TXT 4N6 473
potential authors of messages, thus avoiding some of the
population sampling issues which otherwise arise.
B. Consistency and DistinctivenessAll comparative authorship analysis depends upon two
theoretical assumptions.^20 The first assumption is that there is a
sufficient degree of consistency of style within relevant texts by
an individual author. The second assumption is that this
consistency of style inherent in an author’s writings is sufficiently
distinctive to discriminate the one author from other relevant
authors. Ultimately, the idea that comparative authorship analysis
rests upon a strong theoretical assertion of an idiolect is false.
The empirical discovery of consistency and distinctiveness can,
however, be a sufficient foundation for such work.^21
The first assumption, that there is “a sufficient degree of
consistency of style within relevant texts,” requires further
discussion. It is not necessary to identify features of an author’s
language that are wholly consistent. As shall be seen in the
Birks case, a weight of evidence for authorship may be built
upon a degree of consistency. It must be recognized, however,
that the greater the degree of consistency in any comparison
corpus, the greater the weight of evidence there will be for an
attribution. Identifying consistency within relevant texts also
requires the creation of a linguistically relevant comparison
corpus, which accounts for genre^22 as well as other sources of
linguistic variation. For example, it must take account of
accommodation effects between different recipients of messages
and between the possible modes of production—whether the text
message was created using a twelve-key alphanumeric system
(as is found on more old-fashioned phones), a touch sensitive
qwerty keyboard such as is found on an iPhone, or even through
a speech-to-text system.
The second assumption raises different considerations. There
may be degrees of distinctiveness between pairs of individuals or
(^20) Grant, TXT 4N6, supra note 3, at 509.
(^21) Id. at 521–22.
(^22) To avoid comparison, for example, of text messages with emails.