Corpus-Paul

["Sheila E. McGinn, Ph.D." <smcginn AT jcu.edu>]

Here is the review of Neumann as promised.  SEM
--
Sheila E. McGinn, Ph.D.
Associate Professor
Biblical Studies & Early Christianity
John Carroll University
Cleveland, Ohio 44118
smcginn AT jcu.edu
www1.jcu.edu/religion/mcginn/mcginn.htm
ÿWPC6


ûÿ2G
ÿÿBGPZ¦
Courier 10cpiCourier 
10cpiPrestige Elite 12cpi (D)Prestige Elite 12cpi Italic 
(D)ÿÿÿÿÿÿÿÿÿÿÿÿþÿÿHP LaserJet Series 
IIHPLASEII.PRSÂ
xþ6X@ɓ8

ü 

ê€
0s«eX@#|xestige Elite 12cpi (D)Prestige Elite 12cpi 
Italic (D)ûÿ2ÿÿZy
ÇÓ
ÿÿšX
ªHP LaserJet 
Series IIHPLASEII.PRSÂ
xþ6X@ɓ8

ü 

ê€
0cüYX@Ò 
ÈX0
1ÃÍÄÄ     
ÃÍÄÄ°°
Èh
0
1ÃÍÄÄ     
ÃÍÄÄ°° ÒÐ
ЊÐ
ÈÈ
ÐÐ=Ð#|xdÿÿ‰?xxx,

Ùkô
xþ6X@ɓ8Å;X@þþþþþþþÿÿÿÿÿþÿÿþÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿþÿÿØÿÿtTddd,

WaÂ
dpðø@>pQô
@þþþþþþþÿÿÿÿþÿÿþÿÿÿþÿÿþþþÿÿÿÿÿÿÿÿþÿÿTÿÿtTddd,

'Ç
 Â
dpðøûhQQô
Hþþþþþþþÿÿÿÿÿÿÿÿþÿÿÿþÿÿþþþÿÿÿÿÿÿÿÿþÿÿÿÿÿÿacross 
groups,
the STEPDISC program selected the six most statistically 
significanûÿ24Ô
tT

ÔÕ

 *´Ô
tT

ÔÑ
#Â
dpðø@>pQ
Waô
@#
ÑSheila E. McGinn©Moorer, Ph.D.
Ô
tTÈ
ÔCritical Review of Bauckham,¸hh(ÂÃÃThe Authority of 
the Pauline Epistles in the
Ô
tT

ÔLight of Stylostatistical AnalysisÄÄÆ(#hÆ

ÕÐ=>ÐÑ
#Â
dpðø@>pQ
Waô
@#
ÑÁàìÁSheila E. 
McGinn©Moorer, Ph.D.ƒ
ÁànìÁJohn Carroll Universityƒ
Áàþì#Á2 February 1993ƒ


CRITICAL REVIEW:
Ô
tT°
ÔNeumann°`    `       ÂÁ€Ánn, Kenneth J.  ÃÃThe 
Authenticity of the Pauline Epistles in the Light of
Ô
tTx
ÔStylostatistical Analysis.  SBLDSÄÄ 120.  Atlanta, GA:  
Scholars
Press.  1990.Æ(#`       Æ


THESIS:
Stylistic issues have not been adequately explored in the debate concerning
the authenticity of the disputed letters of Paul.  The present scholarly
consensus is that the authentic Pauline corpus certainly includes the seven
letters Rom, 1 & 2 Cor, Gal, 1 Thess, Phil, and Phlm, while it certainly
excludes the Pastoral Epistles (1 & 2 Tim, Tit).  Concerning the three letters
Eph, Col, and 2 Thess, scholars are more divided.  Nor is there consensus
concerning inauthentic portions within the authentic letters, and vice versa. 
Neumann views this as a source of "embarrassment" to NT scholars, and hopes
"that sufficient new evidence will be obtained on Ephesians and the other two
disputed letters that will assist scholars to come to a geniune `clear
conviction' [regarding their authenticity/pseudonymity]." (2)


ARGUMENT:
Andrew Queen Morton pioneered the field of computer©assisted study of Paul,
concluding (in 1963) from statistical analysis that Paul wrote only five
Ô
tTà
Ôletters:  the four ÃÃHauptbriefeÄÄ and Philemon.  While no 
contemporary scholar
has accepted Morton's conclusions, neither has anyone attempted extensive
research on the question.  To fill this gap, "[t]he present study will
describe and evaluate briefly many of Morton's studies putting his work in
proper perspective with the other numerous statistical analyses in biblical
and non©biblical areas and will correct or supplement all of them with
additional new procedures and data." (4)
ÁÁSince theology, literary dependency, and historical background have 
been
considered in other studies of these letters, this investigation will focus on
the two elements of the vocabulary and style of the letters.  "By `language'
is meant the particular vocabulary employed in a given writing.  `Lexical'
studies are synonymous.  The frequencies of each dictionary form and of
Ô
tT@
Ôvarious groups of vocabulary items, e.g., ÃÃhapax 
legomenaÄÄ and compound words,
are included." (5)  "Style," a word far more difficult to define, refers to
"the grammar, syntax, and construction of sentences" (6) ©© including
considerations of the deviation from those norms "for language, genre, topic,
media, region, and family" which impinge upon the author. (7)  In sum, style
is `the totality of all quantitatively tangible data concerning the formal
structure of a text'." (8)  Previous discussions of style have been based
largely on subjective judgments; this dissertation will develop an objective
method for judging stylistic differences and evaluating their statistical
significance.
ÁÁRobert Wachal has proposed an "authorship model" to approach the
Ô
tTØ'
Ôquestion of authenticity, and lists three kinds of models:  
the ÃÃconsistencyÔØ'
>0* ( (°°ÔÔ
tT
Ômodel,ÄÄ 
the ÃÃpopulation model,ÄÄ and the ÃÃresemblance model. (19)ÄÄ  Neumann 
employs
Ô
tTÈ
Ôthe ÃÃpopulation modelÄÄ, which entails applying various 
indices of style to six
Ô
tT

Ôof the undisputed letters (the ÃÃHauptbriefeÄÄ, 1 Thess, 
Phil; Phlm is omitted
because of its brevity), and then to a group of texts which definitely are not
Pauline (initially Heb, 1 Clem, and letters of Ignatius, Philo, Epictetus, and
Josephus).  Finally the disputed letters are measured "by those indices which
clearly distinguish the various authors, especially those setting off the
Pauline from the non©Pauline literature.  These indices combined by the
multivariate technique of discriminant analysis should indicate the
probability that the letters belong with the Pauline corpus or with the
writings of other authors." (20)
ÁÁIn Chapter Two, Neumann summarizes the prior mathematical and
statistical analyses in biblical and non©biblical literature, and evaluates
their utility for the problem of authenticity in the Pauline literature. 
Included in his review are lexical studies, morphological©length studies,
syntactic©length studies, morphological©category studies, syntactic©category
studies, and non©grammatical studies.  Chapter Three presents his detailed
plan for the investigation, including sixty©nine types of indices from the six
different types of studies reviewed in Chaper 2; by correlating these, one
arrives at 617 indices from which to choose for this computer©aided analysis. 
Neumann concludes Chapter Three by discussing computer procedures and
statistical models.  Texts were analyzed after removing Biblical quotations,
letter opening and closing formulae and, where applicable in regard to the
non©Pauline texts, dialogue segments.  Chapters Four and Five are devoted to a
presentation of the results of his stylo©statistical analyses and their
significance for Pauline studies.
ÁÁIn all, Neumann tried 101 indices ranging over the six different
categories mentioned in Chapter 2.  By testing the effectiveness of the
combinations of variables, and their degree of correlation, he was able to
reduce the number of variables to 24.  By combining variables across groups,
the STEPDISC program selected the six most statistically significant
variables, one each from the categories of lexical studies (choice of
synonymns) and syntactic©category studies ("D Statistic" [3©word sequences]
for 20 parts of speech), and four from the category of morphological©category
studies (articular participles/verbs; dependent genitives after their
substantives to total dependent genetives; relative pronouns/full©stop
sentence length © words; and demonstrative pronouns and interrogative
pronouns).  However, successive analyses showed that these six variables did
not successfully discriminate between authors, but rather between types of
literature.
ÁÁA subsequent analysis was done with only the Christian authors in the
non©Pauline group.  Of the 88 best indices, the six with the best Wilks'
lambda value were determined, and the DISCRIM test procedure selected the best
four.  These are the VARSET1 criteria:  word length in letters excluding iota
subscripts (WDLNG); position of first noun in modified full©stop sentence
(NOUNPOS); relative and indefinite pronouns (R&IPRON); and initial tau
(INITT).  This VARSET1 classification "works almost perfectly;" an Ignatius
sample is assigned to Hebrews. (190)  The NOUNPOS index showed some
irregularities, so it was replaced with one of the two remaining of the 
sixÔ€%0*H&H&˜˜Ôbest indices ©© the percent of initial 
modified full©stop sentence connectors
to the number of independent and major dependent clauses (INITSC).  This
Ô
tT

Ôformed VARSET2, which "produces the desired ÃÃ100% correct 
identificationÄÄ of the
test samples." (190)  Closer investigation of VARSET2, however, showed that
the variable INITSC is not reliable because it shows such abnormalities in
Clement and Ignatius:  "For Clement's samples the mean and standard deviation
are zero, and Ignatius has a mean near zero." (192)  Alternative combinations
were tried, until finally, VARSET1 was chosen as the most reliable set of
variables for the statistical analysis, while VARSET2 could be used to provide
further information.  Neumann argues that this is reasonable because the one
misassigned text sample was attributed to another non©Pauline author, not to
Paul.  Thus, "These [two variable] sets, especially VARSET1, may be applied
with reasonable confidence in their ability to discern whether the disputed
letters fit with Paul or the other authors.  Any other combinations [of
variables] are considered less reliable." (194)
ÁÁWhen the discriminant analysis is performed on the disputed letters 
of
Paul and three additional random samples from the undisputed letters, both
sets of variables misclassify a Galatians sample due to traditional formulae. 
Two additional sections from Galatians are omitted (4:4ff; 5:19©23), and the
analysis done once again.  This time both sets of variables are 100% reliable
for Paul; VARSET1 is 92.3% (12/13) and VARSET2 is 100% reliable for the four
non©Pauline authors.  Neumann then presents in detail the mean and standard
deviation figures for each of the variables in these two sets (207©210).
ÁÁApplication of these sets of variables to the disputed letters 
results
in all samples being assigned to Paul; "however, all except the third Ephesian
sample are somewhat far from typical Pauline values and have probabilities of
less than 0.10." (213)  For most indices, they fall between the first and
second deviation from the mean, or further.  The Pastoral Epistles are
assigned to non©Pauline authors.  Neumann views his research as showing that
Morton's conclusions about Pauline authenticity "are unjustified and need not
be accepted." (217)


EVALUATION:
Neumann's use of multileveled multivariate analysis is a tremendous
methodological advance over previous attempts at stylistic analysis of the
Biblical material.  The limited sample size of the Pauline corpus and the
disputed letters makes discriminant analysis difficult, but not impossible. 
More problematic, as he realizes, is the limited number of comparative
authors; inclusion of other letters would be a good way to further test the
validity of this research.  Neumann wisely suggests trying 1 Jn, Jas, 2 Pet, 2
Clem, Barnabas ©© "[e]specially potential authors of the disputed
letters. . . ." (222)
ÁÁNeumann's finding that Eph, Col, and 2 Thess are "closest to the 
Pauline
writing style" suggests that "there is little reason on the basis of style to
deny the authenticity of the disputed letters." (217)  But he rightly points
out that "the classification to the Pauline group does not mean that the
disputed letters are definitely authentic.  In fact, the distance
probabilities suggest the possibility of non©Pauline composition." (218) 
Ô€%0*H&H&˜˜ÔThere could be other reasons for the 
similarities in samples (e.g., the amount
of paranesis).  Nor can one assume that because there are no other known
candidates for authorship, therefore Paul must be the author.  Also, there may
be other reasons for disputing authorship of these three letters (e.g., genre,
theoogy, content, imitation) ©© issues which go beyond the scope of this study
but which deserve further attention.
ÁÁNeumann's first trial has contributed a set of indices which seem to
distinguish between the letter genre and other forms of literature.  It would
be worthwhile to attempt to replicate these results with other text samples.
Additionally, it would be beneficial for some to attempt to replicate his
research on the Pauline corpus, and also to test other indices which may be
found more effective.  He suggests that the F©ratio can be used for assessment
of these indices; its validity needs to be further established as well.
ÁÁOverall, this is a dense study which requires close inspection by 
both
statisticians and linguists.  Neumann lists those segments of the Pauline and
non©Pauline texts which he eliminates from consideration.  Pauline scholars
need to review these choices and see to what extent they can concur with his
determination.  Even if a scholarly consensus cannot be reached about the
final issue of authorship of the disputed letters, an agreement on the terms
of the debate will in itself raise the level of discussion.  If agreement can
be reached on these basic issues of text sample and statistical reliability of
the F©ratio, Neumann's research will have brought about a tremendous advance
toward the `clear conviction' which he seeks.