Just as Webern had turned to song composition, one of his earliest and most familiar occupations, to launch his atonal career, so he took it up again at this second time of crisis, finding in composing around a text a temporary answer to the difficulty of creating extended atonal forms. During and after his struggle with the cello sonata he began to write songs once more: in 1913–15 he sketched several on texts from various sources, including some he had written himself. At first these were in the aphoristic style of the op.8 songs, though because of the length of the text they were somewhat longer. But soon the style began to change: lines became longer and individual parts more continuous, and the result was a linear polyphony which was more prophetic than retrospective. During this 11-year period, all of Webern's finished works were inspired by lyric texts; this has prompted Anne Shreffler to write of Webern's ‘lyric impulse’.
The songs written in 1913 and 1914, for voice and orchestra, were published only in 1968; the first songs from this period to be published by Webern are for voice and piano: the Vier Lieder op.12, written in 1915 and 1917. Their texts are from widely divergent sources – the first is a Rosegger text posing as a folksong, the others poems by Li Tai Po (from Hans Bethge's collection Die chinesische Flöte, from which Mahler had taken texts for Das Lied von der Erde), August Strindberg (from his Spöksonaten) and Goethe – and are, as might be expected, heterogeneous, ranging from the simple piety of the ersatz folksong to the sensuous pleasures of the poem from the Chinese. Webern follows the structure of the poetry: the piano supplies short preludes and postludes of two or three bars and interludes of a similar length between verses or lines of the poetry.
In the years 1917–22 Webern composed or sketched a large number of songs, including those of opp.13–15. All of these are for voice and some combination of instruments. The sources are again diverse in the op.13 set, which uses texts by Karl Kraus, two more poets from the Bethge collection, and Georg Trakl – a significant choice. Webern was much engaged by Trakl texts during these years, sketching fragments of several Trakl settings between 1915 and 1921, and his op.14, finished in 1921, was Sechs Lieder nach Gedichten von Georg Trakl. These songs, boasting a myriad of instrumental effects and instructions for articulation, are wide-ranging registrally – the singer is required to negotiate a range of over two octaves routinely, and the instruments are treated in a similar way – and rarely rise above pianissimo. Rhythmic complexities and sometimes rapidly changing metres combine to produce an ametric effect. The texture is linear but very dense. Webern described the Trakl songs as ‘pretty well the most difficult there are in this field’, and it would be hard to disagree.
The next set of songs, Fünf geistliche Lieder op.15 (1921–2), used old German religious texts, at least one of which was taken from Des Knaben Wunderhorn. Webern's composition of these five songs, which are generally speaking in the same style as those of the previous set, holds special significance for his future work. In the first six pages of sketches for the fourth song, ‘Mein Weg geht jetzt vorüber’, he wrote out and used a 12-note row; this was the first occasion from which 12-note sketches survive. (The 12-note version was subsequently abandoned, and the method does not make another appearance until the songs of op.17.) In the fifth song Webern turns once again to canon, another of the preoccupations of his later life. The final song, ‘Fahr hin, o Seel’’ (interestingly, the first of the set to have been written, in 1917), is a double canon in contrary motion. The songs that followed immediately, another set on old religious texts, this time from the Latin breviary, were all strict canons. These were finished in 1924.
Schoenberg's famous public annunciation of his new compositional technique occurred in February 1923; Webern got to know of it at the latest in summer 1922, and possibly much earlier. Schoenberg's reminiscences at the end of his life were contradictory on this matter: on one occasion he claimed to have carefully guarded his secret from Webern until 1923, while on another he said that he was experimenting with 12-note themes in 1914 and corresponded enthusiastically with Webern on this subject at that time. Schoenberg's first works to be based on 12-note rows, the piano pieces opp.23 and 25 and the Serenade op.24, were written in 1920–23, years in which Webern was occupied with the composition of the opp.14–16 songs. As early as 1911 Webern seems to have been thinking along the lines of 12-note fields (see his admittedly retrospective description, quoted above, of the way he proceeded when composing the first Bagatelle), and though a sketch of the sort he describes does not exist for op.9 no.1, a sketch dated April 1914 (of a setting of Stefan George's ‘Kunfttag III’ that was subsequently discarded) is very like this. It contains a list of the 12 notes in which nine have been crossed off, these being the pitches played by the instruments in the final chord while the voice fills in the remaining three. The next step occurred, as we have seen, in summer 1922.
The eight songs of opp.17–19, composed in 1924–6, represent a concentration on various aspects of the 12-note method of composing, none making use of all the possibilities it offers. The first two songs of op.17 proceed by means of chromatic fields; only in the second are the notes within all the fields ordered in the same way. The third song of this set contains the first linear presentation of a row in Webern's music, though the sketches show that he had experimented with linear rows before he began writing op.17. The first two songs of op.18 again proceed through a succession of rows which are shared by voice and accompaniment. In contrast, the style of op.18 no.3 is completely linear, each of the three participants setting forth a series of complete rows. This song also makes use of the row's inversion and retrograde for the first time. Row technique is more sophisticated in op.19, the first work in which the several movements (in this case only two) are based on the same row, and the first to use transposition, though only at the tritone.