The focus here is on developing learners' ability to segment the stream of sound quickly and efficiently in order to identify words, phrases and grammatical information, which can then be interpreted through top-down processes.

Background listening
The speed of word recognition is partly dependent on simple familiarity with target language phonotactics, ie the typical permissible sound combinations of the language. Phonotactic knowledge develops implicitly through repeated exposure to the language. It is this that facilitates word recognition even before meaning can be mapped on to a particular phonemic string.

Thus, there is reason to encourage learners to 'just listen' to as much of the target language as they possibly can, even if they don't understand it. Having a cassette playing in the car, in the background while doing other routine tasks or waiting, can provide useful ear training. In particular, it is helpful to play and replay familiar audio recordings to establish fluent recognition of words and phrases.

Word recognition tasks

• Giving students a list of key vocabulary before listening helps prime their bottom-up processing, but you may need to ensure that learners in fact pay attention to that vocabulary in some way, rather than merely note translation equivalents or dictionary meaning. As noted above, a useful exercise which links to top-down processing is to ask learners to guess from a list which items they think will appear in a passage, given some idea of its content. They are then motivated to scan for those items as they listen.
• Alternatively, give students a list of jumbled words from a passage, combining words they are likely to know and new words. They hear a passage once or twice and identify the order in which the words appear. The same technique can be applied to phrases. The words can then be used in completing, for example, a summary of the passage heard.
• To work on phonemic processing, make a list of ten fairly salient words which occur in a spoken passage. For each word, find another word which sounds like it, eg in English: 'drown/drawn', 'ship/sheep' or in French: 'cheveux/chevaux', 'rue/roue', 'blond/blanc'. Present the list of pairs in the order in which the original words are spoken. Get learners to first review the list of pairs and check that they understand all the words. Then play the passage and ask learners to identify which words they heard. This kind of exercise can be done with fairly short recordings.
• Scanning: learners listen to a recording for specific items of information, eg times, dates, names, addresses, or vocabulary. Vocabulary searches can be cued by definitions (eg find a word beginning with 'con...' referring to material used in building) or translations and synonyms (eg find words which mean...). You can also ask learners to note down how many times a particular word is used, and then ask them to note down the context.
• Who said what? In dialogues, particularly video recordings, give students a jumbled list of key phrases from different speakers. Ask learners to identify the speaker. You might then develop this exercise into an interpretation exercise, asking learners to explain why or how the particular person said the particular phrase (eg click here) (Télé-textes Unit 4C exercise 4).

Gap-filling and correction exercises

• Learners receive a partial transcription of a passage and complete the gaps (you need to take into consideration the time it takes to complete a gap - multiple choice answers can be useful here).
• Learners receive a transcription of a passage with a number of words or items of information altered. They have to correct the transcription (eg click here) (Télé-textes Unit 4C exercise 5).
• A more 'real-life' task is to ask learners to correct erroneous notes, made by one of the participants in a dialogue (eg click here) (En route vers l'Europe, Unit 1 exercise 3)

Dictation and transcription
There is increasing recognition of the value of dictation in developing and testing bottom up processing skills. The task of representing aural input in written form requires detailed processing of the sound stream, but this is a task which shows clear improvement through regular practice and which learners can easily undertake on their own.

It may not train learners in the kind of guess-work skills they need for coping with real-life communication, but it should reinforce the automatic recognition of second language words and structures, and therefore speed up bottom-up processes.

The traditional form of dictation often used rather stilted, specially written passages. Dictation does not have to be like this.

• Use short, funny passages for example, such as those amusing tales often published in local or tabloid newspapers.
• Use authentic news items.
• Use texts which might indeed be dictated like letters or messages or advertising copy. ·
• Use texts from your coursebook which students have already worked on - in this sense, dictation serves as an excellent revision exercise.

Dictation also does not have to be conducted by the teacher in the conventional way. It can be shared around the class. This adds an extra dimension: those dictating have motivation to take care over their pronunciation while those writing may have to be inventive in interpreting faulty pronunciation.

Running dictations are dynamic and interactive listening-writing exercises, which can stimulate a passive class. The principle here is simple: the text to be dictated is stuck on the wall, learners work in teams, in turn a member from each team has to run up to the text, remember a sentence and return to the team to dictate it accurately. The aim of each team is to finish the complete text first.

Dictogloss: Nunan (1991) recommends a variant on dictation known as dictogloss or grammar dictation, which was developed by Wajnryb (1990). In a dictogloss exercise, learners are encouraged to establish the gist of a short spoken passage after a first listening; after two subsequent listenings, they work in groups to try to reconstruct the passage word for word.