In 1995 we started recording, just after the earthquake in Kobe. We had looked at the possibility of recording with a Japanese company, but they had no understanding of the cantatas. When I said it would take 20 years to complete the cycle, they were horrified.

We’d had some contact with the Swedish company BIS, whose founder Robert von Bahr visited us in Tokyo. We talked of our dream to record all the Bach cantatas. He has a passion for Japanese culture, which probably helped, but initially he was a little sceptical. We agreed to do one recording over three or four days to see how it worked out. During that recording session he became very excited. At one point he rushed out of the control room, hugged me, and said “This is good”. There were also quarrels — he’s a very strong man, but I have my own personality, and the members of the Bach Collegium Japan also have their views. The quarrels were really useful because we got to know each other very well. Two more recordings followed within that year. In the end he stopped producing our recordings himself, instead sending excellent producers from his office. Robert is very good at managing the company and has encouraged us a great deal. The project is ongoing, for which we are very thankful.

How often do the Bach Collegium Japan come to Europe?

We have made eight international tours. In 2003 we travelled to Italy, Germany, Spain and Israel, but the cost of travel means it is hard for us to make many international tours. We will be back in the summer of 2005, playing at Ansbach in Germany and the Schleswig-Holstein Festival, and again in 2006.

Some people say that Bach cantatas should be performed with only one voice to a part, but others disagree. What’s your feeling?

That’s a famous argument. The solo approach is right for many of the Weimar cantatas and some of the Leipzig ones, but there are some musicologists in Leipzig who are convinced that it is generally wrong. My feeling is that the quality of the performance depends too much on the quality of the soloists when you perform with one voice to a part.

There’s an interesting situation with the chorale cantatas, where a chorale is sung at the beginning and the end, with recitatives and arias in between. The outer movements need to be relatively neutral, so they announce the idea that has more private expression in the recitatives and arias. This works well if there are a few ripienists on each part who can join with the solo voices for the chorales. It is really important to think in terms of concertists who sing throughout, with one or two ripienists joining in where appropriate, rather than to have people who sing only the solo parts.

In modern times there is great pressure on musicians to perform the Bach cantatas to the highest international standards. In Bach’s Leipzig, musicians were simply performing these as part of the church service. What difference does this make?

The liturgical context is the most natural way to understand the cantatas, which were all written to be part of the liturgy and to have a close relationship with the psalm and the hymns sung by the congregation.

This doesn’t work in services today, even in the Lutheran churches, because they rarely last more than an hour yet most of the cantatas take 20–25 minutes. In Bach’s time worship took up to four hours, so the cantata was a much smaller, and more appropriate, proportion. There is also a problem with the words, which are too old-fashioned to be readily understood today. It is easier to appreciate the cantatas now in the concert hall — though we should always be aware of the context for which they were written.

There is an added difficulty if you experiment with cantatas in worship only occasionally, because worship needs to be familiar. It doesn’t work if you have only the occasional “music worship” or “worship with Bach”.