There were between 10 and 15 participants, all Siemens employees and all adults between the ages of 23 and 60. They met one evening each week, on Tuesday, from 5 to 7.30 pm at the Siemens Training Centre in Perlachsüd. The room and resources were in many ways ideal for a 'student-centred' approach. The participants sat on comfortable office chairs at a kind of 'round table'. There was a video recorder in the room as well as a flip chart and well-kept 'white board'. The participants had all had previous English-training, reportedly of the 'student-centred, activity-based' type. Nevertheless, the level of English was extremely low. Very basic vocabulary and structure caused problems, such as the difference between 'He is unhappy' and 'He is unlucky' or 'What do you do?' and 'What are you doing?'. Nobody seemed to know the past forms of the simplest and most frequent verbs in English. When participants were asked or saw question forms like 'What did you do yesterday?' they were very puzzled. 'Why you don't say what did you?'' they asked. Attempts, however brief, to help them understand the 'rule' (which they repeatedly asked for) were unsuccessful Very often, when they spoke English, it was necessary to understand German in order to interpret what they were really trying to say.

However, despite these 'problems', I greatly enjoyed teaching this group, and they seemed also to enjoy the lesson as long as I used a 'teacher-fronted' style most of the time with regular opportunities for brief, highly-structured pair-work.

One aspect of the dogma - or rather, a delusion that supports it ? is the belief that learners can get the kind of input they need from each other. Some teachers and teacher-trainers even appear to believe that this can and will happen regardless of the level of learner.

I do not deny that students talking to each other in the target language can be a valuable source of input, especially if at least one of two conditions are met ? and preferably if both are met.

1. One of the learners knows more of the target language than the other or others and is more skilled at using it.
2. The two or more learners in the group or pair do not speak the same L1.

Personally, if I decided to study a foreign language I knew little or nothing about, and went to a school in the country in which it is spoken in order to study it, I would not be very satisfied if the teacher - presumably a competent and far, far better speaker of that language never or hardly ever spoke to me or to the class as a whole. Of course, I would hope that the same teacher would give the people sitting around me as well as me a variety of opportunities to speak to the teacher and to the other learners. This, however, is not an argument against teacher-talk. It is an argument for teacher-talk and learner-talk. It is an argument for the kind of teacher-talk that does not exclude learner-talk but helps to promote it.

This is not an argument against all pair- or group work. In Munich I used a great deal of pair-work exercises. They usually began as highly-structured conversations, which students read aloud to each other. I often read the dialogues aloud to them myself - an example of something that is currently deplored by many teacher-trainers and which I will return to later. I read aloud to the class for several reasons. First, I discovered that most of the learners in the group simply could not accurately pronounce even very simple words. By 'accurately', I do not mean 'like a native-speaker'. I mean, 'clearly enough to be understood by native and non-native speakers alike'. The reason many people in the group could not do this was not because the words involved difficult phonemic contrasts for Germans. German has vowel sounds that are the same or almost the same as the vowel sounds in