Chapter Four:   Conclusions and Implications

           Three  questions were raised at the beginning of this  
     thesis:  1)  whether a model of the operational features of  
     language  learning  activities could  be  constructed  that  
     would  predict authoring system requirements for  producing  
     these activities as CALL exercises,   2)  what the range of  
     activities that it would be possible to produce through the  
     use  of an authoring system might be,  and 3)  whether  the  
     features which make activities meaningful and communicative  
     are  affected  by the flexibility of the  authoring  system  
     used to realise them.

           Three  separate purposes of the model were  proposed.   
     It  was  to  serve  a) as a partial guide  to  teachers  in  
     choosing  authoring  systems,  b)  to  assist  teachers  in  
     determining  how  a particular activity might  be  realised  
     through  the  use of an authoring system,  and c)  to  give  
     designers of authoring systems an indication of the type of  
     flexibility that is important to CALL.

     4.1   Examination of the questions raised,  in the light of  
     the study
             In  section  2.2,  it was shown that  an  authoring  
     system  functions by establishing a template,  or  formula,  
     for  an  activity  and then  soliciting  the  content  data  
     required  for each discrete item.  The formal feature model  
     involves  the categorization of exercises according to  the  
     behaviour  of the cue and the response-model,  outlined  in  
     section 3.7,  and the factors influencing the presentation,  
     answer  analysis,  and  branching phases of  the  exercise,  
     considered in sections 3.9,  3.10,  and 3.11.    The  model  
     permits  the identification of the inherent formal features  
     of  an  activity  and  the  translation  of  these  into  a  
     formulaic  description.  Where  the  model  can  reduce  an  
     activity to a formula,  therefore,  that activity should be  
     theoretically  realisable through the use of  an  authoring  
     system.  The question of whether existing authoring systems  
     provide  the degree of flexibility required by the  formula  
     obtained  from the model's application is excluded from the  
     present treatement.

           The   model  presented  here  has  been   tentatively  
     validated  in two ways.   First,  some measure of  internal  
     consistency   is   demonstrated   as   the   categorization  
     established  in  section 3.7 reflects regularities  in  the  
     presentation   and  answer-analysis  phases  discussed   in  
     sections  3.9 and 3.10.    Second,   in section  3.13,  the  
     model   was  successfully  applied  to  the  drill   sample  
     organized pedagogically.  All of these latter drills  could  
     be accommodated according to the model.

           A  further process of  validation,  not conducted  in  
     this  study,  would be to examine  existing  computer-based  
     activities,   to   determine  whether  they  could  all  be  
     adequately  described  in terms of the  model,  or  whether  
     there existed CALL activities that are not predicted by the  

           In  section 2.2,  CALL was defined as  consisting  of  
     structured,   mediated,  interactive,  practice  activities  
     involving the use of computers.   In section 3.1,  the role  
     of  CALL in the practice,  rather than the presentation  or  
     production stages of language learning was re-iterated.

            The  model  can be used to predict whether  a  given  
     activity  is of a suitable form to be converted into a CALL  
     exercise.   For the model to be applicable at all, however,   
     the  activity must be a structured one.    An  unstructured  
     activity does not submit to any formulaic description.  The  
     activity  must be mediated,  as well.   There must  be,  at  
     least,  the  possibility of two divergent  paths.   In  the  
     example of a presentation activity as shown in section 3.6,   
     the answer-set would be identical to the response-set.   In  
     other   words,   all  valid  responses  would  be  correct.    
     Another  prerequisite to computer-mediation of an  activity  
     is the author's ability to provide an indication of how the  
     correct response might be identified.    Candlin's  example  
     3.89,  discussed in section 3.13, exhibits a very-large-to- 
     infinite  answer-set and,  therefore,  must be considered a  
     production activity.  

           The  relationship  between the model and  the  drills  
     selected  as  exemplifying  meaningful  and   communicative  
     practice was explored in section 3.13.    All of the drills  
     could  be described in terms of the model,  indicating that  
     no  new operational formulae are required for  this  second  
     sample.   The  general pattern,  as well as the analysis by  
     discrete pedagogical features, suggests that meaningful and  
     communicative exercises,  as illustrated in the sample, are  
     not operationally different from meaningless,  manipulative  

           In  the  discussion of quality materials  in  section  
     2.3,  the  question  is raised as to the  degree  to  which  
     quality  depends,  on the one hand,  on the background  and  
     expertise  of  the author and,  on the other hand,  on  the  
     flexibility of the software.  Wyatt is quoted as indicating  
     that  authoring systems would fail to provide  the  quality  
     desired because of their lack of flexibility.    It appears  
     from the present study that the sophistication of authoring  
     systems,  in  general,  should  not  have  any  appreciable  
     negative  effect  on the production of quality  courseware.  
     (Individual  differences in authoring systems  will  always  
     affect   the   ability  to  realise  particular  types   of  
     exercises, however, as is pointed out below.) 

           The results would lead one to place the onus for  the  
     creation   of   quality,   meaningful   and   communicative  
     courseware  on  the  shoulders  of  teachers  and  teacher- 
     trainers,   rather   than  on  the  shoulders  of  software  

     4.2   The implications of the model to the purposes of  the  

           Teachers  are  normally cognizant of the  pedagogical  
     aspects  inherent in the types of activities they  wish  to  
     produce.   Before  they begin to examine authoring systems,  
     they need to become cognizant of the operational aspects as  
     well.   Similarly,  where  an authoring system has  already  
     been  selected,   teachers  must  compare  the  operational  
     features  of an activity with the options available on  the  
     authoring  system  chosen  in order  to  determine  if  the  
     activity can be produced in an acceptable manner.

          The   study  illustrates  how  slight  changes  to  an  
     activity  can transform a simple operation into  a  complex  
     one,  or  completely change the operational type.    It was  
     shown,  for example, in section 3.13, how 3.96(Dakin) could  
     be   radically   transformed  simply  by   adding   a   few  
     contextualising words.   A seemingly straight-forward drill  
     such as 3.29(Rivers),  was shown in section 3.9, to involve  
     many   discrete   factors:     variable   response   model,   
     elicitation of multiple responses,  linked responses, and a  
     large answer-set. 

           It  was shown in sections 3.9 and 3.10 how even quite  
     complex  activities could be modified to fit the  multiple-  
     choice approach,  allowing them to be realized on even  the  
     simplest of authoring systems.     Understandably, teachers  
     will  demand  alternatives to multiple choice.   It is  not  
     always appropriate and its overuse would certainly lead  to  
     a  negative  reaction on the part of the  learners.     The  
     second  solution to handling complex activitities or  those  
     with large answer sets was shown to be the construction  of  
     a pseudo-parser.

           Currently  pseudo-parsers  must  be  programmed  from  
     scratch.   The  fact that they can be described formulaicly  
     indicates that they could be included in the repertoire  of  
     "tools"  that an authoring system should  provide.    Their  
     construction  might  be possible by the use of  high  level  
     "verbs" discussed in section 2.2.    Programmers would have  
     to take care,  however, not to make the resulting authoring  
     system  too difficult to use.   As was proposed in  section  
     2.2,  various  levels  of  interaction with  the  authoring  
     system might be provided to teachers.    The beginner could  
     be   content  to  choose  from  a  variety  of   "matching"  
     paradigms,  while  the  more  advanced  user  could  create  

          The  need to generate  item-specific  stimuli,  rather  
     than read them from data,  was shown in several cases, such  
     as when the stimulus depends, in part, on material that the  
     learner  has previously provided.    Similarly,  there  are  
     cases where  the correct response must be determined by the  
     computer  on  the   basis of choices that the  learner  has  
     previously made.   These features must be made available on  
     authoring systems.

          An answer-analysis mechanism capable of distinguishing a  
     large  number  of  discrete errors that the  learner  might  
     make, coupled with a record-keeping mechanism that can keep  
     track of a large number of different factors,  was shown to  
     be  a necessary pre-requisite to allowing the  computer  to  
     build  up  an adequate model of the learner's  performance.   
     Authoring systems must be given this flexibility.

     4.3    The  limitations  of the study  and  directions  for  
     further research

           The  limitation  of  the study to the  more  or  less  
     traditional applications that are currently possible on the  
     computer  was  necessary in order to construct  a  feasible   
     model.   As  the  intelligence  and linguistic  ability  of  
     computers  is  developed,  there will  be  a  corresponding  
     decrease  in  the   need  for  such  a  rigid,  algorithmic  
     structuring  of  activities.   The  model may  have  to  be  
     changed  to allow for a wider range of exercise  types,  as  
     the creation of presentation and production activities with  
     an  authoring  system become possible.   Truly  intelligent  
     authoring systems of the future may contain a model such as  
     the one developed here,  and free the author from having to  
     consider all of these operational factors.   A point may be  
     reached  in  the not-so-distant future  where  authors  can  
     simply  communicate  to  the authoring system the  type  of  
     exercise  they  wish  to accomplish and  have  it  respond,  
     "Okay, done." or "Sorry, cannot do." 

           In order to keep the present study within the  bounds  
     of a reasonable length,  no attempt was made to embark on a  
     review  of existing authoring systems,  to establish  their  
     capabilities  and limitations in the creation of  different  
     types  of  exercises.   Such an investigation would be  the  
     next  logical step.    The present theoretical model  could  
     then  be further validated,  and possibly adjusted  to  fit  
     realities that are not foreseen at this point.    While the  
     model itself can be of some use to teachers in selecting an  
     authoring  system,   an actual survey of existing  systems,  
     based on the model, would be much more readily applicable.

Converted from the original Wordstar format and placed in Html package
Roger Kenner, 2006
[Return to Top Level]