148 andrew w. pitts
(1 cor 16:21), φιλεῖ (1 cor 16:22), προσφιλῆ (Phil 4:8), φιλαδελφίας (1 thess
4:9), φιλοτιμεῖσθαι (1 thess 4:11) and φιλήματι (1 thess 5:26). So we would
expect the Pastorals, dominated by paraenesis, to have a higher density
relative to their size than the other letters if the φιλο-word group is a
characteristic of paraenesis. considering the entire letter length without
consideration of the distribution and restriction of the φιλο-word group in
many of the undisputed letters in relation to the paraenetic sections thus
skews the evidence. this feature seems more likely due to register design,
where Paul accommodates his language for the paraenetic form of his dis-
course, likely motivated by his personal relationship to his recipients.
if we set aside the severe methodological problems associated with
the widely referred to hapax legomena criterion for Pauline style, we
can perform a similar analysis here as well.90 in the letters where they
occur,91 their frequency is not more than 6.2 words per page (Philippians)
and as few as 3.3 per page (2 thessalonians). a calculation of the average
rate of hapax legomena in our register profiles 2–4 yields an average occur-
rence of 4.09, on harrison’s figures. no one can deny the spike in variation
with the shift in register in the Pastorals. harrison calculates two figures
here, words shared and not shared by the Pastorals with each other. the
words not shared with one another generate a much less drastic shift with
an average rate of 12.25. here, we get the same level of variation at 66%.
once we enter words shared within the Pastorals we get a more drastic
variation rate of 73%. But, when we calibrate these figures together, this
increased variation level makes better sense on register not author varia-
tion, because it shows that the Pastorals diverge together, which is exactly
what we would expect for writings sharing a register profile. it seems at
least more plausible than posting as many as two (if 2 timothy was com-
posed by a different author than 1 timothy and titus) additional authors.
if simplicity is taken as an explanatory virtue, certainly we should not
multiply explanations (or authors) beyond necessity.
related to the hapax legomena question, we have what is often referred
to as the semantic richness of the Pastorals, when compared to the eccle-
sial letters. harrison leans heavily on the fact that 36% of Paul’s vocabu-
90 harrison, Problem, 21.
91 i do not (and neither does harrison) calculate into the equation instances with zero-
occurrence to track variation since this would involve calculating a shift against a zero-
variable. this highlights the problem of corpus size and the use of vocabulary in studies of
authorship attribution discussed in Part 1 and the methodological problems, highlighted
by many, with using hapax legomena as a criterion for style.