding to the 'threshold level' required for
  satisfact<?ry academic st~dy. Firstly, the scope of the investigation is too smaIl;
  research IS needed on abirgersample of students over a longer period of time,
  and this work is still in progress. Secondly, the language tests are Used as a
  basis. for recommending students to follow courses designed to improve 'their
  EnglIsh proficiency, and we would need to consider' the effect of this
  additional ieaching on academic performance. Although it is too early to
  assess the impact of these courses on students in this sample, the data available
 does suggest that those who participated tended tO'do better in the coursework
 tests. Finally, iris premature to identify a student as successful or unsuccessful
 on the basis of mid-first year results alone, since those whose initial
 perfo~,mce is relatively weak may in time catch up with their fellow stpdents.
 Ingram bas noted that the relationship between language proficiency and
 academic results is more marked in first year examinations than in finaI8:~
 This is partly explained by the fact that some of the weaker student drop out
 without completing their studies, but another factor is that duri1is their course
 of study, students with a poor coinmand of EnsHsh on entry may improve
 sufficiently to perform well at a later ~tage.jl1 For these reasons, it is not
 possible to state categorically that a specific ~               score represents the
 minimum requirement for academic sucCel$-.We can only say that beloW a
,certain level, linguistic weaknesses may seriously jeOpardise a student's chance
 of satbfa..c.tQJ;yPerformance in ,his studies.             '
       ,In pra,ctice, academic success depends, upon the ability both to learn
 relevant skills and'mrormation, and to demonstrate adequate learning through
 the forms or-assessment which apply in a given course of study. If students are
 assessed ina way which makes, great demands on their langUage',sJdUs,this may
 be reflected in a, close correlation between academic results and stores on a
 l~nguage test. As noted in section 2, the criterion for ~c              performance
 used in this study was a series of nfultiple.choice teats, and it seems likely that
 in the testing situation itself, languaae ability is less crucial than, for example,
 in an essay-type examination. Nonetheless, poor performance in coursework
 may be partIy attributable to the effects of being tested in a foreign language.
 In a study of Spanish-speaking students, D.L. Alderman (1981) found that a
 minimum language score was needed be'fore an academic aptitude test given in
 English reflected the student's true ability as measured by a similar aptitude
    ftnWgn    extpIf".
 test given in Spaqish, and concludes that this is because:
    ,tM /anIUIIg. sldlts ~
                           with "poor commtI1td oftM E1I6lUh ~
                                                                       .
                                      for Mmonstrtlti1r8 thWr ability. •
                                                                          ••• m4Y ltJc1c
 It would be interesting to investlpte whether this effect is ff;)U.Dd in tests and
 examinations of the University of Dar es Salaam. If, for example, students
 with low scores on UST performed better on coursework tests cond\lcted in
 Kiswabili than on ~            tests conducted in English, this would indicate how
 far the ~act of beini _till thfough the med!um o~ English obscured th.eir true
 academic attainment. On -the O'ther hand, lf they performed P.OOflym both
 """""g
 tests, this would sugest that their difficulties were more ciosely related to
            throllgh the medium of EnaIish. ~                             '
     , 10 the present study, teSting effects were minimised not,only by the use of
 multinJe..choice coursewod: tests but also by the considerable differences in
 fofaW.~CKm::          '~k              ~d Janauale tests. The fmdings clearly


  contradict Anderson's claim that in such a case, evidence: for the assoCiation
. between lang;uage performance             and academic       achievem~nt     will. be
  inconclusive.'30 On the contrary, the strength of the correlatIOns obtamed,
  makes it unlikely that they show the effect of language in the testing s!tuation
  alone. More probably, achievement in coursework has been also affected by
  the students' ability to use English effectively in the process of learning.
         In the learning process, the receptive language skills of reading and
  listening would appear to be of more importance than the productive skills of
  speaking and writing. Although a student's languageability!s          most evident to
  lecturers through his speech and writing, the less observable receptive skills
  account for a much greater proportion of his study time, and constitute the
  major learning activities. 31 Some support for this order of priority is
  provided by the results of the SSPT, in whi~h reading comprehension proved
  to be the best predictor of academic success, followed by notetaking from an
  aural input (involving a large eletnent of listening comprehension),              and
   paragraph writing. Although differences in predictive validity between these
   skills were not great, it would clearly be unwise to.focJs on writing as requiring
   particular attention. A course in'study skills in Englishshould rather etnphasise
   reading and listening comprehension.
         Results obtained in the Faculty of Medicine should not' be assumed to
   apply equally to all faculties. The overall itnportance of the language factor in
   both testing and learning situation,s. tnay be greater or smaller, the 'threshold
   level' may be higher or lower, and'tne relative weighti~g of language skills may
   also vary. As Upshur has shown, the prediction of w;adetnic success frotn
   language ability differs according to the teaching situation and the type of
   language to be comprehended.32           In general, the highest correlations were
   found when the language used was relatively unrestricted expository prose,
   with literary language showing lower correlations, and specialised expository
   prose least of 'all. Laboratory courses showed low~r: correlations than seminar
   or lecture courses. The situation in the Faculty of Medicine lies towards the
   lower end of this scale, since it involves tnainly laboratory and lecture work,
   using specialised expository language. Thus, if the pattern described above
   holds true, we might expect to find an even stronger association between
   language proficiency and acadetnic achievement in, for exatnple, the Faculties
   of Arts and Social Sciences, and Commerce and Management.
          A second source of variation between faculties is the nature of the student
   body, as determined by university selection procedures. D. Douglas (1977)
   draws attention to differences in the predictive validity of School Certificate
   English among the various faculties of the University of Khartoum, and warns
   that they do not necessarily indicate that English proficiency is less important
   in one faculty than another, but rather reflect differences in admissions
   policies. 33 If students are selected in such a way.that their range of language
   ability is relatively restricted (if, for example; they are better than average),
   then naturally this Jeature will be less significant in differentiating between
   them than it would be when the range is less restricted.34          Analysis of UST
   res.~lts throughout the University of Dar es Salaam suggests that language
    abilIty does. vary from .faculty to faculty, and this may cOlnsequently affect
    the correlatIon between language ability and academic performance:
    .. Although f!lr!her investigation may reveal varying language requirements
    In different facultIes.   it seems likely -that English proficiency is a significant
    factor in. acattem'ic achievement throughout the university, andrese~rch
    currently 10 progress on the validatiQn of tbe UST does indicate an association
                                               38


  ,m aIr faculties  between language ability and academic performance. Indeed,
   the results of this st:uqy merely provide support for the already widely held
   Yiew that linguistic weaknesses contribute to underachievement                at the
   Unive~sity. As the preceding discussion .has suggested, however," English
   profiCIency should be ~een as a factpr relatmg not only to the student's ability
    to express himself in essays and other assessable work, but to the whole range
    of communicative demands imposed by tertiary study. Poor performance may
    be determined less by the ability t~, m,eet the specific linguistic requirements of
    essays, projects, tests and examinations, than by the student's overall ability to
    learn through the medium of English.
   S. CONCLUSION
         The language factor in university education is inextricably bound up with
   the whole educatio}} system, and in a multilingual society such as Tanzania, the
   situation is complex. An indication of the far-reaching issues involved is given
   in Macmillan's discussion of a similar state of affairs in the Sud8.l1,35 and a
   recent collection of papers from the University of Dar es Salaam. 36
         In the post~independence period, attention has rightly been given to the
   development of Swahili as the national language, which is gradually taking
   over many of the previous functions of English, including most educational
   functions. Though it will eveptuaIIy become the medium of instruction at the
   University, English will undoubtedly retain a role as a means of access to
   internationally disseminated information, as in many other countries, both
   developed and developing. In particular, if the University is to maintain its
   academic,standards,     students will continue to need a high level of proficiency'
   in reading in Eng~ish .
       .Whatever happens in future, however, the. needs of the present generation
   of students cannot be ignored. If English proficiency is as crucial a factor in
   academic success as this study suggests, then continuing attention must be paid
   to alleviating the problem of language deficiencies at the University. One
   approach is to raise the standard of English in secondary schools, and in
 . a speech in 1982 to secondary school headmasters,37                the Minister for
   Education called for greater emphasis on English at school, reaffirming its role
   as a tool for learning. A second possibility would be to change the university
   entrance requirements to include a minimum qualification in English .. It is
   doubtful, however, whether this is a feasible step. The only avaIlable
   qualification isthe Form IV examination in English, and as Macmillan points
   out,38 it is not certain that such examinations are reliable, nor that they
   measure the skills relevant to university study. A third approach, and one
• currently followed at the University of Dar es Salaam, is to deal .with t~e
   problem 'in situ', by providing remedial courses for students WIth baSIC
  weaknesses in the English language, and functional courses to develop relevant
   study skills in English •.
        Every student who f~ils to complete h~s uni~ersity ed!lc:yn ~~Pli'::~~~i~
  waste of national resources, .and. when faI:ur~ IS ~r1n~.:i1uJtalents           badly
  needed in the national developtpent of ~~        r
  rather than intellectual defiClenCJ~s, a ,,:as e so
                                                      vel
                                                          ::d~~
                                                                 wer For this reason,
                                                                 study should be seen
  development of the language skills r.equI~t or ~~ulum contributing to the
  as an important part o! ~he. urover.sI '! ?um                ,
  successful training of SpeCIalIsts III all dISCIplInes.
                                              39


 I. I should like to thank the staff of the Comp\1ter ~entre, University of Dar es Salaam
     for their assistance with the statistical calculations. and Prof. P.M. Rea for valuable
     comments on the first draft of this paper.
2. See, for example. J.W. Oller, "The lanauage factor in evaluation of bilingual
   . education". In J.E. Alatis (Ed.), Intemational Dimensions of Bilingual Educailon,
     Georgetown University Round Table on Languages and Linguistics. Washinaton:
     Georgetown University Press, 1978.
3. These tests include the Test of English as a Foreign Language (TOEFL). the Test of
     Emilish as. a. Second Language (ALI/GU), and the Michigan Test of EDJlUh
     Language Proficiency ..
4. B. Spolsky, "Do they know enough English? Some notes on the problems. of
     assessing the proficiency in English of foreian students". In NAFSA, selected
     ConferencePapers of the Association of English as a Second LAnguage. New York:
     National Association for Foreign St~dents' Affairs, 1967.
5.   AACRAO, The AACRAO-Aid Participant Selection and Placel1Jent'.Study.
     Washington: American Association of Collegiate Registrars and Admissions
     Officers, 1971.
6.   A. Sen, Problems of OverseasShidentsand Nurses. London: National Foundation
     for Educational Research, 1970.                                   '
7.   A. Moller, "A case for acrudetestoverseas"~ InA.P. Cowie and J.b. Heaton (Eds)
     English for Acatlemic Pu;pOses: PaPers on the language problems of ove~
     students in Higher Education in the U.K. ~eading: BAAL/SELMOUS, 1977.
8.   E. Ingram, "English standards of f<;»reignstudents". University of Edinburgh
     Bulletin, Vol. 9, No. 12, May 1973:4-~.
     K. James, "Survey of University of Manchester overseas post-graduate students'
     initial level of competence'in English and (heir subsequent acaderiUcperfordlance:
     calel;ldaryear 1977"..In the British Council, ELT Document 109 - Study Modes and
     Academic Development of OverseasStudents. London: the British Council, 1980'-
9.   A. Davies, "The English proficiency of Qverseas students". British Journal of
     Educational Psychology, Vol. XXXVII, part 2, June 1967: 165-174.
10. A. Heron, "The relative ir1tportanceof reasoning ability and of-English vocabulary
      and COrilpr~hensionin first .year"university achieveritent by Zarltbian students".
      In T.P. Gorman (Ed.), LAnguagein Education in Eastern Africa. Nairobi: Oxford
      University Press, 1970..              "                 .
11. M. King and J. King, "S~e co~rel!ltesof ~versityperfonitance            in a developing
      country: the ,case of Ethiopia". Journal of Cross-Cultural Psychology, Vol. 2,
      No.3, Septailber 1971:293-300..               '                                 .
12. D. Douglas, From School to University: The Study Habits Research Project Final
      Report. Khart(lur!l:Khartour1tUniversity ~ress, 1977•.                           "
13. E.L. Klingelhofer" Studies. of ta~nlan             Students. Dar es Salaarlt University
      College, 1967(mimeo): 2-34.
14. R. Anderson, "The langul!.ie factor ~ educationalperforrltance at university
      level". Journa~of tM LAnguage Association 01 Eastern Africa, .Vol. 3, No.2,
      1975: 138-163.
IS. Heron (1970), op. cit.
16. King and King (1971),op. cit.
17. Douglas (1977), op. cit.
18. l.A. Upshur, "English lan~e,tests             and prediction of acaderiUc success". In
      NAFSA (1967), op. cit.
19. Heron (1970), op. cit.                                '                  .
20. Anderson (1975), op. elt.
21. This issue is reviewed in P.M. Rea, "The CoI1lI1lunicationSkills Unit: 1978-
      1980". In LAnfiuagefor EducatIOn, Vo!.'I, University of Dar es Salaam, CSU,
      1980..


   22. Some ptbblew associated witb the interpr~tiQn of student essays are discussed in
         G~Taylor., "Error analy~: three types of ~            levelllll1BU8le error". In Language
                                                          es
        [pr Educ~tlon. vol.;2L Univemty of Da,r Salaam, <.:SU, 1983.
   23. Faculty Board P~per FBP S.2b, Faculty of Medicine, Universtty of Oar es Salaadl.,
        1982 ....
   24. ~SPT ~<,lre.sal(lne ~~ ~           in ~l'     •. S~ce ~e puqx>se of the UST is to. screen
        m~g          students for serltius graDuilAtical proplediS, the tiajority of students
        score ~y,        and tht;.resul~      distri~u~on Of 'p~'         a,nd 'fails' makes it inap.-
        propnate for calculating adiff~r~ce In ~eans on this sample •
. 2S. Klin&eJhofer (1967), op. cit.
   26. IngrlUil (1973),.op. cit •.......
   27. In this.connection,it is worth netting that whDe dtature entrants in the Faculty of
        MedicWe b~ve a higher fail1:lfC~te tban d)roct entrants in the first year, this
        gap has disa~ed           by ~he (mal.year (Faculty Dqard Paper EXFBP 1/11/82,
        Faculty of Medicine, Universi~ of Dar esSaIaaUt, 1982). If, as sUilJested in
        section 3, the per(on1lance of mature entrants. is affected by their lower level of
        l~~uage protl~e~cy .. the~ this Wlj)t(~Veri1c:nt        ..re18tive to direct entrants i:ftay
      . slUUlar!Yreflect an lncr~m          lan~e      pron~ency during thejr course of studv.
   28. D.L. Alderrban,JJangl)ll~P.rOJIcJe,;cyt1SaModerat01'            YarliIbII in Tc,1tJ1.fIIicadel1Jic
 .      Aptitude •. TOEFL ~~Jt              ~eport 10. Pfi.t!.cdon': Etiucational Testilll ServicC,
        19~:~                                                          .
   29. Such an experm.ent would          need  to be designed to control for the fact that Swahili
        is itself a second lan~e       to dumy stui:lents.
   30. Anderson (1975), op.      at. '....
   31. In a surVey at $an :piego S4lte U~versity, ins~ctors !Vere 'asked to tank: the
        langtUlge skills rilost.~eeded in ~       courses.  ~eading and listening w,ere plac~ well
        above the productive sltiUs, with writing tanked third, and speaking Jasgi.ng far
        behind (A.M. Johns, "N~                  English: A faculty survey". TESOL Quarterly,
        Vol. IS, No. I, March 1981: SI-S1)
   32. Upshur (1967), op. cit.
   33. DouJlas (1977), op. cit •.
   34. J.P. Guilford and B. Fruchter, Pund4inJental Statistica in Psychology and
        Education, .6th edition, Tokyo:" McGraw~HU1Koa!lkuma 197~: 32S-327, '.
   3S; M,. Mactitil1an. '''A.specisof bilingua1isril in wjiversity education in Sudan" .
    .    lQ. Qprman (1970), op. at                                                               .
   36. H.R. TrappeS-Lorilax,R.M.             Desha and Y.Y. Mcha (Etds), Changing lAnguage
        Me({ia.Universityof~es~,1982.,                                                   .
   37. Reported in Dai!y News~Tanzama, 11th Oct~ber, 1982
   38. M~iUan         (1970), op. at.