What criteria do composers use to choose the key of a song?

Question

I would like to know what criteria composers use when choosing the key of a song. In the case of music composed for orchestra or, in general, for instruments with limited tessitura or tuned in a certain key, I understand that it may be more convenient to choose one key than another. In the piano, however, even though there are no such limitations, great composers often chose more complicated keys than necessary. I think that this would be the case, for example, of the third prelude of the first book of Bach's Well-Tempered Clavier, which is in C# major (7 altered notes), and could have been written in the enharmonic key of Db major (5 flats); or just transposing it only one semitone up, in D major, with only 2 alterations.

So, given that a song, when transposed one semitone up or down, is (IMHO) practically indistinguishable from the original for the vast majority of musicians, what is the reason for choosing one key over another?

Answer 1

Some keys are easier for some instruments. Wind instruments are generally more comfortable to play in flat keys while strings tend to be more comfortable in sharp keys. There's one orchestral piece (I forget what, but someone will doubtless say in the comments) where the composer made a point of picking a key where G, D, A and E are all sharped so that the string players would be unable to use open strings. Apparently it's not uncommon for players performing this to tune a half-step sharp so they can get their open strings back.

As noted in the comments on the question, the Well-Tempered Clavier was written to demonstrate the ability to play in any key which is another reason why so many different keys are present in that collection.

Finally, there's some belief among some composers that different keys have different emotional resonances (e.g., https://ledgernote.com/blog/interesting/musical-key-characteristics-emotions/ ) Personally, I tend to view this to be roughly as valid as horoscopes (i.e., not at all), although it can be fun to encode this sort of information in a piece regardless.

Answer 2

In my experience I would say the high level categories are:

• Who will be performing it, as in the musicians themselves, their ability and any personal preferences.

• What instruments are used, which has already been mentioned. Obviously certain keys are easier on certain instruments.

• The character of the key so that it matches with the character of the piece, like major/minor. This one is fairly obvious.

• The time period the piece was written in. The more "complicated" keys didn't sound the same during the early Baroque as they did in the early 1900s. C# major was a strange exotic key that just didn't sound good on early keyboard instruments. Now it is about as common as any other key.

• Any sort of chromaticism or peculiar tonality that is perhaps easier in one key than another. For a simple example a piece that tends to resolve up using a sharp key. This is usually more of an issue in modern music. Bartok comes to mind.

• To your comment on unnecessarily complicated keys for piano pieces, especially in the specific case of Bach and his Well-Tempered Clavier, I think using the more complicated one might have been the point. Composers tend to be more daring and experimental in solo pieces for their own instrument. Beethoven and his piano sonatas are another great example.

• Some people just have favorite keys.

• Like @don-hosek said some people believe that keys have resonances or impacts or even meanings. It is a potential criteria even though I personally have never experienced this phenomenon.

Answer 3

For a popular song, the choice is simple. I chose the key that the singer will be most comfortable in. I will transpose up and down to match the singers (and the previous night's activities.)