Sei sulla pagina 1di 39

THE GREEK VERBAL NETWORK VIEWED

FROM A PROBABILISTIC STANDPOINT:


AN EXERCISE IN HALLIDAYAN LINGUISTICS
STANLEY E. PORTER AND MATTHEW BROOK ODONNELL
This study explores numerical or distributional markedness in the
verbal network of the Greek of the New Testament. It extends the systemic
analysis of Porter (Verbal Aspect in the Greek of the New Testament, 1989),
making use of the Hallidayan concept of probabilistic grammar, which
posits a typology of systems where features are either equiprobable
both features are equally distributed (0.5/0.5)or skewedone feature
is marked by its low frequency of occurrence (0.9/0.1). The results con-
firm that the verbal aspect system of the Greek of the New Testament is
essentially independent of other verbal systems, such as voice and mood.
1. Introduction
The use of numerical methods in both traditional grammar and
modern linguistics has a chequered history within this century
1
. Early
field studies by such anthropologists and linguists as Boas emphasized
data-gathering as of paramount importance, with recognition of the dif-
ferences in the structures of languages
2
. This approach developed into a
1
This paragraph is dependent upon P.H. Matthews, Grammatical Theory in the
United States from Bloomfield to Chomsky (Cambridge Studies in Linguistics, 67;
Cambridge: Cambridge University Press, 1993), esp. pp. 5-48. For histories of the ear-
lier period, see J.T. Andresen, Linguistics in America 17691924: A Critical History
(London: Routledge, 1990); K.R. Jankowsky, The Neogrammarians: A Re-Evaluation of
their Place in the Development of Linguistic Science (Janua Linguarum, Series Minor, 116;
The Hague: Mouton, 1972); and R.H. Robins, A Short History of Linguistics (LLL;
London: Longman, 2nd edn, 1979), esp. pp. 164-240. For a contrasting history of the
development of linguistics in Britain, see R. Harris (ed.), Linguistic Thought in England
19141945 (London: Duckworth, 1988), where it is obvious that there was far less
systematic development, at least in part explaining why American linguistics has come
to dominate linguistic discussion. A better overview is G. Sampson, Schools of Linguistics
(Stanford: Stanford University Press, 1980).
2
F. Boas, Introduction, in Handbook of American Indian Languages, I (Washington,
DC: Government Printing Office, 1911; repr. Washington, DC: Georgetown University
Press, n.d.). The most noteworthy proponents of recognition of the differences in lan-
guages, and their relation to how humans think and speak, are E. Sapir, Language: An
Introduction to the Study of Speech (New York: Harcourt, Brace, 1921); and B.
Malinowski, The Problem of Meaning in Primitive Languages, in C.K. Ogden and I.A.
Richards, The Meaning of Meaning: A Study of the Influence of Language upon Thought and
of the Science of Symbolism (New York: Harcourt, 1923), pp. 296-336. Malinowski had a
significant influence on J.R. Firth (e.g. Papers in Linguistics 19341951 [Oxford: Oxford
University Press, 1957]), and Firth on M.A.K. Halliday (see below), and what has beco-
me known as the London School of functional linguistics.
Filologa Neotestamentaria - Vol. XIV - 2001, pp. 3-41
Facultad de Filosofa y Letras de Crdoba (Espaa)
more universalistic analysis of the constituent structures of observed lan-
guage and their relation to meaning, such as performed by Bloomfield in
the 1930s
3
. However, it was not until the post-Bloomfieldians of the
1950s era that quantification of the empirical results of such study began
to take place in a concerted and serious way. This resulted in what has
been widely recognized as descriptivist linguistics, although to a large
extent at the expense of meaning. These descriptivist efforts, led by Harris
in his distributional method
4
, were relatively short lived, with the deve-
lopment and subsequent dominance of Chomskyan thought in the
1960s
5
(what some have characterized as a revolution)
6
. Although buil-
ding upon the analysis of predecessors, Chomskyan linguistics soon came
to be equated with theoretical linguistics, grounding linguistic theory in
formal rules and emphasizing deductive as opposed to inductive discovery
procedures. Since the early numerical studies did not progress far beyond
analysis of the level of the phoneme, and to some extent the morpheme,
4 Stanley E. Porter and Matthew Brook ODonnell
3
L. Bloomfield, Language (London: George Allen & Unwin, 1935). On the relation
between Boas, Sapir and Bloomfield, often overlooked in discussion of these significant
scholars, see M.R. Haas, Boas, Sapir, and Bloomfield, in W.L. Chafe (ed.), American
Indian Languages and American Linguistics (Lisse: de Ridder, 1976), pp. 59-69; repr. as
Boas, Sapir, and Bloomfield: Their Contribution to American Indian Linguistics, in
A.S. Dil (ed.), Language, Culture, and History: Essays by Mary R. Haas (Stanford:
Stanford University Press, 1978), pp. 194-206.
4
See Z.S. Harris, Methods in Structural Linguistics (Chicago: University of Chicago
Press, 1951; repr. as Structural Linguistics [Chicago: University of Chicago Press, 1960]);
and his Papers in Structural and Transformational Linguistics (Formal Linguistics Series,
1; Dordrecht: Reidel, 1970). Other important work in this area includes C.C. Fries, The
Structure of English: An Introduction to the Construction of English Sentences (London:
Longman, 1952); H.A. Gleason, An Introduction to Descriptive Linguistics (New York:
Holt, Rinehart and Winston, 1955; rev. edn, 1961).
5
Some of his most important works are N. Chomsky, Syntactic Structures (Janua
Linguarum, Series Minor, 4; The Hague: Mouton, 1957); idem, Aspects of the Theory of
Syntax (Cambridge, MA: MIT Press, 1964); idem, Current Issues in Linguistic Theory
(Janua Linguarum, Series Minor, 38; The Hague: Mouton, 1964); idem, Topics in the
Theory of Generative Grammar (Janua Linguarum, Series Minor, 56; The Hague:
Mouton, 1966); idem, Language and Mind (New York: Harcourt Brace Jovanovich,
1968; 2nd edn, 1972); among others.
6
See N. Smith and D. Wilson, Modern Linguistics: The Results of Chomskys
Revolution (Harmondsworth: Penguin, 1979). Those not conquered by the Chomskyan
revolution include tagmemics (K.L. Pike, Language in Relation to a Unified Theory of the
Structure of Human Behavior [Janua Linguarum, Series Maior, 24; The Hague: Mouton,
2nd end, 1967]), stratificational and relational grammars (e.g. L. Hjelmslev, Principes de
grammaire gnrale [Copenhagen: Hoest, 1928; ET Prolegomena to a Theory of Language
(trans. F. Whitfield; Madison: University of Wisconsin Press, 1953)] and S. Lamb,
Outline of Stratificational Grammar [Washington, DC: Georgetown University Press,
1966]), and the London School (e.g. Firth and M.A.K. Halliday, Halliday: System and
Function in Language [ed. G. Kress; Oxford: Oxford University Press, 1976]). For a
synopsis of Firths work, much of it in Firths own words, and Hallidays work, again
much of it in his own words, see R. De Beaugrande, Linguistic Theory: The Discourse of
Fundamental Works (LLL; London: Longman, 1991), pp. 187-222, 223-64, respectively.
they were easily disregarded by Chomsky as irrelevant to his major con-
cernthat of syntactical structures, especially at what soon came to
be called the deep structure
7
. From this point on
8
, in many if not
most mainstream linguistic circles, numerical language studies have been
looked down upon, with the resulting tendency to characterize them as
simply counting exercises or mere data collection, representative of a
previous generation of linguistic study. However, this is more a reflection
of the influence of a particular school of thought, Chomskys, on the
questions of linguistic investigation, rather than an accurate picture of the
situation of the time. Already in 1965, Ivic devoted a chapter of his work
on trends in linguistics to what he called mathematical linguistics, in
which he discussed quantitative or statistical linguistics
9
. Although Ivic
saw the closest ties between statistical linguistics and such areas as infor-
mation theory and machine translation, other grammarians, especially
those with functionalist interests such as Firth, had also continued to
study language numerically
10
. In fact, the major development of nume-
rical language studies, bringing the descriptivist agenda back into the
forefront of discussion although this time not as an area in itself but as a
methodological basis for linguistic research, has been in the field of cor-
pus linguistics
11
. Nevertheless, it is only fairly recently that such studies
have become recognized, although not widely and in all linguistic circles.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 5
7
In his first major work, Syntactic Structures, Chomsky wrote, Despite the unde-
niable interest and importance of semantic and statistical studies of language, they appear
to have no direct relevance to the problem of determining or characterizing the set of
grammatical utterances. I think that we are forced to conclude that grammar is autono-
mous and independent of meaning, and that probabilistic models give no particular
insight into some of the basic problems of syntactic structure (p. 17).
8
Chomskyan linguistics and its offspring have developed in many different direc-
tions, especially in terms of the question of meaning. Some of the recent differences of
opinion are chronicled in R.A. Harris, The Linguistics Wars (New York: Oxford
University Press, 1993); G.J. Huck and J.A. Goldsmith, Ideology and Linguistic Theory:
Noam Chomsky and the Deep Structure Debates (London: Routledge, 1995).
9
M. Ivic, Trends in Linguistics (trans. M. Heppell; Janua Linguarum, Series Minor,
42; The Hague: Mouton, 1965), esp. pp. 212-24; cf. J. Whatmough, Language: A
Modern Synthesis (New York: New American Library, 1956), pp. 179-97.
10
See M. Stubbs, British Traditions in Text Analysis: From Firth to Sinclair, in M.
Baker, G. Francis and E. Tognini-Bonelli (eds.), Text and Technology: In Honour of John
Sinclair (Philadelphia: Benjamins, 1993), pp. 1-33; rev. to include Halliday in M.
Stubbs, Text and Corpus Analysis: Computer-Assisted Studies of Language and Culture
(Oxford: Blackwell, 1996), pp. 22-50.
11
See G. Leech, Corpora and Theories of Linguistic Performance, in J. Svartvik
(ed.), Directions in Corpus Linguistics: Proceedings of Nobel Symposium 82 Stockholm, 48
August 1991 (Trends in Linguistics: Studies and Monographs, 65; Berlin: Mouton de
Gruyter, 1992), pp. 106-22. Other useful introductions to corpus linguistics, besides
those cited above, are J. Sinclair, Corpus, Concordance, Collocation (Describing English
Language; Oxford: Oxford University Press, 1991); D. Biber, S. Conrad and R. Reppen,
Corpus Linguistics: Investigating Language Structure and Use (Cambridge Approaches to
Linguistics; Cambridge: Cambridge University Press, 1998).
The history of numerical language studies of the Greek of the New
Testament does not follow the same path of development, but the results
are not significantly different in many ways. In the study of the Greek of
the New Testament, and biblical studies in general, it is evident, as many
have recently pointed out, that most of the grammatical reference tools
are pre-linguistic and thus have been unaffected by the major develop-
ments within modern linguistics, several of which have been traced
above
12
. These tools are instead shaped by both traditional grammar,
which traces its way back to the ancient Greek grammarians, and nine-
teenth-century comparative philology
13
. Roberstons gigantic, A Grammar
of the Greek New Testament in the Light of Historical Research, is perhaps
the most comprehensive example of the application of comparative phi-
lology to New Testament grammar
14
. Roberston attempts a detailed clas-
sification of features of the Greek of the New Testament, listing numerous
examples, but without specific numerical figures in the main body of his
grammar. For example, discussing the use of the, so-called, declarative
., he states, the verbs that use declarative . in the N. T. are very
numerous
15
. The grammar is full of statements such as in general,
usually, the most frequent use of X, and so on. A number of charts,
compiled by a Mr H. Scott, were appended to the third edition, but these
are essentially tables that simply chronicle instances
16
. The same situation
holds true for most of the other standard grammars, including Moulton
17
,
BlassDebrunner
18
, and Turner
19
. As desirable as it might be to have
more complete and precise information, it is probably better to have none
than have inaccurate information. Turner includes a number of charts in
6 Stanley E. Porter and Matthew Brook ODonnell
12
See, for example, D.D. Schmidt, Hellenistic Greek Grammar and Noam Chomsky:
Nominalizing Transformations (SBLDS, 62; Atlanta: Scholars Press, 1981), esp. pp. 3-13;
S.E. Porter, Verbal Aspect in the Greek of the New Testament, with Reference to Tense and
Mood (SBG, 1; New York: Peter Lang, 1989), pp. 50-65; cf. idem, Studies in the Greek
New Testament: Theory and Practice (SBG, 6; New York: Peter Lang, 1996), pp. 39-48.
13
See S.E. Porter and J.T. Reed, Greek Grammar Since BDF: A Retrospective and
Prospective Analysis, FN 4 (1991), pp. 143-64; cf. S.E. Porter, Studying Ancient
Languages from a Modern Linguistic Perspective: Essential Terms and Terminology,
FN 2 (1989), pp. 147-72.
14
A.T. Robertson, A Grammar of the Greek New Testament in the Light of Historical
Research (Nashville: Broadman, 4th edn, 1934).
15
Robertson, Grammar of the Greek New Testament, p. 1034.
16
Robertson, Grammar of the Greek New Testament, pp. 1385-1431.
17
J.H. Moulton, A Grammar of New Testament Greek. I. Prolegomena (Edinburgh: T.
& T. Clark, 3rd edn, 1908); idem and W.F. Howard, A Grammar of New Testament
Greek. II. Accidence and Word-Formation (Edinburgh: T. & T. Clark, 1929).
18
F. Blass and A. Debrunner, A Greek Grammar of the New Testament and Other Early
Christian Literature (trans. R.W. Funk; Chicago: University of Chicago Press, 1961); F.
Blass and A. Debrunner, Grammatik des neutestamentlichen Griechisch (ed. F. Rehkopf;
Gttingen: Vandenhoeck & Ruprecht, 17th edn, 1990).
19
N. Turner, A Grammar of New Testament Greek. III. Syntax (Edinburgh: T. & T.
Clark, 1963).
his syntax of the Greek New Testament, but it has been shown that a
number of his statistics and analyses are in fact wrong
20
.
The one area of New Testament studies where numerical methods
have been used has been in stylistic studies. The standard New Testament
Greek reference grammars contain references to stylistic features, and
often include some numerical figures for specific linguistic elements. The
most comprehensive work of this kind is Turners Style, which com-
pletes the grammar started by Moulton
21
. In his treatment of the stylistic
features of the New Testament documents and comparisons of different
books, Turner includes counts of vocabulary and grammatical items. A
more specific branch of stylistics, sometimes referred to as stylometrics,
is concerned to describe the stylistic features of a given author, often with
the application of these findings to determine the authorship of dispu-
ted documents. Here linguistic features, such as word frequency (parti-
cularly the number of hapax legomena), word-order, the number and
position of grammatical words (such as -c.), and other similar factors
are counted and compared for the documents under investigation. A
number of early stylometric studies argued for non-Pauline authorship of
the Pastoral Epistles from the fact that they contain a higher number of
hapax legomena than the main Pauline epistles
22
. Recent studies have
argued in the opposite direction, however
23
. One must simply note here,
however, that for most of these studies, as currently undertaken, the sam-
ple sizes do not allow for the kinds of analyses made, and clearly not for
the certainty of conclusions that are often drawn
24
.
During the past two decades, as already noted above, there has been
a gradual re-emergence of numerical linguistic studies, and the question-
The Greek Verbal Network Viewed from a Probabilistic Standpoint 7
20
See G.H.R. Horsley, New Documents Illustrating Early Christianity. V. Linguistic
Essays (New South Wales: Macquarie University, The Ancient History Documentary
Research Centre, 1989), pp. 49-65, esp. pp. 52-55.
21
N. Turner, A Grammar of New Testament Greek. IV. Style (Edinburgh: T. & T. Clark,
1976). In volumes 3 and 4, Turner took Moultons grammar in a very different direction
than was originally conceived by Moulton (see n. 17 above), arguing for a Semitized
Greek, over the Greek of the New Testament being that of the koine. Schmidt, in his
overview of the history of Hellenistic Greek grammar, refers to Turners contribution to
Moultons grammar as a definite linguistic regression (Schmidt, Hellenistic Greek
Grammar and Noam Chomsky, p. 9). See also A.B. Spencer, Pauls Literary Style: A
Stylistic and Historical Comparison of II Corinthians 11:1612:13, Romans 8:9-39, and
Philippians 3:24:13 (Lanham: University Press of America, 1998).
22
A.Q. Morton and J. McLeman, Paul, the Man and the Myth: A Study in the
Authorship of Greek Prose (New York: Harper & Row, 1966).
23
See A. Kenny, A Stylometric Study of the New Testament (Oxford: Clarendon Press,
1986).
24
For a critical survey and linguistic evaluation of such studies, see M.B. ODonnell,
Linguistic Fingerprints or Style by Numbers? The Use of Statistics in the
Determination of Authorship of New Testament Documents, in S.E. Porter and D.A.
Carson (eds.), Linguistics and the New Testament: Critical Junctures (JSNTSup, 168;
SNTG, 5; Sheffield: Sheffield Academic Press, 1999), pp. 206-62.
ing of Chomskyan methodological priorities. Advances in computer
technology and the availability of powerful and relatively inexpensive
desktop machines have acted as a catalyst for this revival. This area of lin-
guistics, called corpus linguistics due to the central role of machine read-
able corpora of real language serving as the databases for grammatical
study
25
, is in many ways ideally suited to the study of the Greek of the
New Testament. In most forms of contemporary linguistic analysis,
judgments are made on the basis of the intuitions of native speakers
regarding the grammaticality of utterances. For the study of the Greek
New Testament, there are no native speakers of the language available to
verify the grammaticality of constructed sentences, so one must rely
upon the corpus of available texts. In addition, the New Testament cor-
pus with grammatical (part-of-speech) annotation is available in ma-
chine readable form with a number of software tools to access it, such as the
GRAMCORD and BibleWorks programs. The grammarian of the Greek
New Testament is now able quickly to compile results from searches for
numerous combinations of words and grammatical constructions. This
presents exciting options for future grammatical work, but it also has
potential pitfalls, with the real danger of an increase in the number of
unstructured numerical studies that lack a theoretical grounding
26
.
The ability to collect and collate vast amounts of linguistic data means lit-
tle without a theoretical framework in which these results can be evaluated
27
.
Attempts to apply syntactical theories, such as various forms of Chomskyan
derived grammar, to the Greek of the New Testament have achieved only
modest success
28
. However, a number of recent studies applying models of
8 Stanley E. Porter and Matthew Brook ODonnell
25
On the history and theoretical basis of corpus linguistics, see G. Leech, The State
of the Art in Corpus Linguistics, in K. Aijmer and B. Altenberg (eds.), English Corpus
Linguistics: Studies in Honour of Jan Svartvik (London: Longman, 1991), pp. 8-29; J.
Svartvik, Corpus Linguistics Comes of Age, in Svartvik (ed.), Directions in Corpus
Linguistics, pp. 7-13.
26
On the use of annotated corpora for the study of the Greek of the New Testament,
see M.B. ODonnell, The Use of Annotated Corpora for New Testament Discourse
Analysis: A Survey of Current Practice and Future Prospects, in S.E. Porter and J.T.
Reed, Discourse Analysis and the New Testament: Approaches and Results (JSNTSup, 170;
SNTG, 4; Sheffield: Sheffield Academic Press, 1999), pp. 71-117.
27
There is obviously a need for interaction between theory and data, in that the data
test and, if necessary, correct the theory, while the theory provides limits and structure
to the collection of data. There can be no such thing as an entirely inductive approach,
one that begins with no theory, but instead aims to construct one only on the basis of
the data. See K.R. Popper, Conjectures and Refutations: The Growth of Scientific
Knowledge (London: Routledge and Kegan Paul, 1963; 4th edn, 1974), esp. pp. 3-59; I.
Lakatos, Falsification and the Methodology of Scientific Research Programmes, in I.
Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge: Proceedings of
the International Colloquium in the Philosophy of Science, London, 1965, volume 4
(Cambridge: Cambridge University Press, 1970), pp. 91-196, as well as other essays in
this volume.
28
Examples are Schmidt, Hellenistic Greek Grammar and Noam Chomsky; R.
Wonneberger, Generative Stylistics: An Algorithmic Approach to Stylistic and Source
functional grammar, particularly Hallidays systemic functional grammar,
have shown the descriptive power and flexibility of a functional linguistic
approach for analyzing the Greek language of the New Testament
29
. It is
interesting to note that Halliday, unlike Chomsky, has never dismissed the
statistical approach to linguistic investigation. On the contrary, he states:
It has always seemed to me, ever since I first tried to become a gramma-
rian, that grammar was a subject with too much theory and too little data
30
.
In a number of recent studies, Halliday has begun to investigate the proba-
bilistic nature of grammar, and how the relative frequency of grammatical ele-
ments in a corpus of naturally occurring language can be incorporated into the
paradigmatic system networks that form the basis of systemic linguistics
31
.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 9
Data Retrieval Problems Based on Generative Syntax, in W. Vandemeghe and M. Van
de Velde (eds.), Bedeutung, Sprechakte und Texte: Akten des 13. Linguistischen
Kolloquiums, II (Tbingen: Niemeyer, 1979); idem, Syntax und Exegese: Eine generative
Theorie der griechischen Syntax und ihr Beitrag zur Auslegung des Neuen Testaments, dar-
gestellt an 2. Korinther 5.2f und Rmer 3.21-26 (Beitrge zur biblischen Exegese und
Theologie, 13; Frankfurt: Peter Lang, 1979); J.P. Louw, Semantics of New Testament
Greek (Philadelphia: Fortress Press; Chico, CA: Scholars Press, 1982); R.A. Young,
Intermediate New Testament Greek: A Linguistic and Exegetical Approach (Nashville:
Broadman Press, 1994); M.W. Palmer, Levels of Constituent Structure in New Testament
Greek (SBG, 4; New York: Peter Lang, 1995).
29
For example, see Porter, Verbal Aspect; idem, Idioms of the Greek New Testament
(BLG, 2; Sheffield: JSOT Press, 2nd edn, 1994); J.T. Reed, A Discourse Analysis of
Philippians: Method and Rhetoric in the Debate over Literary Integrity (JSNTSup, 136;
SNTG, 3; Sheffield: Sheffield Academic Press, 1997); G. Martn-Asensio, Hallidayan
Functional Grammar as Heir to New Testament Rhetorical Criticism, in S.E. Porter and
D.L. Stamps (eds.), The Rhetorical Interpretation of Scripture: Essays from the 1996 Malibu
Conference (JSNTSup, 180; Sheffield: Sheffield Academic Press, 1999), pp. 84-107; idem,
Foregrounding and its Relevance for Interpretation and Translation, with Acts 27 as a
Case Study, in S.E. Porter and R.S. Hess (eds.), Translating the Bible: Problems and
Prospects (JSNTSup, 173; Sheffield: Sheffield Academic Press, 1999), pp. 189-223; idem,
Transitivity-Based Foregrounding in the Acts of the Apostles: A Functional-Grammatical
Approach to the Lukan Perspective (JSNTSup, 202; SNTG, 8; Sheffield: Sheffield
Academic Press, 2000); and various essays in Porter and Reed (eds.), Discourse Analysis
and the New Testament.
30
M.A.K. Halliday, Language as System and Language as Instance: The Corpus as a
Theoretical Construct, in Svartvik (ed.), Directions in Corpus Linguistics, pp. 61-77,
quotation p. 61.
31
See Halliday, Language as System and Language as Instance; idem, Towards
Probabilistic Interpretations, in E. Ventola (ed.), Functional and Systemic Linguistics:
Approaches and Uses (Trends in Linguistics, 55; Berlin and New York: Mouton de Gruyter,
1991), pp. 39-61; idem, Corpus Studies and Probabilistic Grammar, in Aijmer and
Altenberg (eds.), English Corpus Linguistics, pp. 30-43; idem, Quantitative Studies and
Probabilities in Grammar, in M. Hoey (ed.), Data, Description, Discourse: Papers on the
English Language in Honour of John McH. Sinclair (London: HarperCollins, 1993), pp. 1-
25; idem and Z.L. James, A Quantitative Study of Polarity and Primary Tense in the
English Finite Clause, in J.M. Sinclair, M. Hoey and G. Fox (eds.), Techniques in
Description: Spoken and Written Discourse (London: Routledge, 1993), pp. 32-66.
In this article, we build on the theoretical work by Porter in Verbal
Aspect in the Greek of the New Testament that systemically analyzed the
Greek verbal network, focusing particularly on the morphological fea-
tures of tense-form and mood. Presenting his findings in terms of marked-
ness theory, he introduced, but did not pursue in the kind of detail
currently achievable, the area of distributional markedness
32
. In this
study, we take the systemic verbal network developed in Verbal Aspect
33
and add frequency information to the systems, both individually (for
example the ASPECTUALITY system) and in combination (for example FI-
NITENESS and ASPECT1). We test our findings against Hallidays hypotheses
about the two main types of system in language, and his viewpoint on
language as both a system and an instance (see below). We also present a
tentative analysis of the voice-system in Greek from a distributional
standpoint, pending future work on voice
34
. The study demonstrates the
integration of empirical and theoretical linguistic analysis, and aims to
provide a suggestive paradigm for future numerical studies of the Greek
of the New Testament.
2. Theory: Probabilistic Lexicogrammar
Recent work by Halliday has stressed the fact that grammar is
inherently probabilistic
35
. He contends that this has always been his
view, but that the counting and analysis of sufficient amounts of lin-
guistic data has only recently become available with the advent of
affordable computer equipment, and, more importantly, access to
machine readable corpora
36
. He is somewhat mystified by the fact
that linguists are comfortable with the counting of words (lexical
items) and with assigning probabilities of occurrence to these words,
but they often object when the same task is attempted for grammati-
10 Stanley E. Porter and Matthew Brook ODonnell
32
See Porter, Verbal Aspect, pp. 178-81, esp. p. 181.
33
See Porter, Verbal Aspect, p. 109.
34
See S.E. Porter, Voice in the Greek of the New Testament (in preparation).
35
Halliday, Language as System and Language as Instance, p. 65.
36
In a 1961 article, Halliday stated that It is not simply that all grammar can be sta-
ted in probability terms, based on frequency counts in texts: this is due to the nature of
a text as a sample. But the very fact that we can recognize primary and secondary struc-
turesthat there is a scale of delicacy at allshows that the nature of language is not to
operate with relations of always this and never that (M.A.K. Halliday, Categories of
the Theory of Grammar, Word 17 [1961], pp. 241-92, quotation p. 259; repr. in
Halliday: System and Function, pp. 52-72, quotation p. 63). Elsewhere he observes that:
It seemed to me clear in 1960 that useful theoretical work in grammar was seriously
hampered by lack of data; we depended on the corpus as a resource for further advance.
Moreover it would have to be computerized, in the sense that some part of the work
would be performed computationally to permit large-scale frequency studies (Halliday,
Language as System and Language as Instance, p. 64).
cal features and classes
37
. People, according to Halliday, have little
difficulty accepting that they use the word go more often than the
word walk, and walk more often than stroll to describe motion.
However, when the same understanding is applied to grammatical fea-
turesthat a person is more likely to use an active construction than
a passive, or a positive rather than a negative clauseHalliday has
observed that many people object very strongly, and protest that they
have a perfect right to choose otherwise if they wish
38
.
In order to apply a probabilistic understanding to the grammar of a
language, one must adopt a paradigmatic view of grammar. Such a model
arranges grammatical categories (aspect, voice, person, gender, etc.) into
combinations of either/or choicesfor instance a noun is either singular
or plural. The language user is most often completely unaware that such
choices are being madeand thus the choices are unconsciousyet the
language system requires that such choices be made
39
. Each grammatical
category in a particular system is given meaning by its relationship to the
other categories within the system. Nesbitt and Plum state that: A system
is defined as an entry condition together with a set of mutually exclusive
options or features, one of which must be selected
40
. So, for instance,
plurality has little semantic value unless it is viewed as a choice against (or
instead of ) singularity. The use of system in this technical, linguistic sense
The Greek Verbal Network Viewed from a Probabilistic Standpoint 11
37
Halliday, Corpus Studies and Probabilistic Grammar, p. 31; idem, Quantitative
Studies, p. 2. There has been a considerable amount of investigation into numerical
behaviour of lexis in language, resulting in a number of vocabulary measures, such as
Zipf s Law (G.K. Zipf, The Psychobiology of Language [Boston: Houghton Mifflin,
1935]) and various forms of the Type-Token ratio. These measures have often been uti-
lized in authorship attribution studies (e.g. G.U. Yule, The Statistical Study of Literary
Vocabulary [Cambridge: Cambridge University Press, 1944]; A. Ellegrd, A Statistical
Method for Determining Authorship: The Junius Letters, 17691772 [Gteborg: Elanders
Boktryckeri Aktiebolag, 1962]). See ODonnell, Linguistic Fingerprints, pp. 215-16,
230-39.
38
Halliday, Quantitative Studies, p. 3. Halliday suggests that: The resistance seems
to arise because grammar is buried more deeply below the level of our conscious aware-
ness and control; hence it is more threatening to be told that your grammatical choices
are governed by overall patterns of probability.
39
See Porter, Verbal Aspect, p. 9, on conscious and unconscious choice in terms of
systemic linguistics.
40
C. Nesbitt and G. Plum, Probabilities in a Systemic-Functional Grammar: The
Clause Complex in English, in R.P. Fawcett and D. Young (eds.), New Developments in
Systemic Linguistics. II. Theory and Application (London: Pinter, 1988), pp. 6-38, quota-
tion p. 7. They continue: The entry condition of a system is itself an option in a prior
system. So the environment of choice is always that of choices already made. In this way
systems form networks of systems organized according to the logical priority of certain
options over other options. See also R.P. Fawcett, Cognitive Linguistics and Social
Interaction: Towards an Integrated Model of a Systemic Functional Grammar and the Other
Components of a Communicating Mind (Heidelberg: Julius Groos and Exeter University,
1980), pp. 19-25; Porter, Verbal Aspect, pp. 7-16.
can be traced back through Halliday to the work of Firth
41
. Firths work
applied the theory of system (paradigmatic choices) and structure (syntag-
matic choices) primarily to the phonological level of language
42
. Halliday
has extended Firths work to include both grammar and lexis. In fact,
systemic linguistics rejects the traditional distinction between lexis (lexi-
cal semantics treated in a lexicon) and grammar (morphological patterns
discussed in grammar books). Instead, systemic theory talks about the
lexicogrammar of languagethat is, a continuum (or cline) of paradig-
matic systems, with grammar (as traditionally described) at one end and
lexis at the other
43
. The fundamental concept of a system as a choice be-
tween semantic features is present throughout the lexicogrammatical con-
tinuum. Nevertheless, the systems at the grammatical end consist of a
small, finite number of feature selections, and can thus be described as
closed-systems, while systems at the lexis end consist of numerous sub-
systems (consider, for instance, how many verbs of motion there are), and
are described as open-systems. An example of a closed-system is the aspec-
tual system in Hellenistic Greek (see section 3). There are at most four
choices to be made by the language user who wishes to speak of a process,
with the choice from a previous system becoming the entry condition for
the next systemic choice: (1) a choice must be made with regard to the
system of ASPECTUALITY (+expectational or +aspectual), (2) if the choice is
+aspectual then the ASPECT1 system is the entry condition for the next
choice (+perfective or perfective), (3) if perfective is selected then two
further co-ordinated systems must be entered: (a) ASPECT2 becomes the
entry condition for one set of choices (+imperfective or +stative) and (b)
REMOTENESS for the other (remote or +remote). It is clear how at each
point an either this or that selection must be made, and that there are
only a finite number of such choices, thus the aspectual system can be
described as a closed-system.
It should not be difficult to see how probabilities could be incorpo-
rated into a systemic model of language
44
. Given that for each system in a
network a (usually) binary choice is made, then it is simply a matter of
12 Stanley E. Porter and Matthew Brook ODonnell
41
See J.R. Firth, A Synopsis in Linguistic Theory, 19301955, in J.R. Firth et al.,
Studies in Linguistic Analysis (Special Volume of the Philological Society; Oxford: Basil
Blackwell, 1957), pp. 1-32. He notes that the discussion of grammatical categories in
closed systems for any given language highlights the fact that meanings are deter-
mined by their inter-relations in the systems set up for that language (Synopsis, p. 22).
42
See Firth, Synopsis, pp. 17-22.
43
Halliday, Corpus Studies and Probabilistic Grammar, p. 32; idem, Language as
System and Language as Instance, p. 63. Cf. R. Hasan, The Grammarians Dream:
Lexis as Most Delicate Grammar, in M.A.K. Halliday and R.P. Fawcett (eds.), New
Developments in Systemic Linguistics. I. Theory and Description (London: Pinter, 1987),
pp. 184-211.
44
Nesbitt and Plum suggest that with choice as the basis of our theory of language,
grammar can be modelled as sets of possibilities, as a potential for making meaning
(Probabilities and a Systemic-Functional Grammar, p. 7).
assigning a probability to each potential outcome (either A or B). If it is
equally likely that either outcome will occur, then each will be given a pro-
bability of 0.5. If outcome A is expected to occur 9 times out of every ten
choices, then A will be assigned a probability of 0.9 and B will be one
minus the probability of A, thus 1 - 0.9 = 0.1. These probabilities should
not be viewed as operating in a predictive manner, in saying the next choi-
ce should be A with odds of nine to one, but rather as providing an inter-
pretative framework for choices once they have been made
45
. Consider
system X, in which the probabilities of outcomes A and B are 0.9 and 0.1
respectivelythat is, A is nine times more likely to occur than Bif B is
chosen, then we can define B not only as not A, but not A against odds
of nine to one. These probabilities not only provide information about
general patterns of grammatical features within a language, but they can
also be used to help define the semantics of these features.
In the introduction we noted the influence of Chomskys syntactical
theory upon linguistics in the past nearly forty years, and particularly how
his negative view of probabilistic modelling has discredited the use of
numerical methods in theoretical linguistics. In addition, his restatement
of Saussures langue and parole distinction in terms of linguistic compe-
tence and linguistic performance has resulted in the consignment of data
obtained from real texts to the realm of applied linguistics, while theore-
tical linguistics has concerned itself with logical and theoretical concepts
such as grammaticality
46
. De Beaugrande has counted the number of sen-
tences which are found in Chomskys early works, Syntactic Structures and
Aspects of the Theory of Syntax, and found that only twenty-eight inven-
ted sentences in SS [Syntactic Structures] and twenty-four in AT [Aspects
of the Theory of Syntax] are analysed
47
. It is not therefore surprising that
Chomsky not only rejected the value of naturally occurring language for
the study of syntactic structure, but concluded that probabilistic models
could bring little insight to this task
48
.
In his work, Halliday rejects the Saussurian langueparole dualism, and
especially Chomskys competenceperformance division, and instead offers
The Greek Verbal Network Viewed from a Probabilistic Standpoint 13
45
Halliday, Corpus Studies and Probabilistic Grammar, pp. 32-33.
46
A standard introduction to theoretical linguistics is J. Lyons, Introduction to
Theoretical Linguistics (Cambridge: Cambridge University Press, 1968). An examina-
tion of this work fails to uncover a single textual example larger than the individual
sentence, and all of these examples are invented instead of being drawn from texts or
instances of speech. As Stubbs indicates, another influential work by Lyons, his two
volume Semantics (2 vols.; Cambridge: Cambridge University Press, 1977), does not
analyse a single authentic text or text fragment (Stubbs, Text and Corpus Analysis,
p. 30).
47
De Beaugrande, Linguistic Theory, p. 176. Halliday suggests that Chomskys sar-
castic observation that I live in New York is more frequent than I live in Dayton Ohio
was designed to demolish the conception that relative frequency in text might have any
theoretical significance (Corpus Studies and Probabilistic Grammar, p. 30).
48
Chomsky, Syntactic Structures, p. 17.
the complementary ideas of language as system and language as instance.
These are not two different entities, but rather two viewpoints upon one
phenomenon, that is, the phenomenon of language. Viewing language as a
system involves taking a long-term view on multiple instances of language
use and making generalizations, whereas viewing language as an instance is
an examination of just one occurrence of language in use. Halliday offers
the illustration of examining the weather in terms of climate (weather viewed
as a system over an extended period) and in terms of day by day weather
patterns (weather viewed as an instance). He suggests that:
There is only one set of phenomena here: the meteorological processes
of precipitation, movement of air masses and the like, which we observe in
close-up, as text, or else in depth, as system. But one thing is clear: the
more weather we observe, as instance-watchers, the better we shall perform
as system-watchers when we turn to explaining the climate
49
.
Halliday began simple frequency counts for the Chinese language
(Mandarin) in the early 1950s (he completed his work on Chinese in
1955, just before the publication of Chomskys Syntactic Structures), and
assigned probabilities to the terms in the grammatical systems
50
. From
these frequency counts and probabilities, he wanted to discover the extent
of the association between different systems
51
. One such association
investigated by Halliday was negative interrogatives, where there is a com-
bination of a choice from the polarity (negative vs. positive) and the indi-
cative (declarative vs. interrogative) systems
52
. From this early work, and
from later work carried out on small samples of English, he developed a
typology of systems. He noticed that when probabilities where applied
to the grammatical systems he studied, the systems generally fell into one
of two groups: (1) those where the two terms in the system have equal pro-
bability of occurring (0.5), and thus there is no marked or unmarked term,
and (2) those where there is an unmarked term with a probability of
roughly 0.9, and a marked term with a 0.1 probability. These two types of
system are referred to as equiprobable and skewed systems, respectively
53
.
Halliday further suggests that the fact that a language will primarily pos-
14 Stanley E. Porter and Matthew Brook ODonnell
49
Halliday, Language as System and Language as Instance, p. 66. Cf. idem, A Brief
Sketch of Systemic Grammar, in Halliday: System and Function, pp. 3-6, esp. p. 3.
50
M.A.K. Halliday, The Language of the Chinese Secret History of the Mongols
(Publications of the Philological Society, 17; Oxford: Basil Blackwell, 1959), esp. pp. 207-26.
51
Halliday, Quantitative Studies, p. 3.
52
He wondered whether it was possible to predict the number of instances of nega-
tive interrogative by intersecting the probabilities of negative (versus positive) with those
of interrogative (vs. declarative) (Halliday, Quantitative Studies, p. 3).
53
Halliday, Language as System and Language as Instance, p. 65. He notes that this
typology of systems arose initially out of his work on Chinese and was then confirmed
by later work on a small corpus of English. He also notes that Svartviks work on voice
in English (J. Svartvik, On Voice in the English Verb [Janua Linguarum, Series Practica,
63; The Hague: Mouton, 1966]) provided further support for his theory.
sess equiprobable and skewed systems is supported by the semiotic func-
tion of language. If all of the systems where equiprobable, there would be
no redundancy in the language and any interference (noise) would ham-
per effective communication. On the other hand, if systems fell into a
whole range of distributions, from 0.5/0.5 all the way up to 0.99/0.01
then the semiotic systemwould be virtually impossible to learn
54
. The
language learner soon recognizes, and is thus able to master, systems where
there is no unmarked term (equiprobable [0.5/0.5]) or where one term is
highly marked in comparison to the other (skewed [0.9/0.1])
55
.
Halliday uses the concept of markedness, as noted above, but does not
define these categories, apparently opting for an intuitive understanding
of the concept
56
. As we have noted above, distributional markedness
along with material (morphological) markedness, implicational marked-
ness and semantic markednesswas introduced and defined in Porters
Verbal Aspect as a means of describing tendencies in Greek tense-form
usage, although this was not pursued in detail. The concept of marked-
ness was first introduced by Trubetzkoy
57
, and then extended by
Jakobson
58
, in Prague-school linguistics to describe binary distinctive
phonemic features or, by extension, distinctive conceptual features.
According to this model, an item is marked if it displays this distinctive
The Greek Verbal Network Viewed from a Probabilistic Standpoint 15
54
Halliday, Corpus Studies and Probabilistic Grammar, p. 36.
55
Halliday was given a chance to test this theory in a computerized language genera-
tion project called the Penman project. He developed a grammar of English based on a
network of 81 systems each with a probability attached to the individual terms
(Halliday, Language as System and Language as Instance, p. 65). He designated each
system in the network as either equiprobable (probabilities 0.5/0.5) or skewed (0.9/0.1).
This grammar was later named the Nigel Grammar, after his son Nigel, who was the
subject for his book Learning How to Mean: Explorations in the Development of Language
(London: Edward Arnold, 1975). Describing the results of the grammar implemented
in the Penman project, Halliday notes: It was run as a random generator, without the
probabilities attached; and it produced garbage as unconstrained grammar generators
always do. But when it was run with the probabilities also being implemented, then (as
the Director of the project, Bill Mann, expressed it to me afterwards) it produced gar-
bage that now actually looked liked Englishit bore some family resemblance to possi-
ble human language (Halliday, Language as System and Language as Instance, p. 65).
56
See M.-L. Kean, Markedness: An Overview, in W. Bright (ed.), International
Encyclopedia of Linguistics (4 vols.; New York: Oxford University Press, 1992), II, pp. 390-
91, esp. p. 390: In some work one finds precise definitions of markedness; elsewhere, the
concept is taken as being antecedently well defined, or at least intuitively well understood.
57
N. Trubetzkoy, Die Aufhebung der phonologischen Gegenstze, Travaux du Cercle
Linguistique de Prague 6 (1936), pp. 29-45; repr. in J. Vachek (ed.), A Prague School Reader
in Linguistics (Bloomington: Indiana University Press, 1964), pp. 187-205. For an analy-
sis of the fundamental concepts within their larger programme, see J. Vachek, The
Linguistic School of Prague (Bloomington: Indiana University Press, 1966), esp. pp. 55-56.
58
R. Jakobson, Zur Struktur des russischen Verbums, in Charisteria G. Mathesio
(Prague: Cercle Linguistique de Prague, 1932), pp. 74-84; ET Shifters, Verbal
Categories, and the Russian Verb, in his Selected Writings. II. Word and Language (The
Hague: Mouton, 1971), pp. 130-47.
feature. Although this model is still utilized by some linguists
59
, it has
proved ineffective in a number of ways: it depends upon successful exten-
sion of phonological features to other, more abstract notions, such as case
and aspect; it is not necessarily true that every opposition can be de-
scribed in terms of the presence and absence of a given feature the way this
can be used in phonology, but there may be degrees of its appearance; this
model is based on a hierarchy of linguistic structures that proceed from
simple to complex, neglecting many of the interconnected and context-
dependent features of language; and it neglects other factors such as how
the linguistic item is distributed in the language. Thus, most linguists,
and we follow this line of reasoning, build upon the Prague concept of
markedness and use what has been called a cross-linguistic distributional
analysis first pioneered by Greenberg
60
, and developed further by many
others since then
61
. This form of markedness does not require a single
feature notation, but is able to take into account a cline of combined fac-
tors, including morphology, semantics, and, most importantly here, dis-
tribution. Distribution might at first seem to be inappropriate as a means
of determining markedness, since it does not appear at first related to
either morphology or semantics, but what might be seen as simply ran-
domness. However, Givn has developed the concept in terms of what he
calls iconicity. He believes that substantive grounds, such as varying con-
textual, socio-cultural, cognitive and communicative factors, determine
the distribution of a given linguistic item, such that The marked cate-
gory (figure) tends to be less frequent, thus cognitively more salient, than
the corresponding unmarked one (ground)
62
. Thus, one is justified in
16 Stanley E. Porter and Matthew Brook ODonnell
59
See, for example, E. Battistella, Markedness: The Evaluative Superstructure of Language
(Albany: State University of New York Press, 1990); E. Andrews, Markedness Theory: The
Union of Asymmetry and Semiosis in Language (Durham: Duke University Press, 1990), esp.
pp. 9-43, 136-39; Martn-Asensio, Foregrounding and its Relevance, esp. pp. 209-10.
60
J.H. Greenberg, Language Universals: With Special Reference to Feature Hierarchies
(The Hague: Mouton, 1966), esp. chaps. 3 and 4.
61
See, for example, B. Comrie, Aspect: An Introduction to the Study of Verbal Aspect
and Related Problems (CTL; Cambridge: Cambridge University Press, 1976), pp. 111-
22; A.M. Zwicky, On Markedness in Morphology, Die Sprache 24 (1978), pp. 129-
43; C. Bache, Verbal Aspect: A General Theory and its Application to Present-Day English
(Odense: Odense University Press, 1985), pp. 60-73, where he shows that Jakobson pro-
vides a half-way point in the discussion, as well as having difficulty in defining verbal
aspect; T. Givn, Markedness in Grammar: Distributional, Communicative and
Cognitive Correlates of Syntactic Structure, Studies in Language 15 (1991), pp. 335-70;
idem, Functionalism and Grammar (Amsterdam: Benjamins, 1995), esp. pp. 25-69.
62
Givn, Markedness in Grammar, p. 337; cf. idem, Functionalism and Grammar, p.
28. On figure and ground, see S. Wallace, Figure and Ground: The Interrelationships of
Linguistic Categories, in P.J. Hopper (ed.), Tense-Aspect: Between Semantics and Pragmatics
(Amsterdam: Benjamins, 1982), pp. 201-23; P.J. Hopper, Aspect and Foregrounding in
Discourse, in T. Givn (ed.), Syntax and Semantics. XII. Discourse and Syntax (New York:
Academic, 1979), pp. 213-41; assessed and modified in Porter, Verbal Aspect, pp. 92-93.
using the terminology of markedness on the basis of distribution, with the
more heavily marked term the one that is less frequent than the less heavily
marked term.
In summary, Hallidays work provides three theoretical constructs that
present themselves in support of the use of quantitative data in linguistic
analysis: (1) grammar as paradigmatic choice, that is, the view of lan-
guage as a network of systems, (2) language as both system and instance,
and (3) the typology of systems as either equiprobable (both of the terms
are equally likely to occur) or skewed (one of the terms is unmarked and
the other is marked). In the remainder of this paper we will utilize these
three concepts for the quantitative analysis of the Greek verbal network.
3. The Greek Verbal Network
The verbal network of the Greek of the New Testament lends itself to
paradigmatic modelling, as has been illustrated at length elsewhere
63
. The
diagram in Appendix A is a system network that contains 14 systems and
captures the semantics of the Greek verb. Networks must contain realiza-
tion statements, which show how selection of particular semantic features
(selection expressions) are translated into the formal substance of the lan-
guage itself (realization). The terms of each system of the verbal network
find realization in verbal forms
64
. The original network presented in
Verbal Aspect covered only the systems for aspectuality and aspect realized
by tense-forms, and for finiteness and attitude, realized by mood forms.
There was also the major sub-system for remoteness, realized by present,
imperfect, perfect and pluperfect indicative forms. All of this was in line
with the major thesis of the work
65
. The remaining three verbal systems
have now been added to the network: CAUSALITY, NUMBER, and PARTICI-
PATION. CAUSALITY is one of the three major systems in the Greek verbal
network, and involves a set of simultaneous choices with the other two,
leading to two further sub-systems, realized by the Greek voice system.
The first system requires choice between +active (realized by the active
voice form) and active. The choice of active is the entry condition for
the required choice of +passive (realized by the passive voice form) or
+ergative (realized by the middle voice form)
66
. The NUMBER system, rea-
lized by singular and plural forms, requires that one have chosen either
The Greek Verbal Network Viewed from a Probabilistic Standpoint 17
63
See Porter, Verbal Aspect, esp. pp. 89-90, 93-97, 109.
64
Porter, Verbal Aspect, p. 13. Cf. Fawcett, Cognitive Linguistics and Social Interaction,
pp. 50-53, 115-24; C.S. Butler, Systemic Linguistics: Theory and Applications (London:
Batsford, 1985), pp. 59-62.
65
Porter, Verbal Aspect, p. 109.
66
The concept of ergativity and the middle voice is being developed in Porter, Voice
in the Greek of the New Testament.
the +finite or the +factively presuppositional features as a singular entry
condition (i.e. factively presuppositional, realized by the infinitive, does
not lead to the NUMBER system). The NUMBER system is realized by sin-
gular and plural forms. Lastly, the PARTICIPATION system, realized by per-
sonal reference, has the entry condition of +finite, and is a system on the
same level as attitude. The participation system has two sub-systems, the
first requires choice of included (realized by the third person form) or
+included, and the second requiring choice of +direct (realized by the first
person form) or direct (realized by the second person form). We have
also taken the opportunity to refine the labelling system of the entire net-
work. The names of the terms for each system are now consistently stated
as English adjectives, which describe their semantic feature; and the
names of the systems are nouns (eg. ASPECTUALITY and REMOTENESS). For
the purpose of this study, the systems of the network have been numbered;
however, there is no significance in the number which has been assigned
to each system.
The network chart in Appendix A also contains predictions for each
system as to whether its terms are equiprobable (E) or skewed (S). These
follow the typology of systems suggested by Halliday, and are based upon
Porters previous study as indicated in his Verbal Aspect and his Idioms of
the Greek New Testament, where statements regarding markedness have
been made on the basis of a variety of factors, including elementary analy-
sis of distribution, as well as morphology and semantics
67
. We have then
18 Stanley E. Porter and Matthew Brook ODonnell
67
It is worth noting that no major work on what we define as verbal aspect has ap-
peared since Porters Verbal Aspect that has developed this method of analysis, so far as we
know, except for Y. Duhoux, Le verbe grec ancien: lements de morphologie et de syntaxe
historiques (BCILL, 61; Louvain-la-Neuve: Peeters, 1992), esp. pp. 497-505; idem, La
dynamique des choix aspectuels en grec ancien, CILL 18.2-4 (1992), pp. 45-66; idem,
tudes sur laspect verbal en grec ancient, 1: Prsentation dune mthode, BSL 90
(1995), pp. 241-99, although he deals with a different time period, and does not treat
the interactive verbal systems as we do. Works to consult that have appeared in the last
ten years include the following (there are, of course, other works that treat verbal aspect,
but they are often speaking in terms of lexis or what is better categorized as Aktionsart):
D. Cohen, Laspect verbal (Linguistique nouvelle; Paris: Presses Universitaires de France,
1989); H.B. Thelin (ed.), Verbal Aspect in Discourse (Amsterdam: Benjamins, 1990); S.
Fleischman, Tense and Narrativity: From Medieval Performance to Modern Fiction
(Croom Helm Romance Linguistics Series; London: Routledge, 1990); R.I. Binnick,
Time and the Verb: A Guide to Tense and Aspect (New York: Oxford University Press,
1991), C.S. Smith, The Parameters of Aspect (Studies in Linguistics and Philosophy, 43;
Dordrecht: Kluwer, 1991); H.J. Verkuyl, A Theory of Aspectuality: The Interaction be-
tween Temporal and Atemporal Structure (Cambridge Studies in Linguistics, 64;
Cambridge: Cambridge University Press, 1993); idem, Aspectual Issues: Studies on Time
and Quantity (Center for the Study of Language and Information Publication Lecture
Notes, 98; Stanford, CA: CSLI Publications, 1999); W. Klein, Time in Language
(Germanic Linguistics; London: Routledge, 1994); C. Bache, H. Basboell, and C.-E.
Lindberg (eds.), Tense, Aspect and Action: Empirical and Theoretical Contributions to
Language Typology (Empirical Approaches to Language Typology, 12; Berlin: Mouton de
Gruyter, 1994); C. Vet and C. Vetters (eds.), Tense and Aspect in Discourse (Trends in
tested these predictions by actually counting the occurrences of each of
the forms that realize the features of the systems. For instance, we have
predicted that, in the REMOTENESS system (system 4), there will be a skewed
probability that the form that realizes the semantic feature of remote will
be far more frequent than the form that realizes the semantic feature of
+remote. To test this prediction, we have counted the number of present
and perfect indicatives (remote) and the number of imperfect and
pluperfect indicatives (+remote) in the Greek New Testament. From our
calculations these figures are 11138 and 1245 occurrences, respectively.
This results in probabilities of: 11138/12383 = 0.899 for remote and
1245/12383 = 0.101 for +remote
68
. The prediction that the form that
realized the semantic feature of +remote (imperfect and pluperfect tense-
forms) would be more frequent than that realizing the semantic feature of
remote (present and perfect tense-forms) has been proved correct, and
the semantic feature of +remote, realized by the imperfect and pluperfect
tense-forms, is the marked member of the opposition.
Hallidays typology of probabilities works with the simple set of calcu-
lations in which distributions are either equiprobable (0.5/0.5) or skewed
(0.9/0.1). Before beginning the calculations, it was optimistic to think
that every opposition would fall to either of these results. However, it
soon became obvious, both after further reflection and after a few of the
The Greek Verbal Network Viewed from a Probabilistic Standpoint 19
Linguistics, 75; Berlin: Mouton de Gruyter, 1994); C. Bache, The Study of Aspect, Tense
and Action: Towards a Theory of the Semantics of Grammatical Categories (Frankfurt: Peter
Lang, 1995); R. Bartsch, Situations, Tense, and Aspect: Dynamic Discourse Ontology and
the Semantic Flexibility of Temporal System in German and English (Groningen-
Amsterdam Studies in Semantics; Berlin: Mouton de Gruyter, 1995); Y. Tobin, Aspect in
the English Verb (LLL; London: Longman, 1994); C.M.J. Sicking and P. Stork, Two
Studies in the Semantics of the Verb in Classical Greek (Mnemosyne, 160; Leiden: Brill,
1996), esp. pp. 3-118; N. Bermel, Context and the Lexicon in the Development of Russian
Aspect (University of California Publications in Linguistics, 129; Berkeley: University of
California Press, 1997); M.J.A. Olsen, A Semantic and Pragmatic Model of Lexical and
Grammatical Aspect (New York: Garland, 1997); L.A. Michaelis, Aspectual Grammar and
Past-Time Reference (Routledge Studies in Germanic Linguistics; London: Routledge,
1998); R.J. Decker, Temporal Deixis of the Greek Verb in the Gospel of Mark with Reference
to Verbal Aspect (SBG, 10; New York: Peter Lang, 2001).
68
These figures have been calculated from the Analytical Greek New Testament prepared
by Timothy and Barbara Friberg. It is a grammatically annotated text of the UBS 4th
Edition Greek New Testament. The figures were calculated by a number of specially writ-
ten scripts and may show certain discrepancies from searches carried out using search
programs such as acCordance and BibleWorks. Given the scope of this article, we has
discounted all occurrences of aspectually vague verbs (... /.. . [and prefixed forms],
j c. [prefixed forms], -.. c. and j.; on aspectual vagueness, see Porter, Verbal
Aspect, pp. 442-47). We have checked many of our figures using these programs and
though there is slight variation for some counts this does not affect the systemic classifi-
cation used (skewed/equiprobable) for any of the systems studied. The variations are par-
tially due to the searching method used, but primarily a result of the different machine-
readable texts and annotation schemes they utilize. For a discussion of existing machine-
readable texts of the Greek New Testament with warnings of potential pitfalls of their
use, see ODonnell, The Use of Annotated Corpora, pp. 93-95.
calculations were done, that such was not entirely realistic, language being
what it is. We will draw attention to any systems that reveal distributions
that do not fall within the parameters that Halliday has outlined. For the
sake of our discussion we consider any ratio of greater than 0.7/0.3 to be
skewed, and any ratio less than that to be equiprobable. In other words,
for a distribution to be skewed, it must be a significantly higher distribu-
tion than simply a slight numerical advantage of one term over another.
There is, as a result, a significant possibility of a distribution that is close
to this dividing line, rendering judgments as to markedness on the basis
of distribution alone tentative.
4. Predicted and Actual Results
a. Predicted Results
The chart in Appendix A shows the revised system network, with the
predictions for each system indicated. Here we summarize how we ar-
rived at these predictions, drawing upon information already discussed in
Porters Verbal Aspect and Idioms of the Greek New Testament. System 1:
In the ASPECTUALITY system, the +expectational/+aspectual opposition
is realized in the formal choices between future forms and other verbal
tense-forms (e.g. aorist, present and perfect). The future form is mor-
phologically restricted in its forms, but is morphologically and seman-
tically, as well as syntactically (e.g. conditional statements), closely rela-
ted to non-indicative forms, as well as having syntactical and functional
relations to the indicative forms. Rather than place it in the category of
the already-established indicative or non-indicative forms, that is, as a
tense-form or as a mood, or even as an aspect, it is better considered a
part of the Greek verbal system but is not fully aspectual (that is, no
paradigmatic choice is offered). On this basis, we predicted that the
future form, realizing the semantic feature of +expectational, would
have a skewed distribution in relation to the +aspectual forms
69
. System
2: In the ASPECT1 system, the perfective opposition, realized in aorist
and non-aorist (i.e. present [imperfect] and perfect [pluperfect]) tense-
forms, was predicted to be equiprobable, on the basis of distributional
figures and other criteria, such as semantics, already known from pre-
vious study
70
. System 3: In the ASPECT2 system, the +imperfective/+sta-
tive opposition, realized in present (imperfect) and perfect (pluperfect)
tense-forms, was predicted to be skewed, with the +imperfective term to
20 Stanley E. Porter and Matthew Brook ODonnell
69
Porter, Verbal Aspect, pp. 94-95, 97, 404-16.
70
Porter, Verbal Aspect, pp. 89-90 (where the opposition is labelled equipollent), pp.
178-81.
be distributionally more frequent, on the basis of distributional, mor-
phological, implicational and semantic criteria
71
. System 4: In the
REMOTENESS system (see above), the remote opposition, realized in
imperfect/pluperfect or present/perfect tense-forms, was predicted to be
skewed, with the +remote term to be distributionally more frequent, on
the basis of distributional, morphological, implicational and semantic
criteria. System 5: In the CAUSALITY system, we were unable to decide
whether the active opposition, realized in the active and non-active
voice verb forms, would be equiprobable or skewed. We were aware of
other distributional figures
72
, but considering a variety of factors, such
as origins of the forms and semantic relations, we were unsure whether
what was clearly a larger number of active voice forms was large enough
to indicate skewed probabilities
73
. System 6: The +passive/+ergative
opposition, realized in the passive and middle voice verb forms, was at
first predicted to be skewed, but recent work on the concept of ergati-
vity and voice for another project has caused us to believe that this
system may be closer to equiprobable than we first thought
74
. System
7: In the FINITENESS system, with the finite opposition, realized in ver-
bal forms limited by person, it was not possible to predict whether the
terms would be equiprobable or skewed
75
. System 8: The factively pre-
suppositional opposition, realized in the participle and infinitive verbal
forms, was predicted to be skewed, on the basis of rudimentary distri-
butional analysis
76
. What we predicted for the skewed probabilities
runs counter to the semantic markedness and labelling conventions.
System 9: In the NUMBER system, the singular opposition, realized in
The Greek Verbal Network Viewed from a Probabilistic Standpoint 21
71
Porter, Verbal Aspect, pp. 89-90, 245-51.
72
See Reed, Discourse Analysis of Philippians, p. 115.
73
Porter, Verbal Aspect, pp. 10-11; idem, Idioms of the Greek New Testament, pp. 62-70.
74
See Reed, Discourse Analysis of Philippians, p. 115. As noted above, it is hoped that
this work will result in a monograph on voice by Porter. On the concept of ergativity as
it is being used here, see M.A.K. Halliday, An Introduction to Functional Grammar
(London: Edward Arnold, 2nd edn, 1994), pp. 161-75; K. Davidse,
Transitivity/Ergativity: The Janus-Headed Grammar of Actions and Events, in M.
Davies and L. Ravelli (eds.), Advances in Systemic Linguistics: Recent Theory and Practice
(London: Pinter, 1992), pp. 105-35; M.B. ODonnell, Some New Testament Words for
Resurrection and the Company They Keep, in S.E. Porter, M.A. Hayes and D. Tombs
(eds.), Resurrection (JSNTSup, 186; RILP, 5; Sheffield: Sheffield Academic Press, 1999),
pp. 136-63, esp. pp. 154-61; cf. P.J. Hopper and S.A. Thompson, Transitivity in
Grammar and Discourse, Language 56 (1980), pp. 251-99.
75
Porter, Verbal Aspect, p. 94.
76
See Porter, Verbal Aspect, pp. 390-91, with reference especially to D. Lightfoot,
Natural Logic and the Greek Moods: The Nature of the Subjunctive and Optative in
Classical Greek (Janua Linguarum, Series Practica, 230; The Hague: Mouton, 1975); G.
Horrocks, Review of Lightfoot, Natural Logic, Linguistics 185 (1977), pp. 68-83; D.D.
Schmidt, The Study of Hellenistic Greek Grammar in the Light of Contemporary
Linguistics, Perspectives on Religious Studies 11 (1984), pp. 27-38; G. de Boel, Aspekt,
Aktionsart und Transitivitt, Indogermanische Forschungen 92 (1987), pp. 33-57.
the singular and plural forms, was predicted to be equiprobable, espe-
cially since the Greek of the New Testament no longer has a dual num-
ber. System 10: In the PARTICIPATION system
77
, the included opposi-
tion, realized in the non-third person and the third person forms, was
predicted to be equiprobable. System 11: The direct opposition, real-
ized in the first and second person forms, was predicted to be equiproba-
ble. For both systems 10 and 11, prior to this study, we had only intui-
tions on which to base these predictions, but equiprobability seemed
likely on the basis of the mix of narrative and expositional material in the
New Testament. System 12: In the attitude system, the assertive oppo-
sition, realized in the indicative and non-indicative (excluding participle
and infinitive) mood forms, was predicted to be skewed, on the basis of
what is already known of frequency of the indicative verbal form, as well
as morphological, implicational and semantic criteria
78
. System 13: The
+projective/+directive opposition, realized in the subjunctive/optative or
imperative mood forms, was predicted to be equiprobable, on the basis
of rudimentary distributional figures, as well as morphological, implica-
tional and semantic criteria
79
. System 14: The contingent opposition,
realized in the optative and subjunctive mood forms, was predicted to be
skewed, on the basis of well-known distributional figures regarding the
disappearance of the optative in Hellenistic Greek, including that of the
New Testament, as well as morphological, implicational and semantic
criteria
80
.
b. Actual Results
The chart in Appendix A also shows the Greek verbal network with
calculated probabilities from the Greek New Testament attached to each
of the terms in the systems. A summary of the relationship between the
predictions offered and the results is that virtually all of the predictions
were warranted, on the basis of the analysis offered in Verbal Aspect, taking
into consideration not only rudimentary distributional calculations, but
morphological, implicational and semantic criteria as well. Systems 2, 9,
10, 11, and 13 were predicted to be equiprobable, and all of them fell
within the parameters set above for equiprobable distribution. In systems
9 and 10, the distributions were closer to the boundary than in the others
22 Stanley E. Porter and Matthew Brook ODonnell
77
See Porter, Idioms of the Greek New Testament, pp. 76-79, with reference to Lyons,
Introduction to Theoretical Linguistics, pp. 276-78.
78
Porter, Verbal Aspect, pp. 163-77; idem, Idioms of the Greek New Testament, pp. 50-61.
79
Porter, Verbal Aspect, p. 181, as well as pp. 167-77, 321-62; idem, Idioms of the
Greek New Testament, pp. 52-61; cf. Reed, Discourse Analysis of Philippians, p. 115.
80
Porter, Verbal Aspect, pp. 181, 321-35; idem, Idioms of the Greek New Testament, pp.
56-61.
(being around 0.65 or 0.66/0.35 or 0.34), but each was still within the
range. The others were very similar in their distribution (system 2:
0.48/0.52, system 11: 0.55/0.45 and system 13: 0.54/0.46). Systems 1, 3,
4, 6, 8, 12, and 14 were predicted to be skewed, and all of them except
system 6 fell within the parameters set above for skewed distribution
81
.
Systems 3 and 4 (0.87/0.13 and 0.90/0.10, respectively) were more skewed
than systems 8 and 12 (0.75/0.25 and 0.80/0.20). System 1 was even
more skewed at 0.94/0.06, but system 14 was the most highly skewed of
all at 0.97/0.03. Of the two systems that were predicted as uncertain,
systems 5 and 7, system 5 was skewed, with 0.72/0.28, although not by
a large margin. System 7 at 0.66/0.34 was similar in distribution to
systems 9 and 10, and so should probably be considered equiprobable in
distribution. In other words, virtually all of the predictions based upon
Porters Verbal Aspect and Idioms of the Greek New Testament, formulated
on the basis of analysis of morphology, implicational markedness, rudi-
mentary distributional figures, and semantic features, have been shown to
be correct in so far as distributional patterns over the whole of the Greek
New Testament are concerned.
These predictions are also supported by distributional probabilities
when individual corpora within the corpus of the Greek New Testament
are considered as well. In all but a few instances, the probabilities for the
whole of the New Testament also apply to the sections within it. We
have performed the same statistical analysis for the Synoptic Gospels
and Acts, the Pauline Letters, the Johannine Writings, and the other
books of the New Testament as we have for the whole of the New
Testament. In only perhaps five instances does one of these corpora
depart from the probabilities for the New Testament as a whole. As
Appendix B indicates in system 2, ASPECT1, the probability for the
whole New Testament is equiprobable that perfective would be selec-
ted (0.48/0.52). This ratio is very similar for all of the sub-corpora exam-
ined, apart from the Pauline Letters where a distribution of 0.36/0.64
is still within the range for equiprobable systems, but not as nearly equal
The Greek Verbal Network Viewed from a Probabilistic Standpoint 23
81
The features of system 6, CAUSALITY sub-system (CAUSALITY2), are realized by the
middle and passive voices in Greek. Statistical analysis of this system is hindered by the
fact that the middle and passive forms of the present, imperfect, perfect and pluperfect
tense-forms are formally/morphologically ambiguous. Machine-readable texts of the
New Testament differ in the manner in which they select to annotate these forms. The
Friberg text has a complex classification scheme that allows for active (a), middle (m),
passive (p), middle or passive (e), middle deponent (d), passive deponent (o) or middle
or passive deponent (n). The GRAMCORD text in contrast classifies an ambiguous
form as either middle (m) or passive (p) (though a recent revision has included both
options in cases of ambiguity). Given the current state of textual annotation of the New
Testament we decided to count all unambiguous middle and passive forms (aorist and
future) and then add half the frequency of all ambiguous forms to each of these totals.
We realize that this means of calculation may affect the results.
as the ratios of the other sub-corpora. The preference for perfective
forms in these letters might be explained by the exhortative function of
a letter. Further research might examine the distribution of various
aspectual forms according to the letter form
82
. The FINITENESS system
(system 7) exhibits a weak equiprobable distribution for the entire New
Testament (0.66/0.34). The Johannine Writings, at 0.81/0.19, a skewed
distribution, is the only examined sub-corpus in this system to vary great-
ly from this ratio. These writings have noticeably fewer finite forms
(participles and infinitives) than +finite forms. Again, more detailed
analysis is required to explain this variance, but it might be a function
of Johnannine style. In system 10, PARTICIPATION, the probability for
the whole of the New Testament is equiprobable that included would
be selected (0.65/0.35) but in the Synoptic Gospels and Acts, it is
slightly more probable that included rather than +included would be
selected (0.72/0.28), a slightly skewed distribution. This is probably
best explained by the narrative nature of this material and its subse-
quent dependence upon third person forms. In contrast, the Pauline
Letters exhibit a clear equiprobable distribution for this system
(0.48/0.52). Again, this may be explained by the genre of the mate-
rialletters have a relatively higher use of first and second person forms
than do narratives. Similar comments can be made regarding the use of
the first and second person as indicated by system 11, DIRECTION. In
the whole New Testament, the system is equiprobable (0.55/0.45), but
in the Synoptic Gospels and Acts the ratio is closer to the divide between
skewed and equiprobable at 0.65/0.35, although still within the
equiprobable distribution. In system 13, an ATTITUDE sub-system (NON-
ASSERTION), the system is of equiprobable distribution in the whole of
the New Testament (a relatively equal number of imperative and sub-
junctive/optative forms) but ambiguous for the Johannine writings
(0.70/0.30). These are the only possible departures within various sub-
corpora of the New Testament from the probabilities for the whole of
the New Testament. In fact, in only two (systems 7 and 10) is there a
clear changing of category of distribution (we discount system 5, CAU-
SALITY1, since the system and all of its sub-corpora, except for the
Johannine Writings, are very close in their distributions, even if they fall
on either side of the dividing line between skewed and equiprobable).
In the light of the probabilities of the systems involved, the departures
noted above seem understandable, and do not jeopardize the results in
any way.
24 Stanley E. Porter and Matthew Brook ODonnell
82
Preliminary attempts along these lines have been made by Reed, Discourse Analysis
of Philippians; and S.E. Porter and M.B. ODonnell, Semantics and Patterns of
Argumentation in the Book of Romans: Definitions, Proposals, Data and Experiments,
in S.E. Porter (ed.), Diglossia and Other Topics in New Testament Linguistics (JSNTSup,
193; SNTG, 6; Sheffield: Sheffield Academic Press, 2000, pp. 154-204).
5. Implications of Results of Distributional Analysis
There is more that can be said on the basis of these calculations, how-
ever. These comments will be divided into two parts. The first part is con-
cerned with the findings regarding several individual systems. The second
is concerned with more extended patterns of distribution, with particular
reference to their implications regarding verbal aspect.
a. Individual Systems
As a result of the above analysis, several questions regarding certain
individual systems should be addressed. The first is the question of how
the semantic feature labelling is possibly affected by the distributional pat-
terns. This issue appears in several different forms. The first is where one
has an equiprobable distribution, but positive and negative terms are uti-
lized to indicate the semantic features, such as in system 2, where there is
an equiprobable distribution but the terms +perfective and perfective are
used. This labelling would appear to indicate a skewed distribution, since
the use of + feature notation may seem to imply markedness. The same
question arises for systems 7, 10, and 11. Systems 1 and 3 are skewed in
their distribution, yet they have two positive semantic features indicated,
such as in system 1 with +expectational and +aspectual. There is also the
question of how one labels a system where one cannot clearly establish
equiprobable or skewed distribution. These questions of labelling are ones
that have been of interest to systemic linguists from the advent of this gra-
phic systemic display, since the convention seems to warrant concise and
consistent labelling conventions in the light of their use as semantic fea-
ture notation for the forms grammaticalized by these features
83
. There has
been no resolution to the difficulties regarding the labelling of systems,
with some systemicists using systemic networks for purely semantic cate-
gories, others for formal categories, and others for a combination. The
major criticism of networks seems to revolve around instances where the
labelling terms are not either formally or implicationally motivated
84
.
Our network has both formally and implicationally motivated terms, and
moves from the broadest to the most delicate semantic features, with the
more delicate features implicated by the previous entry conditions of the
network, and specific forms that serve as the unique realizations of the
semantic features. So long as this double condition is realized, as ours is,
The Greek Verbal Network Viewed from a Probabilistic Standpoint 25
83
Besides the sources mentioned in Porter, Verbal Aspect, pp. 8-11, see Butler,
Systemic Linguistics, pp. 40-45; R. Melrose, Systemic Linguistics and the
Communicative Linguistics Syllabus, in Fawcett and Young (eds.), New Developments,
II, pp. 78-93.
84
See J.R. Martin, The Meaning of Features in Systemic Linguistics, in Halliday
and Fawcett (eds.), New Developments, I, pp. 14-40; P.W. Davies, Modern Theories of
Language (Englewood Cliffs, NJ: Prentice-Hall, 1973), p. 298.
the positive or negative features of the network do not seem to be of
importance, and such semantic labelling is found in a variety of networks.
A second major question to ask is with regard to the implications of
the distribution of system 2 and the aorist tense-form. The question is
whether the equiprobable distribution indicates that the aorist is not
unmarked. There has probably been too much loose use of language with
regard to the so-called unmarked character of the aorist. The equiproba-
ble distribution indicates, as predicted, that the forms grammaticalizing
the +perfective semantic feature are not in privative opposition to those
that are perfective, but that they are equipollent. In this sense, the aorist
tense-form is not unmarked. However, on the basis of a number of other
factors, such as morphology, implication and semantics, as well as the pat-
terns of distribution when one differentiates the aorist from the present
and from the perfect tense-forms, it is accurate to say that the aorist is
probably the least heavily marked of those forms
85
.
b. Patterns of Distribution
The probabilistic studies that have been performed also allow for a test-
ing of several particular dimensions of verbal aspectual usage. In Verbal
Aspect, Porter defined verbal aspect in the following way: Greek verbal
aspect is a synthetic semantic category (realized in the forms of verbs)
used of meaningful oppositions in a network of tense systems to gram-
maticalize the authors reasoned subjective choice of conception of a pro-
cess
86
. This definition was refined further in Idioms of the Greek New
Testament as follows: verbal aspect is defined as a semantic (meaning)
category by which a speaker or writer grammaticalizes (i.e. represents a
meaning by choice of a word-form) a perspective on an action by the
selection of a particular tense-form in the verbal system
87
. There have
been two major objections raised against these definitions
88
. One of these
revolves around the issue of lexis
89
. This response takes several different
26 Stanley E. Porter and Matthew Brook ODonnell
85
See Porter, Verbal Aspect, pp. 89-90.
86
Porter, Verbal Aspect, p. 88.
87
Porter, Idioms of the Greek New Testament, pp. 20-21.
88
Objections based upon simplistic or traditional temporal conceptions of the verb
are not considered here, since they have already been adequately dealt with in Porter,
Verbal Aspect, passim.
89
The major proponents of this position with regard to the Greek of the New
Testament are J. Mateos, El Aspecto Verbal en el Nuevo Testamento (Estudios de Nuevo
Testamento, 1; Madrid: Ediciones Cristiandad, 1977); cf. idem, Metodo de Analisis
Semantico Aplicado al Griego del Nuevo Testamento (Estudios de Filologa
Neotestamentaria, 1; Crdoba: Ediciones el Almendro, 1989); and S.M. Baugh, An
Introduction to Greek Verbal Aspect in the Non-Indicative Moods (unpublished
manuscript; Westminster Theological Seminary in California, 1995). These categories
are not absolute, and several of those listed here could be placed in the next category,
with some of those below placed here as well.
forms, but essentially argues that choice of verbal aspect is to a large
extent, if not entirely, determined by the choice of lexical item. Therefore,
the aspectual systems in the Greek verbal network cannot have the status
or position in the Greek language that they are given in Porters model,
but aspectual choice comes only as a latterly or even terminal choice dic-
tated by previous lexical choice. The probabilistic results regarding the
various corpora as opposed to the whole of the New Testament go some
of the way to showing that lexis, at least as it would be influenced by indi-
vidual writers and genre, does not have a significant effect upon verbal
aspect. This is especially the case since none of the aspectual systems showed
significant variance in its probabilities between the sub-corpora and
the whole of the New Testament. ODonnell has also shown in a more
detailed study that such holds true
90
.
The second response sees verbal aspect as dependent upon a variety of
other grammatical features, such as voice and mood
91
. The contention of
Verbal Aspect was that the Greek verbal network has aspect as one of its
major system components, and that verbal aspect as a morphologically
based semantic system functions largely independently of these other fac-
tors. In the rest of this paper, we have tested this theory by analyzing a
number of instances where aspectual systems in the Greek verbal network
interact with other verbal systems.
As the tables discussed below indicate, aspectual choice seems for the
most part to be independent of any other set of grammatical choices
within the Greek verbal network. Not all of the possible paths through
the systemic network have been analyzed, but those that haveall of
them concerned with various aspect related systemstend to confirm
this analysis. For each analysis, we first define the parameters of the inter-
active systems (they are not necessarily simultaneous choices in the net-
work, but are choices that must be made to realize the various semantic
features concerned), and then comment upon the possible influences of
one on the other, paying particular attention to how choice of verbal
aspect may or may not be affected by the other interactive system
92
.
Distributional statistics for these individual systems in isolation are also
The Greek Verbal Network Viewed from a Probabilistic Standpoint 27
90
See M.B. ODonnell, Aspect and Lexis: An Empirical Approach (forthcoming).
91
See B.M. Fanning, Verbal Aspect in New Testament Greek (OTM; Oxford:
Clarendon Press, 1990), esp. pp. 126-96; cf. M. Silva, God, Language and Scripture:
Reading the Bible in the Light of General Linguistics (Foundations of Contemporary
Interpretation, 4; Grand Rapids: Zondervan, 1990), pp. 111-18; idem, A Response to
Fanning and Porter on Verbal Aspect, in S.E. Porter and D.A. Carson (eds.), Biblical
Greek Language and Linguistics: Open Questions in Current Research (JSNTSup, 80;
SNTG, 1; Sheffield: JSOT Press, 1993), pp. 74-82 (also found in WTJ 54 [1992], pp.
179-83); idem, Explorations in Exegetical Method: Galatians as a Test Case (Grand Rapids:
Baker Book House, 1996), pp. 68-79, who especially in his later work entertains Baughs
ideas (see n. 89 above).
92
See Halliday, Language as System and Language as Instance, pp. 72-76.
provided. These statistics sometimes differ from those of the individual
systems in the Greek verbal network as a whole, since other systems that
affect the statistics have been eliminated. As mentioned above, the results
are essentially confirmatory that choice of verbal aspect is not affected by
competing grammatical choices; however, there are several other interes-
ting statistical results that come to the fore, virtually all of them in con-
firmation of previous research found in Porters Verbal Aspect. Each table
is presented with the two systems that are being tested. The system at the
top is the one seen to influence the system at the side. For each set, each
system is tested in relation to the other, to determine whether the proba-
bilistic distribution for a given system is or is not affected by the other
system. We determine that a system is not affected if the distribution
still falls within its distributional type, that is, as either equiprobable or
skewed, even if the exact figures vary. A system is affected if its distribution
changes types. Those that fall on the border of the two types are also
noted.
1. FINITENESS (7) and ASPECTUALITY (1)
Realization Statements:
+finite +expectational >> Future Indicative
finite +expectational >> Future Participle & Infinitive
+finite +aspectual >> Non-Future Indicative, Subjunctive, Imperative & Optative
finite +aspectual >> Non-Future Participle & Infinitive
The first set of examples concerns the relation of the FINITENESS
system and the ASPECTUALITY system. This set of systemic choices con-
cerns the options of finite and +expectational/+aspectual (systems 7 and
1). Choice of finite does not affect the skewed distribution of +expecta-
tional/+aspectual (0.08/0.92 and 0.01/0.99, respectively), in conformity
with the distribution for this system on its own (system 1). Choice of
+aspectual results in an equiprobable distribution regarding finite
28 Stanley E. Porter and Matthew Brook ODonnell
+expectational +aspectual
+finite ........................ 1421 0.99 15450 0.64 16871 0.66
finite ........................ 13 0.01 8537 0.36 8550 0.34
+finite finite
+expectational ............ 1421 0.08 13 0.01 1434 0.06
+aspectual .................. 15450 0.92 8537 0.99 23987 0.94
(0.64/0.36), consistent with the distribution of system 7. However,
choice of +expectational results in a highly skewed distributional ratio
toward +finite (0.99/0.01), realized in the future finite form (the so-called
future indicative), with the future finite form distributionally the
unmarked term and the non-finite forms (participle and infinitive) marked.
Thus, selection of +expectational apparently affects selection of finite,
and the forms that realize these selection statements. This is in keeping
with what has already been discussed regarding the limited range of forms
of the future, as well as its awkward place within the Greek verbal net-
work
93
. In any case, the ASPECTUALITY system is unaffected by the FI-
NITENESS system, as confirmed by the bottom two rows of the table.
2. FINITENESS (7) and ASPECT1 (2)
Realization Statements:
+finite +perfective >> Aorist Indicative, Subjunctive, Imperative & Optative
finite +perfective >> Aorist Participle & Infinitive
+finite perfective >> Present, Imperfect, Perfect, Pluperfect Indicative, Present,
Perfect Subjunctive, Imperative & Optative
finite perfective >> Present, Perfect Participle & Infinitive
The second set of examples concerns the relation of the FINITENESS
system and the ASPECT1 system. This set of systemic choices concerns the
options of finite and perfective (systems 7 and 2). Choice of finite
does not affect the equiprobable distribution of perfective. However,
there are some slight variations in the distributional statistics that might
be worth noting, in the light of discussion above. One is that the distri-
bution changes for that of system 2 (perfective 0.48/0.52) to perfec-
tive 0.52/0.48 when +finite is also selected. In other words, there is a
slightly larger frequency of +perfective +finite forms realized by aorist
indicative forms as opposed to non-aorist finite forms, than there is overall
The Greek Verbal Network Viewed from a Probabilistic Standpoint 29
93
See Porter, Verbal Aspect, pp. 94-95, 97, 409-10, 417-19.
+perfective perfective
+finite ........................ 8078 0.70 7372 0.60 15450 0.64
finite ........................ 3526 0.30 5011 0.40 8537 0.36
+finite finite
+perfective ................. 8078 0.52 3526 0.41 11604 0.48
perfective ................. 7372 0.48 5011 0.59 12383 0.52
for aorist forms as opposed to non-aorist forms. Another variation is that,
in the distribution of forms grammaticalizing the feature finite, there
is an equiprobable distribution (0.41/0.59) that favours the forms
that grammaticalize perfective semantic features. Choice of perfective
has distributions that further support these tendencies. Choice of per-
fective results in an equiprobable distribution for finite (0.60/0.40),
similar to that of system 2. However, choice of +perfective produces a dis-
tribution of finite at 0.70/0.30, a figure right on the border between
equiprobable and skewed distribution, rather than the clearly equiproba-
ble distribution of system 7. The tendency is to see a relation here between
+perfective and +finite. This would be somewhat confirmatory of the
view of the aorist indicative as the so-called default tense-form in the
indicative mood
94
, as well as in the non-indicative moods
95
. This is con-
sistent with analysis of the Greek verbal network arrived at through non-
distributional means. Whatever one concludes regarding these slight
variations in distribution, however, the ASPECT1 system is unaffected by
the FINITENESS system.
3. FINITENESS (7) and ASPECT2 (3)
Realization Statements:
+finite +imperfective >> Present & Imperfect Indicative, Present Subjunctive,
Imperative & Optative
finite +imperfective >> Present Participle & Infinitive
+finite +stative >> Perfect & Pluperfect Indicative, Perfect Subjunctive, Imperative
& Optative
finite +stative >> Perfect Participle & Infinitive
30 Stanley E. Porter and Matthew Brook ODonnell
94
See Porter, Verbal Aspect, pp. 102-107; idem, Idioms of the Greek New Testament, pp.
21-24, 50-52.
95
See Porter, Verbal Aspect, pp. 163-77; idem, Idioms of the Greek New Testament, pp.
52-61.
+imperfective +stative
+finite ........................ 6436 0.60 936 0.56 7372 0.60
finite ........................ 4289 0.40 722 0.44 5011 0.40
+finite finite
+imperfective ............. 6436 0.87 4289 0.86 10725 0.87
+stative ...................... 936 0.13 722 0.14 1658 0.13
The third set of examples concerns the relation of the FINITENESS system
and ASPECT2 system. This set of systemic choices concerns the options of
finite and +imperfective/+stative (systems 7 and 3). The results are simple
to summarize here. Choice of finite does not affect the skewed distribu-
tion of +imperfective/+stative (0.87/0.13 and 0.86/0.14, respectively), and
choice of +imperfective or +stative does not affect the equiprobable distri-
bution of finite (0.60/0.40 and 0.56/0.44, respectively). Similarly, these
results are consonant with the distributional probabilities of system 3.
4. ATTITUDE (12) and ASPECT1 (2)
Realization Statements:
+assertive +perfective >> Aorist Indicative
assertive +perfective >> Aorist Subjunctive, Imperative & Optative
+assertive perfective >> Present, Imperfect, Perfect, Pluperfect Indicative
assertive perfective >> Present, Perfect Subjunctive, Imperative & Optative
The fourth set of examples concerns the relation of the ATTITUDE
system and ASPECT1 system. This set of systemic choices concerns the
options of assertive and perfective (systems 12 and 2). Choice of asser-
tive results in an equiprobable distribution of perfective, as is found in
system 2. Choice of perfective is skewed toward +assertive over asserti-
ve (0.83/0.17), consistent with the distribution of assertive in system 12.
Choice of +perfective is also skewed toward +assertive over asser-
tive, although the distribution is closer to the border between equiprobable
and skewed (0.73/0.27). This result is consistent with that noted above,
and already described in Verbal Aspect, in which there is a tendency
toward skewed distribution, confirming the markedness of the forms that
grammaticalize the semantic features of perfective (present and perfect
tense-forms) and assertive (non-indicative mood forms)
96
. The ASPECT1
system, however, is not affected by the ATTITUDE system.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 31
96
See Porter, Verbal Aspect, pp. 97, 181, 323-24, 336-43, 360-61.
+perfective perfective
+assertive .................... 5879 0.73 6145 0.83 12024 0.78
assertive .................... 2199 0.27 1227 0.17 3433 0.22
+assertive assertive
+perfective .................. 5879 0.49 2199 0.64 8078 0.52
perfective .................. 6145 0.51 1227 0.36 7372 0.48
5. ATTITUDE (12) and ASPECT2 (3)
Realization Statements:
+assertive +imperfective >> Present & Imperfect Indicative
assertive +imperfective >> Present Subjunctive, Imperative & Optative
+assertive +stative >> Perfect & Pluperfect Indicative
assertive +stative >> Perfect Subjunctive, Imperative & Optative
The fifth set of examples concerns the relation of the ATTITUDE system
and ASPECT2 system. This set of systemic choices concerns the options of
assertive and +imperfective/+stative (systems 12 and 3). The results of
this distributional analysis simply confirm what has been noted above, and
has already been discussed in Verbal Aspect. Choice of assertive is skewed
toward +imperfective over +stative, consistent with the distributional pro-
babilities of system 3, although to an even higher degree. Similarly, choice
of +imperfective is skewed toward the +assertive over assertive
(0.81/0.19), and choice of +stative is also skewedalthough even more
heavilytoward +assertive (0.99/0.01). The skewed statistics concerning
the +stative feature are commensurate with the very few perfect non-indi-
cative forms (assertive) to be found in the Greek New Testament
97
.
Again, the ASPECT2 system is not affected by the ATTITUDE system.
6. ATTITUDE subsystem (NON-ASSERTION) (13) and ASPECT1 (2)
Realization Statements:
+projective +perfective >> Aorist Subjunctive, Optative
+directive +perfective >> Aorist Imperative
+projective perfective >> Present & Perfect Subjunctive, Optative
+directive perfective >> Present & Perfect Imperative
32 Stanley E. Porter and Matthew Brook ODonnell
97
See Porter, Verbal Aspect, pp. 361-62. Instances of the perfect subjunctive are only
ten in the Greek of the New Testament, and imperative forms occur four times. There
are no perfect optatives.
+imperfective +stative
+assertive .................... 5221 0.81 924 0.99 6145 0.83
assertive .................... 1215 0.19 14 0.01 1227 0.17
+assertive assertive
+imperfective............... 5221 0.85 1215 0.99 6436 0.87
+stative ....................... 924 0.15 14 0.01 936 0.13
The sixth set of examples concerns the relation of the ATTITUDE sub-
system (NON-ASSERTION) (13) and ASPECT1 system. This set of systemic
choices concerns the options of +projective/+directive and perfective
(systems 13 and 2). Choice of perfective results in an equiprobable dis-
tribution of +projective/+directive choices. In this regard, the results are
consistent with those for system 13 on its own. However, the numerical
distribution is reversed. Choice of +perfective results in a distribution of
+projective (0.65)/+directive (0.35), similar to that of system 13, but choi-
ce of perfective results in a distribution of +projective (0.33)/+directive
(0.67). Both are within the range of equiprobable distribution (if only
marginally), but now at opposite ends of the range. This may well have to
do with the relationship between the subjunctive and imperative forms in
terms of negation
98
. Choice of +projective results in a skewed distribution
of +perfective over perfective (0.78/0.22), whereas choice of +directive
results in an equiprobable distribution of perfective (0.48/0.52). The
effect of selection of +projective, realized by subjunctive and optative
mood forms, is the first statistical test that has shown that choice of an-
other system within the Greek verbal network can affect choice of verbal
aspect. The result is a skewed, rather than an equipollent distribution.
However, the results of this analysis are consistent with those tendencies
noted elsewhere regarding the choices of aspectual and modal forms.
Whereas it was noted above that there was a tendency for the finite and
aorist forms to be linked, this result extends that conclusion. Here it is
noted that, although there is an equiprobable distribution regarding the
semantic features of perfective when +directive is selected, when +projec-
tive is selected the perfective feature is much more highly marked. In other
words, there is a tendency in the subjunctive and optative forms to select
the aorist tense-form more often than the present and perfect, indicating
that the present and perfect subjunctive and optative forms are more highly
marked than the aorist subjunctive and optative forms. This is in confor-
mity with analysis of the Greek verbal network presented in Verbal Aspect
99
.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 33
98
See Porter, Verbal Aspect, pp. 97, 167-73.
99
See Porter, Verbal Aspect, pp. 181, 323-24.
+perfective perfective
+projective .................. 1435 0.65 405 0.33 1840 0.54
+directive .................... 764 0.35 822 0.67 1586 0.46
projective +directive
+perfective .................. 1435 0.78 764 0.48 2199 0.64
perfective .................. 405 0.22 822 0.52 1227 0.36
7. ATTITUDE subsystem (NON-ASSERTION) (13) and ASPECT2 (3)
Realization Statements:
+projective +imperfective >> Present Subjunctive, Optative
+directive +imperfective >> Present Imperative
+projective +stative >> Perfect Subjunctive, Optative
+directive +stative >> Perfect Imperative
The seventh set of examples concerns the relation of the ATTITUDE
sub-system (NON-ASSERTION) (13) and ASPECT2 system. This set of
systemic choices concerns the options +projective/+directive and
+imperfective/+stative (systems 13 and 3). Choice of either +projec-
tive or +directive results in a skewed distribution with +imperfective
over +stative (0.98/0.02 and 0.97/0.03, respectively). Choice of
+imperfective results in an equiprobable distribution (0.33/0.67),
within the same overall distributional category, although at opposite
ends of the range, as +projective/+directive in system 13. The choice of
+stative, however, results in a slightly skewed distribution, with +pro-
jective over +directive (0.71/0.29). Thus, it would appear that choice
of verbal aspect in the ASPECT2 system does affect the choice of +pro-
jective/+directive, at least where the perfect form is concerned. This is
consistent with what has already been noted in Verbal Aspect regarding
the relative infrequency of the perfect imperative
100
. However, the very
small number of instances may mitigate the validity of these statistics.
Nevertheless, choice in the ASPECT2 system is unaffected by choice in
the ATTITUDE subsystem.
34 Stanley E. Porter and Matthew Brook ODonnell
100
Porter, Verbal Aspect, p. 362. We take the instances of . c. in Eph. 5.5 and Jas 1.19
as indicatives, not imperatives. See S.E. Porter, . c. ,... c-.., in Ephesians 5,5:
Does Chiasm Solve a Problem?, ZNW 81 (1990), pp. 270-76.
+imperfective +stative
+projective .................. 395 0.33 10 0.71 405 0.33
+directive .................... 820 0.67 4 0.29 824 0.67
+projective +directive
+imperfective .............. 395 0.98 820 0.97 1215 0.99
+stative ....................... 10 0.02 4 0.03 14 0.01
8. CAUSALITY (5) and ASPECTUALITY (1)
Realization Statements:
+active +expectational >> Future Active
active +expectational >> Future Middle & Passive
+active +aspectual >> Non-Future Active
active +aspectual >> Non-Future Middle & Passive
The eighth set of examples concerns the relation of the CAUSALITY
system and ASPECTUALITY system. This set of systemic choices concerns
the options of active and +expectational/+aspectual (systems 5 and 1).
Choice of active does not affect semantic choice of +expectational or
+aspectual in the ASPECTUALITY system, with a similar skewed distributio-
nal ratio being present for both +active (0.05/0.95) and active
(0.08/0.92) semantic features. Choice of +aspectual has a very similar dis-
tribution regarding active (0.73/0.27) as the CAUSALITY system 5 does on
its own (0.72/0.28). Choice of +expectational results in an equiprobable
distribution regarding active (0.58/0.42), rather than a skewed distribu-
tion. This is consistent with one of the noteworthy morphological fea-
tures of the future form, especially regarding having a more limited set of
voice forms (often evidencing what is called, whether rightly or wrongly,
deponency)
101
. Regardless of how one analyzes the effect of selection of
the feature of +expectational, the ASPECTUALITY system is not affected by
the CAUSALITY system.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 35
101
Porter, Idioms of the Greek New Testament, pp. 70-73.
+expectational +aspectual
+active ........................ 837 0.58 17495 0.73 18332 0.72
active ........................ 597 0.42 6492 0.27 7089 0.28
+active active
+expectational ............. 837 0.05 597 0.08 1434 0.06
+aspectual ................... 17495 0.95 6492 0.92 23987 0.94
9. CAUSALITY (5) and ASPECT1 (2)
Realization Statements:
+active +perfective >> Aorist Active
active +perfective >> Aorist Middle & Passive
+active perfective >> Present, Imperfect, Perfect, Pluperfect Active
active perfective >> Present, Imperfect, Perfect, Pluperfect Middle & Passive
The ninth set of examples concerns the relation of the CAUSALITY
system and ASPECT1 system. This set of systemic choices concerns the
options of active and perfective (systems 5 and 2). The results here are
quite easy to summarize. In both sets of choices, the results are similar to
those noted above for the simple systems. Choice of active results in an
equiprobable distribution for perfective (0.49/0.51 and 0.48/0.52, res-
pectively), and choice of perfective results in a similarly slightly skewed
distribution for active (0.73/0.27, for each), as for the causality system on
its own. Again, the CAUSALITY system does not affect the ASPECT1 system.
10. CAUSALITY (5) and ASPECT2 (3)
Realization Statements:
+active +imperfective >> Present & Imperfect Active
active +imperfective >> Present, Imperfect Middle & Passive
+active +stative >> Perfect & Pluperfect Active
active +stative >> Perfect, Pluperfect Middle & Passive
36 Stanley E. Porter and Matthew Brook ODonnell
+perfective perfective
+active ........................ 8508 0.73 8987 0.73 17495 0.73
active ........................ 3096 0.27 3396 0.27 6492 0.27
+active active
+perfective ................... 8508 0.49 3096 0.48 11604 0.48
perfective .................. 8987 0.51 3396 0.52 12383 0.52
+imperfective +stative
+active ........................ 8032 0.75 955 0.58 8987 0.73
active ........................ 2693 0.25 703 0.42 3396 0.27
+active active
+imperfective ............... 8032 0.89 2693 0.79 10725 0.87
+stative ....................... 955 0.11 703 0.21 1658 0.13
The tenth set of examples concerns the relation of the CAUSALITY
system and ASPECT2 system. This set of systemic choices concerns the
options of active and +imperfective/+stative (systems 5 and 3). Choice
of active does not affect the choice of +imperfective/+stative (0.89/0.11
and 0.79/0.21, respectively), resulting in skewed distributional results
consistent with system 3. Choice of +imperfective similarly results in a
slightly skewed distribution regarding choice of active (0.75/0.25), con-
sistent with system 5. However, choice of +stative results in a more equi-
probable distribution in choice of active (0.58/0.42), rather than a ske-
wed distribution. The distributional results here indicate some effect on
the CAUSALITY system by the ASPECT2 system, but no effect of the CAUSA-
LITY system on the ASPECT2 system.
102
11. CAUSALITY (5) and REMOTENESS (4)
Realization Statements:
+active remote >> Present & Perfect Active
active remote >> Present, Perfect Middle & Passive
+active +remote >> Imperfect & Pluperfect Active
active +remote >> Imperfect, Pluperfect Middle & Passive
The eleventh set of examples concerns the relation of the CAUSALITY
system and REMOTENESS system. This set of systemic choices concerns the
options of active and remote (systems 5 and 4). Choice of active
results in skewed distribution in favour of remote over +remote
(0.89/0.11 and 0.91/0.09, respectively). Choice of remote/+remote
results in a generally skewed distribution in favour of +active over active
(0.72/0.28 and 0.77/0.23, respectively) consonant with system 5. Again,
the results are easy to summarize here, in that neither the CAUSALITY
system nor the REMOTENESS system is affected by choice in either system.
The Greek Verbal Network Viewed from a Probabilistic Standpoint 37
remote +remote
+active ........................ 8032 0.72 955 0.77 8987 0.73
active ........................ 3106 0.28 290 0.23 3396 0.27
+active active
remote ........................ 8032 0.89 3106 0.91 11138 0.90
+remote ...................... 955 0.11 290 0.09 1245 0.10
102
See Porter, Voice in the Greek of the New Testament, for discussion.
Of the eleven interactive systems analyzed, in only one set of semantic
choices does it appear that choice of verbal aspectual semantic features is
affected by other verbal semantic choices. In interactive system test 6, the
choice of +projective results in a skewed distribution of +perfective over
perfective, whereas system 2 for ASPECT1 on its own has an equiproba-
ble distribution. Other than this one instance, there is no statistical evi-
dence that choice of semantic features regarding verbal aspect are affected
by other verbal semantic choices in the Greek verbal network.
6. Conclusions
This study has accomplished a number of goals. The first is that it has
introduced Hallidayan statistical analysis into study of the Greek of the
New Testament, in particular statistical analysis of the Greek verbal net-
work. Hallidays suggestive comments have proved a fertile ground for
developing a set of statistical tests for the Greek verbal network, as de-
scribed and graphically displayed in the Greek verbal network of Porter in
his Verbal Aspect. The second accomplishment is that this study has led to
further development of this verbal network, by adding sets of systems for
CAUSALITY, NUMBER and PARTICIPATION, and their respective sub-systems
(Appendix B, below, names these sub-systems for the sake of clarity,
although they are not all so designated above or in the system network in
Appendix A). These add important dimensions to the network and ena-
ble this diagram to represent the complete Greek verbal network.
Although some of the systems may require further development in the
light of subsequent research, at least at this point they can be entered into
discussion as part of an integrated verbal system network. The third
accomplishment is to study the statistical probabilities that attach to the
various systems within this network of Greek verbal structure. The
Hallidayan framework, with necessary modifications, has been used, label-
ling the statistical frequencies in terms of equiprobable or skewed distri-
butional results. These systems have been studied and analyzed from seve-
ral different perspectives. The most obvious perspective is to analyze the
distributional probabilities of each of the individual systems over the
whole of the Greek of the New Testament. Another is to analyze these dis-
tributional patterns over the various sub-corpora that make up the Greek
New Testament. The results of these studies have indicated that initial
predictions made on the basis of Porters previous analysis have in vir-
tually all instances been confirmed through this statistical study. The last
accomplishment is to select for further analysis a number of interactive
systems in the network, especially those that involve choice regarding
semantic features of the aspectual systems, to determine how choice of
verbal aspect may or may not be affected by selection of other semantic
features within the Greek verbal network. The results have indicated that
38 Stanley E. Porter and Matthew Brook ODonnell
in the vast majority of instances (in fact, in all instances but one out of
44), selection of Greek verbal aspect is not statistically significantly affec-
ted by selection of these other verbal semantic features. This therefore
argues, at least in part, against the competing theory that claims that lexis
is fundamental for determining Greek verbal aspect, and, decisively,
against the competing theory that claims that other verbal factors influ-
ence selection of Greek verbal aspect. As a result, the definition and per-
spective on Greek verbal aspect offered by Porter in Verbal Aspect, and elu-
cidated further in his Idioms of the Greek New Testament, is essentially con-
firmed from a distributional statistical standpoint.
Stanley E. PORTER
Matthew Brook ODONNELL
McMaster Divinity College
1280 Main St. W.
Hamilton, ON
CANADA L8S 4K1
The Greek Verbal Network Viewed from a Probabilistic Standpoint 39
40 Stanley E. Porter and Matthew Brook ODonnell
A
p
p
e
n
d
i
x

A
.

S
y
s
t
e
m

N
e
t
w
o
r
k

f
o
r

G
r
e
e
k

o
f

t
h
e

N
e
w

T
e
s
t
a
m
e
n
t
p
r
o
c
e
s
s
A
S
P
E
C
T
U
A
L
I
T
Y
C
A
S
U
A
L
I
T
Y
F
I
N
I
T
E
N
E
S
S
+

e
x
p
e
c
t
a
t
i
o
n
a
l

0
.
0
6

*
1

S
+

a
s
p
e
c
t
u
a
l

0
.
9
4
{
A
S
P
E
C
T
1
+

p
e
r
f
e
c
t
i
v
e

0
.
4
8
2

E


p
e
r
f
e
c
t
i
v
e

0
.
5
2
{
A
S
P
E
C
T
2
+

i
m
p
e
r
f
e
c
t
i
v
e

0
.
8
7
3

S
+

s
t
a
t
i
v
e

0
.
1
3

1
S
y
s
t
e
m
s

a
r
e

n
u
m
b
e
r
e
d

s
i
m
p
l
y

f
o
r

r
e
f
e
r
e
n
c
e

u
s
e

w
i
t
h
i
n

t
h
e

a
r
t
i
c
l
e
.
2
L
e
t
t
e
r
s

E

a
n
d

S

b
y

e
a
c
h

s
y
s
t
e
m

i
n
d
i
c
a
t
e

t
h
e

a
u
t
h
o
r
s


p
r
e
d
i
c
t
i
o
n
s

a
s

t
o

w
h
e
t
h
e
r

t
h
e

s
y
s
t
e
m

w
o
u
l
d

e
x
h
i
b
i
t

s
k
e
w
e
d

o
r

e
q
u
i
p
r
o
b
a
b
l
e

d
i
s
t
r
i
b
u
t
i
o
n
.
3
N
u
m
b
e
r
s

b
y

e
a
c
h

t
e
r
m

i
n

a

s
y
s
t
e
m

i
n
d
i
c
a
t
e

t
h
e

p
r
o
b
a
b
i
l
i
t
y

o
f

s
e
l
e
c
t
i
o
n
,

b
a
s
e
d

o
n

d
i
s
t
r
i
b
u
t
i
o
n

s
t
a
t
i
s
t
i
c
s

f
o
r

t
h
e

N
e
w

T
e
s
t
a
m
e
n
t
.

D
i
s
t
r
i
b
u
t
i
o
n
s

g
r
e
a
-
t
e
r

t
h
a
n

0
.
7
/
0
.
3

a
r
e

c
o
n
s
i
d
e
r
e
d

s
k
e
w
e
d
,

a
n
d

a
n
y

r
a
t
i
o

l
e
s
s

t
h
a
n

t
h
a
t

i
s

c
o
n
s
i
d
e
r
e
d

t
o

b
e

e
q
u
i
p
r
o
b
a
b
l
e
.
+

a
c
t
i
v
e

0
.
7
2
5

E
S


a
c
t
i
v
e

0
.
2
8
+

p
a
s
s
i
v
e

0
.
6
4
6

S
+

e
r
g
a
t
i
v
e

0
.
3
6
}
A
T
T
I
T
U
D
E
+

a
s
s
e
r
t
i
v
e

0
.
8
0
1
2

S


a
s
s
e
r
t
i
v
e

0
.
2
0
}
R
E
M
O
T
E
N
E
S
S


r
e
m
o
t
e

0
.
9
0
4

S
+

r
e
m
o
t
e

0
.
1
0
+

p
r
o
j
e
c
t
i
v
e

0
.
5
4
1
3

E
+

d
i
r
e
c
t
i
v
e

0
.
4
6


c
o
n
t
i
n
g
e
n
t

0
.
9
7
1
3

S
+

c
o
n
t
i
n
g
e
n
t

0
.
0
3
*
+

f
i
n
i
t
e

0
.
6
6
7

E
S


f
i
n
i
t
e

0
.
3
4
{
{
+

f
a
c
t
i
v
e
l
y

p
r
e
s
u
p
p
o
s
i
t
i
o
n
a
l

0
.
7
5
8

S


f
a
c
t
i
v
e
l
y

p
r
e
s
u
p
p
o
s
i
t
i
o
n
a
l

0
.
3
6
{
P
A
R
T
I
C
I
P
A
T
I
O
N
+

s
i
n
g
u
l
a
r

0
.
6
6
9

E


s
i
n
g
u
l
a
r

0
.
3
4


i
n
c
l
u
d
e
d

0
.
6
5
1
0

E
+

i
n
c
l
u
d
e
d

0
.
3
5


d
i
r
e
c
t

0
.
5
5
1
1

E
+

d
i
r
e
c
t

0
.
4
5
N
U
M
B
E
R
Appendix BSystem Probabilities by Sub-Corpus of the New Testament
Synoptic Pauline Johannine Other Whole New
Gospels & Letters Writings Testament
Acts
(1)ASPECTUALITY
+expectational 767 0.06 294 0.06 175 0.05 198 0.06 1434 0.06
+aspectual 13015 0.94 4635 0.94 3401 0.95 2936 0.94 23987 0.94
(2)ASPECT1
+perfective 7222 0.55 1651 0.36 1413 0.42 1318 0.45 11604 0.48
perfective 5793 0.45 2984 0.64 1988 0.58 1618 0.55 12383 0.52
(3)ASPECT2
+imperfective 5203 0.90 2576 0.86 1595 0.80 1351 0.83 10725 0.87
+stative 590 0.10 408 0.14 393 0.20 267 0.17 1658 0.13
(4)REMOTENESS
remote 4864 0.84 2929 0.98 1780 0.90 1565 0.97 11138 0.90
+remote 929 0.16 55 0.02 208 0.10 53 0.03 1245 0.10
(5)CAUSALITY1
+active 10010 0.73 3244 0.66 2930 0.82 2148 0.69 18332 0.72
active 3772 0.27 1685 0.34 646 0.18 986 0.31 7089 0.28
(6)CAUSALITY2
+passive 2290 0.61 1115 0.66 405 0.63 691 0.70 4501 0.64
+ergative 1482 0.39 570 0.34 240 0.37 294 0.30 2586 0.36
(7)FINITENESS
+finite 8771 0.64 3318 0.67 2884 0.81 1898 0.61 16871 0.66
finite 5011 0.36 1611 0.33 692 0.19 1236 0.39 8550 0.34
(8)FACTIVITY
+fact. presupposit. 3744 0.75 1137 0.71 534 0.77 980 0.79 6395 0.75
fact. presupposit. 1267 0.25 474 0.29 158 0.23 256 0.21 2155 0.25
(9)NUMBER
+singular 5680 0.65 2176 0.66 2000 0.69 1268 0.67 11124 0.66
singular 3091 0.35 1142 0.34 884 0.31 630 0.33 5747 0.34
(10)PARTICIPATION
included 6300 0.72 1608 0.48 1786 0.62 1255 0.66 10949 0.65
+included 2471 0.28 1710 0.52 1098 0.38 643 0.34 5922 0.35
(11)DIRECTION
direct 1608 0.65 724 0.42 536 0.49 365 0.57 3233 0.55
+direct 863 0.35 986 0.58 562 0.51 278 0.43 2689 0.45
(12)ATTITUDE
+assertive 7138 0.81 2417 0.73 2407 0.83 1483 0.78 13445 0.80
assertive 1633 0.19 901 0.27 477 0.17 415 0.22 3426 0.20
(13)NON-ASSERTION
+projective 808 0.49 492 0.55 335 0.70 205 0.49 1840 0.54
+directive 825 0.51 409 0.45 142 0.30 210 0.51 1586 0.46
(14)CONTINGENCY
contingent 790 0.98 461 0.94 335 1.00 198 0.97 1784 0.97
+contingent 18 0.02 31 0.06 0 0.00 7 0.03 56 0.03
The Greek Verbal Network Viewed from a Probabilistic Standpoint 41

Potrebbero piacerti anche