Mahler’s three volumes of Lieder und Gesänge (1880-1889) comprise the only large collection of songs that the composer did not orchestrate himself: from 1892 onwards, the large majority of his songs were written in at least two performing versions, including for voice-piano and voice-orchestra; and while he returned to orchestrate the earlier Lieder eines fahrenden Gesellen, only a single sketch fragment evidences any intention to similarly return to the Lieder und Gesänge. Given the considerable popularity of Mahler’s orchestral songs since the mid twentieth century, the ‘metamorphoses’ of certain Lieder und Gesänge within Mahler’s symphonic oeuvre, and the sharp rise of song arrangements on the concert stage in the past several decades, it is unsurprising that several later composers and arrangers have produced orchestral versions of the Lieder und Gesänge. This paper explores the reception and arrangement history of the songs through a comparative analysis of four sets of orchestrations, focusing in particular on potential models used by arrangers for their versions, and their varying interactions with ideas of Mahler’s ‘early’ song-writing style.

I turn first to versions of the songs by Colin Matthews and David Matthews (1964/2016) and by Detlev Glanert (2014–15), which I suggest belong to a phenomenon of ‘historically informed arrangement’ – both seem to aim, in different ways, to realise how Mahler might have orchestrated the songs himself. I then consider these versions alongside those by Luciano Berio (1986-7) and Eberhard Kloke (2011), both of which I suggest also belong under the ‘historically informed’ umbrella but which use their ‘historical information’ as means for subversion and play. I also introduce texture graphs of the orchestrations, designed to comparatively visualise different approaches and detect emerging patterns, and incorporate an evaluation of some of the prospects and limitations of this analytical method.