Assessing productive skills (writing and speaking)

Writing and speaking skills are assessed by subjectively scored integrative task types, which means that scoring is a matter of judgement, that is why it may vary from person to person.

13.11.1 The most common task types for testing written performance.

According to the extent of controlling test takers’ freedom of expression in their written work three kinds of writing tasks can be distinguished: controlled, guided and free or creative writing tasks. The more controlled the tasks are the easier it is to ensure reliability, but they become less valid measures of writing ability.

Controlled writing tasks impose complete control over both language and content. They cannot be considered real writing tasks in the sense, that the writers do not produce any/or very limited writing of their own. Controlled writing tasks are: multiple-choice sentence completion, transformation, (these are tests of usage testing grammar at sentence level), cloze/modified cloze, translation, editing/error correction/proof-reading.

Guided writing tasks give a full task environment by specifying several features such as

- the content/topic/information to be included in the written text (what), - the role of the reader, the relationship of reader to writer (to whom)

- the role of the writer him/herself or in an assumed role (who)

- the reason for writing, the effect the writer wants too achieve (explanation, request etc)(why)

- text type, style (informal/formal), genre (how)

- details of date to be used/date of background events (when) - details of place, background, context, situation (where)

In this way the test taker can get enough information so that s/he can treat the task as a social writing activity, as a real-life activity. Guided writing tasks can be considered authentic tasks as by giving a full task environment they simulate a real-world writing task. This is the way how we write in the real world. The most common writing tasks are: form filling, writing a summary of texts (it is a typical academic writing task), writing from given context or a fully defined situation, responding to an input text (e.g. incoming letter), information transfer from non-verbal input (e.g. describe pictures, diagrams, maps, flowcharts), agreeing or disagreeing with a given statement.

Free composition writing imposes little linguistic control on the writer. In free writing tasks there is minimum control, only the composition titles are given and the test takers are free to express their ideas in their own words.

13.11.2. Scoring productive writing tests

Earlier written pieces of work were marked based on the tester’s impression of the language output, which is a fully subjective way, consequently very unreliable.

In modern communicative testing we can speak of improved subjective scoring, which means that scoring is made reliable by detailed marking schemes, in which language performance is assessed based on certain categories and subcategories to which marks are assigned. Usually there are three main categories of language performance to be scored: content (the required information included, irrelevant info excluded, content complete, message appropriate), language (accuracy, proportion of errors, vocabulary size, good range of language, accurate spelling and punctuation), and text (style, register and format, textual organisation and cohesion, suitable layout)

Another way of assessing written language performance is the use of yardsticks or scales, which means a set of bands which cover the whole range of language performance from non-user to expert user. Each band has a descriptor of the expected performance in the given band. These descriptors are phrased in

real-13.11.3. Assessing speaking skills

Testing speaking ability is an important aspect of language testing, but it is very difficult to test, as it is a complex skill, moreover speaking skills cannot be separated from the listening skills in many tests of oral production. This interdependence of the speaking and listening skills makes it difficult to analyse precisely and reliably what is tested at any one time. Moreover, spoken language is transient, so without recording the production it is impossible to apply such procedures as in scoring composition, where there is enough time to consult the marking scales. During an oral exam there is a time pressure, examiners are required to make subjective judgement as quickly as possible. Another difficulty in oral testing lies in its administration. Testing large numbers of students employing a limited number of examiners presents large problems and it is a time consuming process.

13.11.4. The most common task types

There is a great variety of tasks that are employed to assess oral production. Just like writing tasks, speaking tasks can also be categorised according to how much potential is given for real communicative interaction, how much the content and the language is controlled. These tasks can be found at different language proficiency examinations

Linguistic (controlled) speaking tasks focus on certain aspects of speaking and the spoken language (e.g. pronunciation, knowledge of appropriate conversational exchanges, clichés,) and are far from authentic interaction. However, they can be found in a few language examinations. Controlled tasks are: reading aloud, response to a recorded grammatical cue or a situational cue, interpretation of what the examiners says, simultaneous translation of text in L1.

Guided speaking tasks give details of the context or situation. Typical tasks are:

taking and relaying message on the tape, summarising a conversation or a spoken text recorded on the tape, problem solving, interpreting or summarising a written text, role-play, discussion based on an object or a picture belonging to the test taker, discussion based on a book read by the test taker.

Free speaking tasks impose minimum linguistic control on the speaker. Widely used tasks are: prepared presentation on a topic chosen by the candidate in advance, monologue on a given topic chosen at the exam, interview (e.g.

discussion of personal information about the candidate, or discussion of a topic or of the candidate’s opinion about the topic), description of a picture chosen.

13.11.5. Scoring speaking tests

The scoring presents the same problem as in the case of writing tasks. There are a lot of elements of subjectivity of assessment, which reduces the reliability of scoring. However, several techniques are available, which can improve reliability, especially if they are employed together: the most effective ones are: the use of marking scales, multiple marking (i.e. the test taker’s performance is marked by two examiners independently), assessor training. Marking scales are usually based on several criteria for the assessment of oral production (e.g. accuracy, appropriacy, range of vocabulary, flexibility of communication, fluency, pronunciation,) to which scores are allocated. Another alternative is to provide yardsticks, which refer to the whole range of language performance achieved in the whole test, describing the required level of ability in each band.