I don't think Roud says anything to deny the creation of songs from within the community itself. I agree he could perhaps have placed more emphasis on this, but I think he is trying to challenge the assumption that most folk songs originated this way, whereas he is trying to show how popular songs from a variety of sources entered the repertoire of 'folk singers'.

One of the problems with definitions is that they are too definite. Perhaps when talking about folk song we should be looking for guidelines, rather than definitions. I am happy to agree with Jim that many of the popular songs from music hall etc should not be considered 'folk songs' even though they are widely found in the oral tradition and in the repertoire of 'folk singers'. They have not undergone sufficient change through the 'folk process' and can be clearly differentiated not only in style but also in emotional content. But what word should we use to describe those songs which can be traced back to a composed origin but which have been transformed in the hands of folk singers? I am happy to regard those as 'folk songs' and I think the 1954 definition supports this view. I'm not sure Jim agrees, although I will happily concede that I'm not sure I'm understanding him correctly and I'm not trying to put words in his mouth.

Coming back to Roud, the waters are possibly muddied by his "Folk Song Index". This label is a convenient shorthand but is I think misleading. It records and indexes songs collected from the oral tradition, and includes many popular songs as well as true folk songs.    I don't think it should be interpreted as claiming that anything collected from a folk singer is therefore folk song, and I don't believe that is Roud's intention.