Computer aided instruction in the skills of linguistic analysis

0360-1315;83!010025-07f03.00/0 Pergamon Press Ltd

COMPUTER IN THE SKILLS

Department

of Communication

(Rrcrired

AIDED INSTRUCTION OF LINGUISTIC ANALYSIS

Studies. Sheffield City Polytechnic. Totley Hall Lane. Sheffield S17 4AB. England 9 March

1982: amended

25 Ma!, 1982)

Abstract-Students on the B.A. (honours) in Communication Studies at Sheffield City Polytechnic study linguistics as part of their degree. These students seem to vary considerably in their abilities to analyse sentences. The difficulties that they experience seem to be the same for each successive intake even though the difficulties may not be shared by many students in any particular year. Thus. the teacher of linguistic analysis often needs to resort to individual student tuition to deal with the same problems year after year. The computer aided instruction program described in this paper seeks to overcome these problems. It enables students to progress at their own rate through a series of exercises designed to instruct them in immediate constituent analysis. The program was found to be very successful in achieving its linguistic objectives and also in producing a remarkably high degree of interest in computers.

INTRODUCTION The problems in reaching linguisric

analpis

On the B.A. (honours) in Communication Studies at Sheffield City Polytechnic, students study human communication for 3 years, using a range of disciplinary approaches including linguistics, psychology and sociology. The students usually enter the degree with three A levels, but not much knowledge of the theory of English grammar. Though they are very interested and willing to learn most aspects of communication studies, there is some resistance to learning linguistic theory and analysis. The students are introduced to the methods of analysis associated with three theories: Immediate Constituent Analysis, Systemic Grammar and Transformational Grammar. Some of the teaching in this area is done by means of lectures. but the successful development of the required analytical skills really rests on the practical work done in seminars. In tackling the practical exercises, the students work at varying rates. Some are able to grasp the concepts of noun phrase, verb phrase, etc. rapidly and so are able to analyse sentences into their constituents with little effort, whilst others struggle for weeks. What is noticeable is that the problems that the students experience are predictable, though not necessarily shared by many students in any one year. Given the problem situation just described, it seemed appropriate to devise a computer program to teach the required skills. Hence the development of the program IMCON. The structure

arld ohjecfiws

qf

IMCON

IMCON is a computer aided instruction program which teaches students to analyse simple sentences into their constituents and to draw tree diagrams representing their structures, In the program. “initial teaching” is carried out in small steps with regular testing of the student’s learning in its application to specific exercises. The various problems experienced by the students are catered for by means of branchings and routings through appropriate remedial parts of the program. In this way, the program allows students to progress at their own speed in achieving the objectives of the program. At the end of the program, the student should be able to: (a) (b) (c) (d) te) (f) (g) (h) (i)

state the “substitution criterion” for recognizing constituents; state the “distribution criterion” for recognizing constituents: analyse a given sentence into its constituents; define the concept of “immediate constituent”: recognize the immediate constituents of a given sentence: define the concept of “bracketing’* as a means of representing constituent structure; state the limitations of bracketing as a descriptive device; define the concept of “tree diagram” as a means of representing constituent structure; draw tree diagrams representing the structures of simple sentences;

STFWART

26

I~.YRSH.ALL

&fins the concept of “structura) ambiguity”: tk) draw tree diagrams of a given sentence exhibiting

(J)

structural

ambiguity.

In addition to these linguistic objectives. the program is also intended to serve as an introductton the use of computers and to create in the students favourablc attitudes towards their use. THE

PROGRAM

to

STRATEGY

Stages. jirmne.5 and structwe

The skills that the student is to acquire can be arranged in a hierarchy as shown in Fig. 1. The program is divided into three main stages. each one being concerned with the teaching of a subset of the skills shown. This teaching is carried out in a series of small units called “frames”. each frame being destgned to be displayed on the screen of a v usual display unit (VDU). In each stage there is a series of teaching frames and one or two practice and criterion frames. Trc~chinyfiames. as the name implies, are those frames in which new information is provided and students are asked to respond to it in some way. Prcdcr and crirrriou j-ames are ones which test that learning has occurred, the former providing a little assistance for the student. Thus. the program has an overall structure and content as shown in Fig. 2. In addition to the basic frames mentioned above. the program has decision frames. remedial teaching frames and summary frames. Decision fkmw are simply those points at which the student is given control of the program. i.e. he or she can choose to drop out. or go to some other part of the program. Ranrdid rwching frcwes provide extra information or explanation for those students with specific difficulties. .Srrntntur~~/rume.> summarize. paraphrase or otherwise repeat some of the information already dealt with so as to consolidate the learning that has taken place. Response\

trnd routings

At the end of each frame the student is asked to make a response. Sometimes

this response is simply to press the return key to continue. but more often the student is actually asked a question relating to the frame. The student’s answer will be evaluated as one of the following: correct: incorrect tin a variety of anticipated unrecognizable (because of mistakes

ways); or because

the response was not anticipated).

Drawing tree diagrams _ for structurally ambiguous sentences

’

Representing structure using tree diagrams

Analysing a structurally ambiguous sentence

Representing structure

Recognizing structural ambiguity L

I r

I

-

r

Analysing a sentence into its immediate constituents i

_ I

Distinguishing constituents and immediate constituents

criterion to identify constituents

Fig. 1. Lrurnlnp

Using distribution criterion to identify constituents

ta& huarchy.

CA1 m the skills of hngulstlc

analysts

‘7

i------I

Introduction

(

IMKEDIATE CONSTllUENT ANALYSIS

Analysing a sentence Substitution criterion Practice frame on substitution Distribution criterion Practice frame on distribution Criterion frame on analysis

J

1 REPRESENTING CONSTITUENT STRUCTURE

Bracketing Practice frame on bracketing Tree diagrams Criterion frames on tree diagrams

I REPRESENTING STRUCTURAL AMBIGUITY Structural ambiguity Criterion frame on ambiguity

Concluding frames

Fig. 2. Structure

and content

of the program

IMCON

Thus. there are usually at least three routes out of each frame. the student being routed along the appropriate one according to the response given. Students giving incorrect or unrecognizable responses are often given extra attempts at the problem and/or provided with remedial frames. Students giving the correct response are congratulated and sometimes provided with a summary frame. A typical set of routines is shown schematically in Fig. 3.

THE

CONTROL

AND

OPERATION

OF

THE

PROGRAM

The IMCON program uses the authoring system STAF[ I]. This system includes a set of files and programs which enable the author to control and record the progress of the student through the CA1 program. The STAF language is relatively easy to use and so is ideal for teachers with little programming experience. The system was developed as part of the CALCHEM project in the National Development Programme in Computer Assisted Learning and was first used on the IBM 370 at Sheffield City Polytechnic in 1974[2]. It now runs on a variety of other computers. including microcomputers such as Cromenco and Research Machines 3802. Any programming language used for CA1 must enable the author to direct the student from one unit of the program to another. In the language STAF. each “node” (this being the smallest unit of a program) has a node “ident$fier” so that students can be routed to it and a “routiflg insrructiorl” to route them to other nodes. Most nodes also contain a “resporlsr rnntching md rouring” device. This compares the student’s response with one anticipated by the author and if it matches then the student

28

STEWART

MARSHALL

. n=2 I _

Remedial frames

Try again

Unrecoqnizable _!_ Question

Response matching

’ Incorrect

I Remedial frames

/ Sorry

Correct Fig. 3. A typical set of routings

is routed to a particular part of the program. A “mrchiny oprim” rnabics the author to specify hou closely the student’s response must correspond for it to “match”. Thus. the author can allow spelling mistakes. extra words or words in the wrong order if he or she wishes. Each node can also include one or more of the numerous “.strhrortriries” available in STAF. For example. the author can set a counter to record the number of correct responses made by the student. Each node would then contam a subroutine to increase this counter by one should the student make a correct response. This information can then be used to control the route taken by the student at any point in the program. It can also be recorded in the program statistics and printed out at the terminal as feedback for the student. The “progrctrn mAaic.s” are a record of the student’s progress through the program. The author can specify the nature of the statistics collected so as to include any of the following: responses made at selected nodes: the identifier of every node visited by the student; numerical data as collected by subroutines (e.g. the number of correct responses and the time taken); and. the author’s own comments.

The IMCON program has a total of I73 nodes to produce and control the various possible teaching sequences. 37”,, of these nodes are concerned with “author responses” to student answers and 22”,, are specificallv for control routings and subroutines. Only about 7”, are concerned with “initial teaching”. Running

the progrtrm

on rhr tnuinjktnw

(IBM

3 70) cornpurer

To run the program on the mainframe. after logging-on the student simply types “EXEC IMCON”. He or she then receives the message “PLEASE TYPE IN YOUR COURSE”. Any response is accepted here for the student to start on the introductory part of the program. In the first node of the program. a subroutine sets a variable to the present time. Another subroutine routes the student to the next node and also specifies the node to be routed to if the student types “STOP” at any point in the program. The next node produces. on the VDU. the title of and an introduction to the program. The student is then asked to type in his or her name and the response is recorded in the program statistics. The student can then choose where to start on the actual teaching program (see Fig. 4). The student proceeds through the program. being routed according to his or her responses. In some cases. for the answer to be accepted as “correct” it must be identical to the “matching response”. e.g. when the students are asked to represent the constituent structure of “AN ORGANIC COMPUTER FOUND THE ANSWER” using bracketing. At other times extra words and spelling mistakes are allowable. Thus. at one point students are confronted with the practice frame shown in Fig. 5. Responses to this frame are regarded as correct even if the students type something like “I THINK THEY ARE IMMEDIATE CONSTITUENTS” or if they misspell “IMMEDIATE CONSTITUENTS”. At the end of the teaching program. or when the student types “STOP”. he or she is routed to the concluding pair of frames shown in Figs 6 and 7. Note that the penultimate frame asks the student to evaluate the program. The comments typed in at this point are recorded in the program statistics.

CA1 m the skills of linguistic analysis

This program covers the following topics:-

: 3 4 z 7

Analysing a sentence into its constituents. Using 'substitution' as a criterion for constituents. Using 'distribution' as a criterion for constituents. Finding the immediate constituents of a sentence. Bracketing as a means of representing constituent structure. Tree diagrams for representing constituent structures. Structural ambiguity in sentences.

If you have not tackled this program before and would like to start at the beginning, please type the number 1. Otherwise, type the number of the section you would like to start at, (Please type your number now.)

Fig. 4. The options given at the beginning

In the

sentenre

of the program.

-

JOHN CRIED “JOHN” and "CRIED" are called the ..... of the sentence.

(Please type the missing words.)

Fig. 5. A practice frame

You might like to know that you have been working on this program for approximately 45 minutes.

I hope you feel that the time you spent was enjoyable and Perhaps you would let me know? educational To help me evaluate the program, I would like to know your reactions to it - any comments, criticisms, suggestions for improvement, words of praise, etc. would be gratefully received. Write them down fairly soon and let me have them (anonymously if you like) or come and see me please Perhaps you would like to make a few comments now? If you would, type them now (up to 11) lines) and finish by typing 99. If you don't want to cormkant now, simply type 99.

Fig. 6. The penultimate

frame.

29

STEWARTMARSHALL

30

Byet and thank you very

much,

Stewart Marshal 1

Fig. 7. The final frame.

EVALUATION The method qj rculuation

A wide variety of techniques was used to gather information during the three weeks that the students were completing the program. These techniques were: direct observation: computer statistics: questionnaire: free response writing; Informal interviews: recorded group discussion; coursework unit: and comments by colleagues. To some extent the “gathering process” was guided by the selection of “key” issues which emerged during the evaluation and which seemed to require further “illumination”. For example. analysis of the student comments typed in at the computer terminal revealed that the “tree diagram” section was regarded as repetitive by some students. This point was then followed-up in subsequent informal Interviews. In this way issues which were seen as important in the whole “learning milieu” were investigated.

From the evaluation (a) (b) (c) (d) (e) (f)

it would seem that the main problems

experienced

by the students

were:

the section on “bracketing” was tedious: there was insufficient explanation or help given when a response was “not recognized”: there was a need for a “review” or “help” mechanism: the program was repetitive in places; some students did not know how to correct typing mistakes: the program w;1s rather long if a lot of mistakes were made (maximum time taken = 53 min: average = 45 min).

Analysis of the completed questionnaires revealed that 52”,, of the students thought that the program was “not very successful” in achieving objective if) {defining the concept of bracketing). For each of the other objectives more than ?O”,, of the students reported that the program was “fairly” or “kery successful”. The percentage of students claiming to have ‘.a lot of interest” m computers increased from 24”,, before completing the program to 48”,, after. Sixty per cent of the students said that they would like to take a computer option as part of their studies. CONCLUDING

REMARKS

The results of the evaluation suggest many places where the IMCON program could and should be needs altering so as to be less tedious. The number of improved. The section on “bracketing” “matching responses and routings” needs to be increased at some nodes so as to produce less unrecognizable responses and more remedial routes. A solution to the problem of the need for a “help” mechanism can be found in Brian Farrington’s program for teachmg French to undergraduates[3]. By incorporating the word “HINT”’ as one of the possible answers to each question. he W:IS ktble to provldc helpful comments ;!t the terminal thus enablmg the students to m;lke intelligent guesses where appropriate. Clexl> this fxilit> IS also available in STAF By providing Lm appropriate

CA1 in the skills of linguistic analysis

:I

“response matching and routing” at each questlon. It would be possible tor students to ask for a HINT or even a REVIEW before typing an actual answer. The technology problems that students experienced because of the novelty of the situation could be overcome by producing a preparatory program. This could introduce students to the hardware and to the associated operations such as “logging-on”. typing in data. deleting incorrect data. and “loggingoff”. Such a program would remove the need for some of the opening frames of IMCON and would thus reduce the running time of the latter. Overall this experiment in CA1 in linguistics was successful. Perhaps the most exciting success of all was in the area of that almost hidden objective relaiing to student interest and confidence in the use of computers. As a result of this experiment six students on the degree are already working on computer projects (producing programs to help children to write stories) and planning is well underwa! for a third year option on computers. A~kno~ledyenlenrs-M) thanks go 10 Quentin Whitloch. Stuart Trickey. Christopher Dean and Janet Watkins for helping me to write the program. and to all the second year of the B.A. (hens, m Communication Studies for providing me with such valuable feedback. REFERENCES Ayscough P.. S7AF Author Guide. Imperial College Computer Centre ( 198 I ). Peterson J. W. M. and Sessions A. E.. STAF L,‘srr Grids (IBM 370 1_7.(--.\4L’SIC). Council for Educational Technology (1977). Farrington B.. Computer based exercises for language learning a1 university level. COUI~LU.Educ. 6. I I3 I I6 (1981).