Testing

Writing and
Speaking
 What are we trying to MEASURE?

The student’s ability to interact

in the target language
in written or spoken form.
This involves
comprehension as well as
production.
(Hughes,
1991)
 Why is it so difficult to test productive
skills?
 Difficult to elicit behavior in early
stages
 Tasks should elicit a representative
sample of the student’s ability
 Formal formats tend to elicit unreal
behavior
 Teachers tend to score by impressions
 Scoring criteria is often inconsistent
among testers
 We can mend these problems by…

 Setting appropriate tasks

 Testing only writing or speaking
abilities, and nothing else
 Restricting the student’s response
 Selecting appropriate test formats
 Obtaining reliable scores
 Setting appropriate tasks means…

Considering questions like these:

 What is the communicative function

I want to elicit from the student?

(expressing… giving… describing…)
 What is the form of interaction

expected? (letter, postcard, email,

voice chat, interview, oral presentation)
 What is the context for the

interaction? (formal, informal, real,

unreal, school context, the street,
home, abroad…)
 Obtaining a representative sample
means…
 Setting a flexible context
 Considering a variety of tasks
 Asking for authentic and meaningful
answers
 Setting the criteria for production (1
paragraph, the following aspects…)
 Giving the student enough time to
complete the task.
 Restricting the student’s response
means…
 Defining tasks in detail (telling exactly
what you expect from the students)
 Establishing boundaries (controlling
ideas that will stop students from going
too far astray)
 Specifying the minimum requirements
for production
 Provide information or cues in the
form of notes, examples or pictures
 Testing the skill only means…

Avoiding the assessment of…

 Opinions

 IQ

 Creativity or imagination

 Previous/general knowledge

 Other skills (reading, listening…)

E.g. Read and explain this quote:“Bigotry

is the sin which most harms the sinner”
 Test formats for Speaking
 Oral questionnaire
 Free interviews
 Peer-interaction/role-plays
 Picture-cued tasks (Describing a
picture, structuring a story, giving
directions, giving instructions…)
 Response to recordings
 Paraphrasing a story, a situation…
 Prepared monologue *
 Reading aloud *
 Test formats for Writing
 Essay questions
 Guided writing (a letter, postcard,
email, note, form, advertisement…)
 Picture-cued tasks (Describing a
picture, structuring a story, giving
directions, giving instructions…)
 Response to written/spoken stimuli
 Paraphrasing a story, a situation…
 Prepared written production *
 Dictation*
 Obtaining reliable scores means…

Having a well-structured criteria based

on bands and descriptors
 5 Excellent writer
 4 Good writer
 3 Average writer
 2 Substandard writer
 1 Poor writer
 However, one question arises…

What makes a student an 'excellent'

writer?

 Is it comprehensibility?
 Is it grammatical accuracy?
 Is it spelling?
 Is it the way they organize the text?
 One solution is…
Either produce overall descriptions of
writing and spoken ability…
Or concentrate on different aspects of
writing and speaking and then separate
these aspects out into individual scales.
 Holistic scales: those containing
different features of ability
 Analytic scales: Those which
separate out different aspects of
ability
 Holistic Scales
5 Constructs grammatically correct sentences and phrases and shows
full mastery of appropriate vocabulary. Text is organized coherently.
Correct spelling at all times. Excellent content and presentation.
Message wholly relevant
4 In general, grammatically correct sentences and phrases but some
errors which do not affect understanding. Does not use. appropriate
vocabulary at all times. Some difficulties with organization of text.
Some errors in spelling. Good content and presentation. Message
mostly relevant.
3 Grammatical errors and use of vocabulary affects understanding as
does organization of text. Many errors in spelling. Satisfactory
content and presentation. Message not always relevant.
2 Text understood with difficulty due to inaccurate grammar and
inappropriate use of vocabulary. Poor content and presentation.
Message generally lacks relevancy. .
1 Impossible to understand text due to frequency of grammatical errors
and incoherence of organization. Poor spelling Message irrelevant.
 Why using Holistic Scales?
 They are relatively easy to construct
 Scoring takes less time than with the
analytic method
 The integration of sub-skills tends
concentrates on the overall effect of the
piece of writing
 Each work can be assessed more than once
 More bands can be added depending on the
context and the purpose of the tester
 Analytic Scales
Grammar
6. Few (if any) noticeable error of grammar or word order.
5. Some errors of grammar and word order which do not
interfere with comprehension.
4. Errors of grammar or word order fairly frequent; occasional
re-reading necessary for full comprehension.
3. Errors of grammar or word order frequent; efforts of
interpretation sometimes required on reader’s part
2. Errors of grammar and word order very frequent; reader
often has to rely on own interpretation.
1. Errors of grammar or word order so severe as to make
comprehension virtually impossible.

Score: Gramm Voc Mech_ Form TOTAL_

 Why using Analytic Scales?
 They consider each student’s particular
differences in terms of sub-skills
 Testers need to consider the different
aspects of performance individually
 The sole fact that the scorer needs to give a
number of individual marks makes the
scoring more reliable
 Each aspect can have a different weight
depending on the purpose of the tester
 Some sub-skills might not fit together in a
single band
 Holistic or Analytic Scales?

 What is the purpose of your test?

 Are you looking for diagnostic information?
 Do you want to assess overall performance?
 Is it for a small and homogeneous group?
 Is it for a large heterogeneous group?
 Will it be carried by yourself?
 Will it be applied in several different places?