SECTION A: INTRODUCTION AND BACKGROUND 1. Framework
3. Methodology
3.2 Analysing Atticism
33 Vilborg’s).120 I then manually updated the version I had downloaded to match Vilborg’s, using a print version of his text.
In searching for examples of particular forms in Achilles’ text, I used a combination of my recreated edition run through concordance software (a program called Concordance, developed by R.J.C. Watt,121 and the TLG edition analysed using the TLG’s own online search tools (more detail on this below).
34 1) It must be described as Atticist by modern scholars writing on the topic
2a) It must have been regularly used (or preferred) by Attic authors of the 5th and 4th centuries BC or Attic inscriptions of the time123
2b) It must have a non-Attic equivalent that was widely used in the Koine
3) It must have been described by the Atticist lexicographers and grammarians or other writers of the Second Sophistic as Attic/ist
I made use of the following methods in order to determine whether or not each token I examined met these criteria.
3.2.1 Modern Scholarship
As mentioned, my starting point was to consider markers that modern scholars writing on the topic have identified as Atticist. I began with the comprehensive list given in Horrocks’ book on the history of the Greek language. I give an abbreviated form of his list below:
Important hallmarks of correct Attic usage included the following:
(a) -ττ- and -ρρ- for -σσ- and -ρσ- in the relevant words (b) ξύν for simplified σύν “with”
(c) The formation of abstract nominals with the neuter article τό and an adjective in agreement
(d) Regular use of the dual number
(e) Extensive use of the dative in all its traditional functions …
(f) Use of the ‘contracted’ forms of nouns … e.g. ὀστοῦν not ὀστέον ‘bone’ etc.
(g) Retention of the Attic declension of λεώς/νεώς in place of λαός/ναός ‘people/temple’
(h) γίγνομαι, γιγνώσκω for simplified γίνομαι, γινώσκω (i) The use of the synthetic perfect rather than periphrasis … (j) Extensive use of middle verb forms …
(k) Use of the optative in its full range of classical functions
(l) The use of the monolectic perfect forms with a ‘stative/present’ rather than a ‘simple past’ meaning 124
123 For examples of such authors, see Evidence from Ancient Use (3.2.2) below.
124 Horrocks 2014: 138
35 Another list is given in Anderson’s discussion of Atticism where he says:
To write … literary Attic of the fifth and fourth centuries BC in the Early Roman Empire … The student would have had to affect various subtle stylistic choices, such as the doubling of Attic tau for Ionic sigma, the preserving of ‘Attic’ declension or the frequent use of the dual, to say nothing of the ‘purist’ declension of such troublesome words as naus; he would also have had to purge his language of deviant tendencies such as the lapse into easy alternative -o forms of -mi verbs preferred by the koiné or on the other hand hypercorrect attempts at restoring the Middle Voice. He would have had to negotiate the syntax of subordinate moods … and maintain a correct vocabulary of words used by ‘classical’ authors.125 [underlining mine]
Kim provides the following list of “peculiarities of the Attic dialect that had largely been lost in the popular language” and which “Atticising authors were careful to maintain”:
preferring ττ over σσ … and ρρ over ρσ… employing the “Attic” second declension…, the contracted forms of certain first and second declension nouns, athematic verb endings, and γίγνομαι and γιγνώσκω for γίνομαι and γινώσκω. … the dual number, the dative case, the middle voice, the perfect tense, the future infinitive, and the optative mood, among others, in their full range of Classical functions … The most striking contrast between Atticist and colloquial language, however, is in vocabulary…126 [underlining mine]
I also consulted the list found in the index of Blass, Debrunner and Funk’s New Testament grammar under the heading “Atticisms: of the Koine of the NT” and the list under the heading “Linguistic Purism and Prescriptivism of the Second Sophistic” in the entry on “Attitudes to Language” in the Encyclopedia of Ancient Greek Language and Linguistics, among others.127
From these, I selected the Atticist tokens that came up most often and placed them under the headings: phonetic (graphemic), morphological, morpho-syntactic and lexical.
It turned out to be impossible to examine all the potential Atticist markers, but I focused
125 Anderson 1997: 88-89
126 Kim 2014: 470
127 Blass, Debrunner, & Funk 1961: 266; Bubenik & Crespo 2013: 206 (EAGLL)
36 on the more common, especially of the phonetic and morphological kind. The same or similar methodologies could be extended to those variations I did not examine.
3.2.2 Evidence from Ancient Use
In order to ascertain whether or not the tokens identified were genuine examples of Atticism or not, I wished to confirm how they were actually used by both Classical Attic and Koine writers. I used a number of different methods for gathering this information. A fully annotated corpus of Greek literature would have made this task easier but in the absence of such a resource, I had to make use of those available.
Three handbooks which were of great help in this task were Threatte’s The Grammar of Attic Inscriptions (2 Volumes), Gignac’s A Grammar of the Greek Papyri of the Roman and Byzantine Periods (2 Volumes) and Blass, Debrunner and Funk’s A Greek Grammar of the New Testament and other Early Christian Literature.128
Threatte’s books helped establish normal Attic practice as evidenced by Attic inscriptions from the Classical (5th-4th century BC) period. As the work deals with inscriptions from the 8th century BC right up into the Roman periods (roughly 300 AD), I had to be careful to focus on examples from the correct period. But this was easily done as Threatte is explicit in differentiating the different periods and sometimes gives good comparative evidence, e.g. from literary texts.129
By comparison, Gignac’s books give examples of evidence from documentary papyri from the Roman and Byzantine eras, giving me examples of Koine usage. Again, I had to be careful with the dates and focussed on instances from the early centuries AD. He also often makes reference to comparative examples from both Attic and Koine literature and to inscriptional evidence of the sort Threatte examined.
One of the difficulties encountered in establishing general Koine usage is the fact that so many literary texts from the period were influenced by Atticist practice that it can be hard to differentiate between what was “normal Koine” usage and what was Atticised.
128 Threatte: 1980, 1996; Gignac 1976, 1981; Blass, Debrunner, & Funk 1961
129 Threatte 1980: 1
37 The documentary papyri help with this distinction as they are less likely to be influenced by linguistic Atticism. The New Testament along with the Greek translation of the Old Testament (the Septuagint) are also useful sources in establishing non- Atticised norms. Even these texts are not completely immune from classicising or Atticist influence. (e.g. Kim suggests that while the gospels and other early Christian literature “reflect[ed] the spoken popular Koine”, Luke was an exception that did not fit into this category, and Kilpatrick discusses examples of linguistic Atticism in some manuscripts of Revelation).130 But these texts tended to be less strongly influenced and so are helpful in establishing something of the norms of Koine practice.131 Previously, the biblical texts would have been avoided out of fear of their representing a “special variety of Greek used by Jews of the Near East,” but Horrocks points out that this is no longer thought to be the case, even for the Septuagint, of which much is translated from Hebrew.132 He describes their language rather as “a reasonably close reflection of the everyday Greek of the majority of the literate population in the early centuries AD,”
although he also recognises that there would have been some “peculiarly regional features”.133 It is for this reason that I have frequently consulted the grammar of the New Testament by Blass, Debrunner and Funk in my establishing of Koine norms.
In addition to the evidence from handbooks on Attic and Koine practice, I also made use of the tools available on the TLG, and in particular the Text Search Tool. This allows one to look up instances of word use in an individual or group of authors either by lemma or as a particular inflected form/type (using the “Word Index”). Since it was not usually practical to search for all tokens of a word throughout the entire corpus (which, as mentioned, contains texts ranging from the 8th century BC to the 15th century AD.), I selected groups of authors that I could examine to establish either Attic or Koine norms. The groups I used were as follows:
130 Kim 2014: 201, 470 does not elaborate on this point but Luke’s writings are often described as exceptions to the natural Koine or non-Atticising language of the New Testament. Adams 2013 addresses this generalisation and considers why this analysis is problematic. On Revelation, see Kilpatrick 1963.
131 Cf. Silk 2009: 22
132 Horrocks 2014: 147
133 Horrocks 2014: 147
38 For Attic use
Dramatists: Aeschylus, Sophocles, Euripides (tragedy); Aristophanes (comedy)
The Attic orators: (The “canon of ten”) Aeschines, Andocides, Antiphon, Demosthenes, Dinarchus, Hyperides, Isaeus, Isocrates, Lycurgus and Lysias
Other prose: Plato, Aristotle, Thucydides, Xenophon (historian) etc.
I consulted the dramatists because their plays were among the earliest and best respected of 5th to 4th century Attic literature. Their texts, however, especially those of the tragedians, are not always the best examples of Attic practice because of their poetic nature and the influence from poetic literature and other dialects. The search tool also does not allow one to differentiate between different parts of a particular work, which means that the Doric influence on choral sections cannot be accounted for separately. I considered the orators an especially useful source for an analysis of Attic practice both because of their texts being prose and because of the respect they held in the minds of Atticists, especially since Atticism (both stylistic and linguistic) was so closely associated with rhetoric. Other prose writers (such as Plato, Aristotle, Thucydides and Xenophon) would be consulted for comparison, especially when a form was not frequently used by the orators. Sometimes I also consulted the use of non-Attic contemporary texts such as Herodotus or the Hippocratic Corpus.
Koine use
Biblical texts: The New Testament and Septuagint
Other: Josephus, Plutarch, Galen, Epictetus
Schmid’s Atticists: Dio Chrysostom, Herodus Atticus, Lucian, Aelian, Aristides and Philostratus II.
As already mentioned, the biblical texts served as crucial examples which I could examine for Koine practice. These, along with the evidence from papyri in Gignac, were my key resources. Sometimes I consulted other contemporary texts which were thought to have little or less Atticism (in particular Galen’s corpus).134 Finally, I also considered the practice observed by the group I call “Schmid’s Atticists” (the authors he examined in his work), to see what the general trends for Koine-period authors who were attempting to Atticise might be.
134 Kim 2014: 478; Whitmarsh 2005: 47-48
39 Because of the wide range of different linguistic markers which I examined, and differences in the frequencies and trends surrounding the different forms, I did not always examine all the categories of authors listed above. Although I often established the number of tokens of a particular form in particular authors, my overall analysis was typically qualitative rather than quantitative. I wished to establish a general trend of usage in the two different periods rather than focussing on absolute statistics, which are not particularly meaningful on their own. I am also aware of the fact that, for these groups of authors, I was relying entirely on the editions of texts as represented on the TLG which might contain errors and does not account for problems in transmission or variations in the manuscript tradition. While I did take these factors into account in my analysis of Achilles’ text, I could not do the same for each and every author represented here.
3.2.3 Evidence from Ancient Witnesses (Lexicographers and Grammarians)
To establish whether the tokens I was examining were actually viewed by people of the time as Attic or Atticising, I consulted the evidence of the ancient lexicographers, grammarians and other authors mentioned in the literature review. My primary source was Moeris’ lexicon because the form in which his work survives is concise and clear.
Most of his entries take the following basic structure:
x Ἀττικοί· y Ἕλληνες.
The Attic speakers (say) x; the (Hellenistic) Greeks (say) y
Moeris is dated to somewhere between the 2nd and the early 3rd century AD, either during or shortly after when it is thought that Leucippe and Clitophon first appeared.135 Moeris’ lexicon gives a wide range of examples of phonetic, morphological, lexical and semantic variations although each example relates to a specific word, and a particular inflected form of that word with very little generalisation applied. For example, you will find him recommending the form βήττειν over βήσσειν (Moeris Atticista β.25), but no generalisation about using -ττ- rather than -σσ-.136
135 See footnote 46 on questions of the dating of Moeris.
136 The version of Moeris’ text which I used is the edition of Hansen 1998: 71-156 available on the TLG.
(#1515.2).
40 Phrynichus and Pollux’s lexicographical works relate more to word use and meaning and as such are less useful, but I do cite from them where appropriate. I also consulted the works of other contemporary grammarians, in particular Aelius Herodianus (or Herodian). He was also writing in the 2nd century AD and, as mentioned, gives interesting information regarding accepted norms of the time, although his focus is not Atticist. Herodian’s most important work was De Prosodia Catholica (On General Prosody) in which he discusses accent and breathing norms in intricate detail, addressing or attesting to other grammatical features along the way. Other works of Herodian which I make reference to include Περὶ παθῶν (On the Modification of Words), Περὶ ὀρθογραφίας (On Orthography/Correct Spelling) and Περὶ τῶν εἰς -μι (On -mi Verbs).137
I also make reference, where applicable, to other contemporary texts that reveal evidence of the attitudes to Attic or non-Attic variants, especially those of Lucian. On occasion, I make reference to examples from slightly later lexicographers and grammarians, especially where evidence from writers closer to the period of composition is sparse. Although this does not reflect directly on the understanding of the day, the Atticist programme continued through the centuries and it is probable that much of the later grammarians’ understanding (and even part of their texts) was derived from the grammarians who preceded them.
Sometimes the evidence of the lexicographers contradicts that of actual use. This information is important as it highlights the distinction between perceived linguistic Atticism and the use of “genuine” Attic forms. Because my analysis is on Achilles’
Atticist intent rather than how truly Attic he was, this evidence is important for my analysis. In most cases, however, there is at least some overlap between actual practice and the recommendations of the lexicographers. I will make note of cases where important deviations occur.
137 For an introduction to the works of Herodian, see Dickey 2007: 75-77. For a full list of all his works, see Dickey 2014.
41 3.3 Achilles’ Use
The terminology I use for analysing the statistics of the different markers found in Achilles’ text are taken largely from Corpus Linguistics norms. I use the term “marker”
to refer to the abstract variable that has a different version in Attic and Koine and for which the one is considered “Atticist” and the other “non-Attic” (e.g. the -ττ-/-σσ- variation, the “Attic declension” and athematic μι-verbs are different examples of markers which I will consider).138 I use the word “lemma” in the sense of a linguistic
“lexical item” to refer to the category that contains all inflected versions of the same
“word” (e.g. θάλασσα “sea”, ναός “shrine” and δίδωμι “I give” are examples of lemmas).139 Lemmas are usually given in their “dictionary entry” form (i.e. nominative masculine singular of nouns; 1st person singular indicative active present for verbs etc.).
Where the dictionary entry for a lemma is different in the Attic or Koine form, I normally use the form given in Liddell, Scott and Jones’ Greek Lexicon (LSJ 9th edition) unless the context makes the other more appropriate. The word “type” refers to all the potential inflected forms of a lemma (e.g. θάλασσαν, θαλάσσης, θάλατταν, θαλάττης etc. are examples of different types of the lemma θάλασσα). Finally, the word
“token” (which will be more important for me than “type”) refers to all instances of a particular lemma in a text (e.g. there are 31 tokens of the type θαλάσσης in Achilles’
text, which in turn is one of the 8 different types which he has for the lemma θάλασσα).
As mentioned, I used Vilborg’s edition of Achilles’ text to gather statistics on each of the Atticist markers I examined. To identify the tokens relevant to each marker, I searched the text in one of two ways. For phonological variants, I searched the concordance of my version of the text, which I had created using Watt’s Concordance programme. This list, which itemised every inflected type separately and provided references to all tokens of each type, allowed me to search for strings of letters such as ττ (Beta Code TT), σσ (Beta Code SS) or ρσ (Beta Code RS) etc.
For morphological markers, searching for strings of letters was not sufficient and so I had to use other methods. For example, searching for all instances of athematic μι-verbs could not be conducted by searching for the string -μι (Beta Code MI) in my
138 I use “variant” in a similar manner, but with a less exclusive/technical implication.
139 While the plural of “lemma” is sometimes given as “lemmata” (as in the TLG), I prefer the Anglicised form, “lemmas” which is common in Corpus Linguistics.
42 concordance, since that string is only found in the 1st person singular present indicative tokens of those verbs. For verbal morphology, then, I made use of the TLG’s Vocabulary Tools which allowed me to create a Lemmata List (a list of all lemmas in their “dictionary entry” forms) found in Achilles’ text. I copied this list into an MS Excel spreadsheet document from where I could search for suitable lemmas, (for example lemmas ending in -μι). From this list, I could then create a list of all types and tokens of the appropriate lemma found in Achilles’ text by searching using the TLG Text Search Tool. For nominal morphology, I used a combination of searching for individual types on the Concordance programme and searching for lemmas using the Lemmata List from the TLG.
Given the statistical results gained from the Concordance programme or the TLG (both of which are based on unannotated versions of Vilborg’s text), I then checked the tokens I had identified against the apparatus in Vilborg’s print edition, where he identifies manuscript variations. Depending on the nature of the variations, I either created a separate list, omitting unreliable tokens, or simply noted the variations in my discussion.
Because Vilborg’s edition only incorporated 3 of the 7 papyrus fragments, I examined these independently for instances of the marker being assessed. I relied on the published transcripts of the fragments for this information, consulting pictures of the original fragment where available. The small proportion of Achilles’ text which the fragments cover and the high number of lacunae in them meant that not very many of the Atticist tokens which interested me were extant in the papyri. I also consulted O’Sullivan’s Lexicon to Achilles Tatius which sometimes contained information on variations.140
For an Atticist analysis of lexical items, which I examined only briefly, I followed a different approach. I began with a sample of entries from Moeris’ lexicon. Using the Vocab Tools Lemmata Lists on the TLG and MS Excel’s formatting and function tools, I searched for which of these entries Achilles used words from and whether the words he used were considered Attic or Hellenistic by Moeris. I give more detail on how I went about doing this in the chapter on Lexical Atticism (18).
140 O’Sullivan 1980