Gripla - 2020, Page 36

35
that C-divergent (in light gray) is again more similar to both parallel docu-
ments than A-divergent is. A-divergent (in dark gray) is slightly more simi-
lar to A-parallel (1.500) than it is to C-parallel (1.506) and C-divergent is
slightly more similar to C-parallel (1.334) than it is to A-parallel (1.355), but
the fact that these values are so close indicates to us that we have removed
most of the interfering noise. Meanwhile, C-divergent is once again more
similar to A-parallel and C-parallel than A-divergent is, indicating that
C-divergent is more stylometrically similar to both parallel documents
than A-divergent is. Having eliminated most of the words which could
have contributed to statistical noise or otherwise were not stylistic, we get
what is probably our clearest result yet. C-divergent is again closer to the
other documents than A-divergent is, supporting Erichsen’s argument for
the internal consistency of the C-redaction.
The fact that the overall trend remains consistent makes us doubtful
that further manipulation of the text or of the test environment would
affect the result significantly. We are therefore confident enough in the
results to argue that stylometry firmly supports the internal consistency
of the C-redaction over the A-redaction.
The results support Erichsen’s understanding of A-redaction: it is a
result of a mix of sources. It is important to note that she remains ambiva-
lent regarding whether the “secondary source” of A-divergent is written
or oral. She also does not explain why this alternative source was sought.
It could have been due to an exemplar, personal choice, or something else.
All told, this explanation is cautious but sufficiently captures the various
possibilities. Furthermore, this explanation best fits our results: stylometry
agrees that the A-redaction is indeed the result of a mix of sources when
compared with the C-redaction. On the other hand, we can safely dismiss,
as Andersson did, Magerøy’s attempt to demonstrate a stronger stylistic
affinity between A-divergent with the parallel chapters than C-divergent.
Nevertheless, we cannot necessarily dismiss Magerøy’s attempt to dem-
onstrate a textual connection between A- and C-divergent, as this was not
tested. Overall, it is our impression that A- and C-divergent’s textual rela-
tionship is minor, such that Erichsen’s argument has much greater explana-
tory power than Magerøy’s argument that the two are textually related.67
A STYLOMETRIC ANALYSIS OF LJóSVETNINGA SAGA
67 Our impression is based on the fact that there is not even a single clause which is exactly
the same in A and C in these chapters.