style and pseudonymity in pauline scholarship 149
lary in the Pastorals is not found in his other letters.92 collins advocates
a more conservative figure, insisting that “almost 20% of the general
vocabulary of the Pastorals is not used by Paul or any other new testa-
ment author.”93 to widen the parameters, we may also turn to o’donnell’s
analysis of Pauline vocabulary for similar results. he gauges types (distinc-
tive vocabulary) in the Hauptbriefe, a seven letter Pauline canon, and a ten
letter Pauline canon and in no instance do we have more than 37% varia-
tion (1 timothy against the Hauptbriefe). But the average variation of the
Pastorals against other letters widely acknowledged to be Pauline is 35.0
(against the Hauptbriefe), 32.0 (against the seven letter canon) and 30.0
(against the ten letter canon). interestingly, the variation levels decrease
with the increase in corpus size, highlighting quantitatively the need for
larger corpus sizes in assessing issues of authorship. if we were able to
continue to increase the quantifiable volume of the Pauline corpus (add-
ing more writings for comparison), would the cline continue downward
in the direction of lower percentages of variation between the Pauline
corpora until the statistical divergences no longer seemed significant?
We will never know. But this data is at least suggestive of the appropri-
ateness of the observation from corpus, statistical, and computational
linguists concerning the need for large corpus sizes for studies in author-
ship attribution—if these studies are even possible at all. again, assuming
with the advocates of pseudonymity, for at least the Pastorals, the validity
of their basic approach to gauging style, these shifts in vocabulary are actu-
ally a bit less than what we find across several other languages in response
to register variation for various other features (including vocabulary).
this indicates that the Pastorals may potentially exhibit a higher level of
linguistic stability across changing registers when measured according to
the criterion of vocabulary than we discover in most single author inves-
tigations of style-shift, at least for the wide range of lexicogrammatical
features that those studies consider.
Register Designed Grammar
harrison and his followers also highlight “grammatical” divergence
between the ecclesial letters and the Pastorals. he notes the absence
of many small particles in the Pastorals that are present in the ecclesial
92 harrison, Problem, 21.
93 raymond f. collins, Letters That Paul Did not Write: The Epistle to the Hebrews and
the Pauline Pseudepigrapha (good news Studies 28; Wilmington, de: michael glazier,
1988), 95.