Forms of Address and Conversational Language in Aristophanes and Menander

Author(s): Eleanor Dickey

Source: Mnemosyne, Fourth Series, Vol. 48, Fasc. 3 (Jun., 1995), pp. 257-271
Published by: BRILL
Stable URL: http://www.jstor.org/stable/4432501 .
Accessed: 10/09/2011 06:18

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

BRILL is collaborating with JSTOR to digitize, preserve and extend access to Mnemosyne.

http://www.jstor.org
 FORMS OF ADDRESS AND CONVERSATIONAL
LANGUAGE IN ARISTOPHANES AND MENANDER

BY

ELEANOR DICKEY

...de? ?a? e??a? ???a?????

t?? de??e??? t??. p?es??te??? t?? t?? ???a?
?pa????'? e???? pat??a ?a? p?ppa[? ????.]
??a??* ??t??'. a? t?? d?a ??s?? t[?? ?? ????,]
????es' ?e??a?. a? ?e??p??[
???t?st??. ??e?? de ??e?a?[
? t?? ??a??a?. ...

'You need a soft approach when you want a favour. An older chap
answers the door: I promptly address him as "Father" or "Dad".
If it's an old woman, then "Ma". If it's a middle-aged woman, I
call her "Madam"1). If a youngish servant, then "My dear chap".
You all deserve to be strung up. Such ignorance!'2)

In this passage from

Menander's Dyscolus (492-8), the cook Sico
professes himselfto be an expert in the use of forms of address
(vocatives). Some of the terms he suggests, however, are rarely or
never found as addresses in extant Greek literature. Is Menander
indicating that Sicospoke oddly? Probably not. Although there is
a cook with
peculiar linguistic habits in Straton's comedy
F??????d??, that cook cannot be taken as representing a general
tendency of cooks in New Comedy. His obscure Homeric words are
quoted not from his own lips but from those of someone else who
is complaining about his language, clearly for comic effect. Sico in
Dyscolus, on the other hand, is not caricatured for his use of
language, and except in this passage, his Greek appears to be
perfectly normal.
We must thus assume that Menander meant the cook's instruc-

1) Literally, 'priestess'; for the extended usage compare Dottore in Italian.

2) N. Miller, Menander: Plays and Fragments (London 1987), 35.

? E. J. Brill, Leiden, 1995 Mnemosyne, Vol. XLVIII, Fase. 3

258 ELEANOR DICKEY

tions to be taken seriously, and this gives us new information about

address habits in ancient Greece. Or does it? To what extent can
we rely on Menander's information about Greek vocative usage?
Can he be taken as a faithful mirror of conversational Greek, or was
he using a language different from that spoken on the streets of
Athens in his day? To the extent that Menander's language differs
from that of Aristophanes, does that difference reflect actual
changes in the language, or did one of these poets adhere to collo-
quial Attic more closely than did the other?
The problem of the relationship of comic language to conversa-
tional Attic has been increasingly discussed in recent years3),
although at an earlier period it was sometimes taken for granted
that the language of comedy must be identical to that of colloquial
spoken Greek. The question of comic
language is of course a large
and complex one, and at the present time it would be extremely dif-
ficult to examine all its aspects in a systematic way. The smaller and
more quantifiable area of vocative usage, however, can provide
some valuable evidence about the type of language used by
Aristophanes and Menander.
The first question for a study of this type is exactly what standard
to use for the evaluation of comic language, and this decision will
*
have important consequences. We are not seeking spoken Greek',
a nebulous concept taking no account of different registers and
genres of speech, but rather conversational Attic. Discussions of
this type often get sidetracked onto the question of the difference
between spoken and written language, but in fact linguists studying
this question now think that the spoken/written difference is not
based on any real linguistic criteria4). The distinction of genre is

3) See D. Bain, Female Speech in Menander, Antichthon 18 (1984), 24-42; K.J.

Dover, Linguaggio e CaratteriAristofanei, Rivista di Cultura Classica e Medioevale
18 (1976), 357-71 [Reprinted in Dover, Greekand the Greeks:CollectedPapers, vol. 1
(Oxford 1987), 237-48]; K.J. Dover, Lo Stile di Aristofane, QUCC 9 (1970), 7-23
[Reprinted in Dover, Greek and the Greeks, 224-36]; K.J. Dover, The Colloquial
Stratumin Classical Attic Prose, in: G.S. Shrimpton, DJ. McCargar (edd.), Classical
Contributions: Studies in Honour of Malcolm Francis McGregor (Locust Valley, NY
1981), 15-25; F.H. Sandbach, Menander'sManipulation ofLanguagefor Dramatic Pur-
poses, Fondation Hardt: Entretiens 16 (1970), 111-43.
4) For a general overview and extensive bibliography of linguistic work on this
subject, see W.L. Chafe and D. Tannen, The Relation between Writtenand Spoken
Language, Annual Review of Anthropology 16 (1987), 383-407.
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 259

more important than that of medium in determining language, and

the tendency of modern European languages to associate certain

genres with specific media is by no means universal5). We shall thus

ignore the question of 'spoken Greek' and aim for conversational

Attic, the language spoken by Athenians to their friends and
families in informal settings.
How then might conversational Attic be related to the genres we

possess, tragedy, comedy, prose? and The traditional wisdom is

that tragedy represents the most elevated type of language and the
one most remote from ordinary speech, comedy the lowest level and
that closest to daily conversation, and prose a level somewhere in
between. If a distinction is made between Aristophanes and
Menander, it is that Menander's language is more elevated than
that of Aristophanes and thus further from the conversational
genre. These assumptions make possible studies of colloquialisms
in tragedy and prose by comparison with comedy, since comic
language is taken to be virtually identical to conversational Attic6).
All the evidence indicates that this ordering of the genres is cor-
rect in terms of its classification of the elevation of language; that
is, tragic language was certainly meant to be elegant, comic
language (at least that of Aristophanes) was often meant to be
vulgar, and Plato's language aimed neither at excessive elevation
nor at great vulgarity. Thus far the traditional view is virtually
unassailable, but what are the grounds for assuming that conversa-
tional language must be equated with the lowest of these levels?
That is, aside from the admitted fact that conversational Attic may
not have been identical to any of these three genres, why should it
necessarily be closest to the 'lowest' type of literary language?
Two answers to this question spring to mind. Aristotle says that
the goal of tragic language is a dignity and elevation gained by the
use of rare and unfamiliar words; that is, by the avoidance of pure
conversational language {Poetics 1458a). We shall not claim that
tragic language had much in common with the conversational

5) Cf. D. Biber, Variationacross Speechand Writing (Cambridge 1988).

6) See for example P.T. Stevens, Colloquial Expressions in Euripides (Wiesbaden
1976; Hermes Einzelschriften 38), and D. Tarrant, Colloquialisms, Semi-Proverbs,
and Word-Play in Plato, CQ 40 (1946), 109-17.
260 ELEANOR DICKEY

genre, but that still leaves us with two other genres. A second argu-
ment is itiore pernicious: conversational language must be equated
with the lowest level of literary language because that is the defini-
tion of conversational language, which is the most 'vulgar', 'collo-
quial9 of
form language.
Yet what is in fact the definition of conversational language? I
would define conversational language as that used in casual conver-
sations by people who were not thinking about their use of language
at the time. In fact such a definition will produce more than one
kind of conversational language, for it is normal for different social
groups to use different varieties of language. In England, the con-
versational language of a university professor will be different from
that of a farmer. In Athens, there were probably linguistic dif-
ferences between men and women, citizens and slaves, and natives
and immigrants. Within the range of the conversational genre,
'low' language would have been that associated with people of
lower status in the society, such as slaves, and 'high' language
would have been
that normally used by people of higher rank.
There is no reason why any one of these types of conversational
language should be regarded as more authentic than any other; the
language that an educated Athenian used in talking to his family
has just as much claim to be considered 'conversational' as the
language used in talking to their families.
that his slaves Thus there
is no reason to assume that all conversational language will be
equivalent to the 'lowest' literary genre available. Such is certainly
not the case in English, where it is not unusual for educated native
speakers to find in books, newspapers, and drama types of English
which they would never themselves use and which they consider
beneath their own conversational language. Often these forms of
English are typical of the conversational usage of a large segment
of society, one which is considered to be of lower status. It is how-
ever also possible to see in print language which is 'below' the nor-
mal conversational usage of virtually all of the population. Such
language can be used deliberately by an author or dramatist in
order to create a specific effect; it may indicate that a given charac-
ter comes from a particular segment of society, deliver a shock to
the audience, or simply be humorous.
If the writers of Greek prose and comedy were being absolutely
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 261

realistic in imitating conversational language, then, we would

expect to find a variety of different types of speech in their works,
ranging from the language of educated citizens to that of slaves and
foreigners. In fact, distinctions in language between Athenians and
foreigners are
prominent in Aristophanes, whose plays contain
Spartans, Megarians, and Boeotians speaking conspicuously non-
Attic dialects. Non-Athenians in Plato and Xenophon, on the other
hand, speak pure Attic, as do Menander's
usually characters. This
might suggest that Aristophanes' depiction of conversational
language was the most realistic, but when it comes to the subtler
and less humorous distinctions between the speech of various
groups of Athenians, Aristophanes is less impressive: Menander
appears to be the one author who consistently uses different
language for different types of character7).
There is however more
one way in which an author might
than
use conversational language in a literary work. He might choose
one type of language, for example the conversational speech of an
educated citizen, and put this language into the mouths of all his
characters. There could be a number of different reasons for doing
this: the author, or his audience, might feel that lower-class or
foreign accents would be in bad taste, or the author might not be
sufficiently familiar with the type of language spoken by certain
groups to enable him to reproduce it, or he might feel that the
audience would recognizenot or understand various types of
language. Alternatively, an author could use a variety of different
kinds of language but without assigning each type to the sort of
character who would normally use it; this could be done for the
humorous effect of incongruity, or because the author wanted to do
other things with language?such as punning?which could not be
accomplished without putting inappropriate language into the
mouths of certain characters.
How does all this apply to Attic Greek? We are attempting to
find out how close the language of Aristophanes and Menander is
to conversational Attic, and which type of conversational language,
if any, these authors represent. Let us consider what the best possi-

7) Sandbach, Manipulation, and Dover, Linguaggio; this difference between

Aristophanes and Menander was first noted by Plutarch, Moralia 853 c-f.
262 ELEANOR DICKEY

ble situation might be. Tragedy, we have agreed, does not reflect
much of conversational language. Prose is not written in the
'lowest' form of Attic, and it does not show significant variation in
language, so it might be a purely literary language, but it might
also represent the normal conversational usage of educated Athe-
nian citizens. Aristophanes generally uses a low linguistic level and
shows regional but not social variation in the language of his
characters8); he might be consistently adhering to the conversa-
tional usage of some level of society, or he might be using vulgar
language inappropriately in order to be funny or to help produce
other linguistic effects. Menander's language is apparently more
elevated than that of Aristophanes, and at the same time he depicts
social variationin the use of language; there is no reason to suspect
that he is not using conversational language realistically, but the
question will bear further investigation. In any case it must be kept
in mind that Menander lived in a different century from
Aristophanes, and some aspects of conversational language might
have changed over time.

So far the discussion has been about conversational language as

a whole, and such generality is perfectly feasible as long as we
remain at a purely theoretical level. If we are to make any concrete
arguments or produce statistics, however, it is necessary to move
to a smaller topic which can more easily be isolated and quantified.
The topic chosen in this case is the use of forms of address in
Aristophanes and Menander. The extant plays of Aristophanes
contain a total of 1168 addresses, while those of Menander provide
5259), giving us a substantial amount of evidence. One difference

8) It could be argued that the language of the Scythian policeman in the

Thesmophoriazusaerepresents a case of social variation in Aristophanes. The
policeman, however, is clearly depicted as Scythian rather than Greek, and thus
his peculiar language is more fairly described as regional than as social variation.
Social variation can certainly be found in other comedians; Amphis has a charac-
ter use t???? ????? and ?t? ????? for tett???? ?????? and ??t? ?????? (fr. 30
Kassel-Austin), but Amphis is considerably later than Aristophanes, and in any
case these words are quoted by one character as examples to illustrate his com-
plaints about another character's bad Greek, rather than being delivered
naturally.
9) The texts used for Aristophanes were as follows: Clouds, ed. K.J. Dover
(Oxford 1968); Wasps, ed. D.M. MacDowell (Oxford 1971); Ecclesiazusae, ed.
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 263

between the two authors is evident at once: in Menander, the most

common type of address is the addressee's name,
standing alone
and unmodified except for ?10) (e.g. ? S???ate?). Such vocatives
account for 40 % of the addresses in Menander but are much rarer
in Aristophanes, where they make up only 12% of the addresses.
Comparison of these figures with those from other literary genres
yields some surprising results. In Attic tragedy, the use of
unmodified names in address is relatively infrequent11), but in
Plato and Xenophon it is extremely common, accounting for 71 %
of the 3487 addresses in Plato and 60% of the 1092 addresses in
Xenophon12). Addresses consisting of unmodified personal names
are rare in Thucydides and the Attic orators, but here other factors
are at work: nearly all the vocatives in these authors are addressed
to audiences comprising large groups of people, so addresses by
name were normally impossible.
Thus the use of addresses consisting of a name alone is very com-
mon in prose, fairly common in Menander, and much less frequent
in tragedy and Aristophanes. What was the situation in conversa-
tional Attic? Two possibilities exist: that there was some type of
conversational language in which the normal form of address was
by name alone, or that there was no such type. If the former is the

R.G. Ussher (Oxford 1973); Lysistrata, ed. J. Henderson (Oxford 1987); Frogs, ed.
K.J. Dover (Oxford 1993); other plays from the Bude series, ed. V. Coulon and
H. Van Daele (Paris 1946-54). No Aristophanic fragments were included in the
corpus of vocatives. For Menander, the corpus consisted of all those fragments
assigned to known plays (but no other fragments), following the text of F.H. Sand-
bach (Oxford 1972) and including all emendations and supplements in that text;
in addition, vocatives were collected from the additional fragment of the Misumenus
published by E.G. Turner, Proceedings of the British Academy 63 (1977),
315-331.
10) The question of the use and omission of ? (discussed by J. Wackernagel,
Vorlesungen?ber Syntax: mit besondererBer?cksichtigungvon Griechisch, Lateinisch, und
Deutsch I (Basel 1926), 310-12) is a complex one and entirely separate from the
topic of the present discussion. Hereafter no regard will be taken for the presence
or absence of ?, and the terms 'alone' and 'unmodified* are to be understood to
include cases where the word in question appears with ?.
11) T. Wendel, Die Gespr?chsanredenim griechischenEpos und Drama der Bl?tezeit
(Stuttgart 1929; T?binger Beitr?ge zur Altertumswissenschaft 6), 56.
12) All figures for Plato are based on the edition of J. Burnet (Oxford 1900-
1907); those for Xenophon are from the text of E.C. Marchant (Oxford
1900-1921).
264 ELEANOR DICKEY

case, the difference between genres is easy to explain. Plato and

Xenophon consistently used this type of conversational language
(at least as far as address usage is concerned), the tragedians
while
found that system of address both repetitive and mundane and so
replaced it with a more elaborate and varied system. Aristophanes
either used a different type of conversational language, one in
which unmodified names were less common as forms of address, or
deliberately avoided the use of names because they were too com-
mon to be funny. Perhaps Menander portrayed some of his charac-
ters as using this system of address and others as using a different
system, in accordance with the actual tendencies of different social
groups, or perhaps he generally had all of his characters use the
same address system but occasionally departed from it in order to
do something else with language.
If on the other hand unmodified names were never common in
any kind of conversational Attic, the usage in Plato, Xenophon,
and Menander is very hard to understand. Why would they have
replaced a more lively and varied system of address with an endless
succession of unmodified names? If this feature did not come from
ordinary speech, why was it invented, and why did both Plato and
Xenophon follow it, when in other respects these authors are very
different? Did Menander then imitate them rather than making his
own observations of conversational language, and if so, how did he
manage to introduce into his works a social variation in language
which is not evident in Plato or Xenophon?
It thus seems more likely that there was some
form of conversa-
tional language in which unmodified personal names were the most
common form of address. Before we attempt to draw too many con-
clusions from this evidence, however, it would be prudent to look
at other aspects of the address system.
Almost all of Menander's addresses (whether by name or by
some other term) consist of a single word, or a single word preceded
by ?; the rare addresses which are longer than this almost never
contain more than two words. In Aristophanes, on the other hand,
fully a third of the addresses are made up of more than one word
(not counting ?), and many of these take up several lines of verse.
It is clear that Aristophanes' vocatives are generally more elaborate
than Menander's, and in this respect as well, Aristophanes is
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 265

definitely closer to the practice of tragedy and Menander to the

practice of prose.
Where could conversational Attic fit into this picture? Did Athe-
nians address each other briefly, or did their salutations stretch for
several lines of verse? Again the only argument available to us is
one of probability. If one-word vocatives were common in conver-
sational language, it is easy to see why the tragedians might have
elaborated and expanded them in order to make the address system
more elegant. Aristophanes might also have used long, complex
addresses for humorous effect. Yet why would Plato, Xenophon,
and Menander systematically shorten the addresses they heard
around them and
produce a repetitive succession of unmodified
names? Such a change could hardly be described as making conver-
sational language more elegant and literary.
There is a third feature of address systems which lends itself to
quantitative analysis, namely that of variety. In order to test fairly
this aspect of vocative usage, one must discount all personal,
ethnic, or divine names (since names will inevitably vary) and omit
from consideration the paratragic addresses in Aristophanes. One
then counts the number of different words (p?te?, ???t?ste, f??e,
etc.)13) used in address by each author and notes how many times
each one occurs.It emerges that Menander's addresses contain 70
different words used on average five times each, while
Aristophanes' addresses contain 427 words used an average of 2.9
times each14). Of course Aristophanes would have more words,

13) The masculine, feminine, and plural forms of each word (e.g., ??t??, a?t?,
??t??, a?ta?) are all counted as the same word.
14) In the case of Aristophanes these statistics are subjective, and I doubt that
another researcher would reach precisely the same totals. The difficulty lies in
which words are held to count as meaningful parts of an address; it is clear that
one must not count the article, or ?a?, but many other cases are less clear, and
when one is faced with a long address going on for several lines and containing
no words found elsewhere in Aristophanes' addresses, the number of words which
it is held to contain is essentially an arbitrary decision. I have in all cases tried to
count as few words as possible, so as not to distort the statistics in my favour;
another researcher might have come up with another fifty words or so. I have also
counted as two occurrences words which are repeated within one address; thus ?
Fa??? Fa??? at Acharnians271, repeated in line 276, is counted as one word occur-
ring four times, even though it appears as a vocative nowhere else in Aristophanes.
Had circumstances like this been taken into account in doing the counting,
266 ELEANOR DICKEY

sin^e the corpus of Aristophanes contains just over twice as many

addresses as the corpus of Menander, but the difference between 70
and 427 is too great to be accounted for in this way. It indicates that
Menander had a limited range of vocatives which he tended to use
repeatedly, while Aristophanes used many different words and was
much more likely to invent a vocative for a particular situation and
then never use it again.
Variety can also be tested in another way. The ten most common
vocatives in Menander
(pa?, p?te?, f??tate, ???t?ste, ??d?e?,
?????pe, ???a?, ?e???????, t??a?, in declining
d?sp?ta, order of fre-
quency) for 56% of the non-name
account vocatives in his plays.
On the other hand the ten most frequent vocatives in Aristophanes
(??t??, pa?, ??d?e?, d?sp?ta, ???a?, f??e, f??tate, a?a??, p?te?, ???e)
account for only 33% of the non-name words in Aristophanes'
addresses. This again suggests that Aristophanes' addresses were
more varied than those of Menander, and by this time the argu-
ment from probability hardly needs to be spelled out. Did
Menander simplify a more varied conversational system, or did
Aristophanes introduce more variety into his address system for his
own literary purposes?

So far we have considered

only quantifiable aspects of address
usage, but numbers
rarely produce a complete picture, so it is
worth looking at some features which are less easy to reduce to
figures. One such feature is consistency of address usage within a
given dyad. For example, in English most people have one form of
address which they habitually use for each person. One may on
occasion use insults or terms of endearment as appropriate, but
basically if, e.g., a girl addresses her mother as "Mum", her
teacher as "Mrs. Smith", and a friend as "Jane", she will use the
same addresses for these three people each time she speaks to them,
in a consistent and predictable way. This consistency is not
characteristic of all languages15), but it is certainly found in Greek

Aristophanes' addresses would have looked even more varied than they do under
the present system.
15) F. Braun, Terms of Address: Problemsof Patternsand Usage in VariousLanguages
and Cultures (Berlin 1988), 184.
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 267

prose. For example, in Xenophon's Oeconomicus the character

Ischomachus addresses his wife fifteen times, on each occasion
using the same address, ???a?. In Plato's Republic Glaucon
addresses Socrates 29 times, always with the vocative S???ate?.
Other examples abound; in fact it is generally possible to predict
with a fair degree of accuracy how most characters will be addressed
in Plato or Xenophon.
In Menander as well, many characters show a high degree of
consistency in
address usage. Thus in Samia Moschio always
addresses his father Demeas as p?te? (128, 452, 467, 486, 520, 537,
725), while his father generally uses ??s???? in return (154, 451,
459, 465 (bis), 537, 694, 709, 720) but can also employ pa? (129,
148) and once uses an insult (481). The neighbours Niceratus and
Demeas almost always address each other by name (?????ate 182,
196, 494, 517, 562, 589, 605; ????a 106, 463, 499, 506, 538, 556,
571, 583), but Demeas once uses ta? (547). Demeas always
addresses his mistress by name (???s? 378, 385, 392, 569, 575,
730). In each of these cases the address (or addresses, in the case
of ??s???? and pa?) generally occurring is the one which would be
used between those characters in Plato
or Xenophon.
Not every pair of characters, however, exhibits such consistency.
Demeas uses four different terms to address the slave Parmeno:
?a?????? (189 (bis), 295, 305), pa? (189, 202), ??t?? (312), ?ast???a
(324). An extreme case of variety is Geta in Misumenus, who uses
five different terms to address his master Thrasonides six times:
d?st???? (A20)16), ?a????e, (A28), t?? (A82), ?a??da???? (?88),
d?sp?ta (?97, ?98). It is notable that the dyads in which a variety
of addresses are used generally contain a slave, cook, or person of
similar status. When varied addresses are used in other dyads, they
appear under the pressure of strong emotion which calls for insults
or terms of endearment, but even these emotions disturb the con-

16) This example raises the issue of the nominative for vocative and what
should be counted as an address. This decision is inevitably subjective, but I have
excluded from the statistics all words, whether in the nominative or in the vocative
case, which seemed to be general exclamations rather than addresses directed at
specific individuals (? ?e??, etc.). I have included all words which did seem to
stand outside the syntax of their sentences (i.e., vocatives and exclamatory
nominatives) and to be aimed at specific addressees.
268 ELEANOR DICKEY

sistency of the address pattern less often than one might expect.
(Consider, e.g., the amount of emotion between Demeas and
Moschio in Samia.)
In Aristophanes, by contrast, almost no consistency in address
usage can be found. In Clouds, Pheidippides addresses his father
Strepsiades five different
ways: p?te? (35, 93, 816, 1325), 80,
da?????e (38, 816), ???e (1192), t?? (1432), a???te s? (858). Strep-
siades uses fourteen vocatives in return: ???e (33, 1338), Fe?d?pp?d?
(80, 827), 1165), t????? (1165,
pa? (87, 1170), fa?? (1168 (bis)),
??a?? (1325, 1388), Fe?d?pp?d??? (80), f?tate (1464), f?tat'
a????p?? ???? (110), ??a?? ?a? pat?a???a ?a? t???????e (1327),
?a???p???te (1330), ??a??tate (1332), ? ? t? s' e?p?; (1378),
a?a?s???te (1380). This variety is in no way atypical; similar figures
can be found for Bdelycleon and his father Philocleon in Wasps, as
well as for most other dyads in which a large number of addresses
occur.
What then can we conclude about conversational Attic? Was
there some type of conversational language in which sons habitually
addressed their fathers as p?te? and received in return pa? or their
names? The evidence of prose would suggest strongly that this was
the case; every address from a son to a father in the genuine and

spurious works of Plato and Xenophon uses the vocative p?te?, as

do all the addresses from sons to fathers in Herodotus, Plutarch,
Luc?an (counting human characters only), Dio Chrysostom,
Dionysius of Halicarnassus, Philo, Flavius Josephus, Epictetus,
and Longus17). In Menander all but one of the addresses from sons
to fathers usep?te?18); the
exception (Dyscolus 856) papp?a, uses
which may be described
fairly as a variant of p?te?. Yet in tragedy
and Aristophanes p?te? is only one of a virtually infinite number
of possible ways for sons to address their fathers.
Once again it seems to me that the similarity among Plato,
Xenophon, and Menander is inexplicable if the three authors were
not imitating some form of conversational language. Why would

17) The data supporting this statement are too cumbersome to insert here; they
are presented in my Oxford D.Phil, thesis and I hope in a forthcoming
monograph.
18) Dyscolus 784, 802, 806, Samia 128, 452, 467, 486, 520, 537, 725, Sicyonius
362.
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 269

each one have independently decided that fathers should be

addressed only as p?te?, if a wide range of vocatives was possible
in conversational usage? Why would they have imitated each other
on a point like this? Yet it is easy to see why Aristophanes and the
tragedians might have wanted to use a more complex and varied
system of address.

We have thus examined four points of address usage: the

prevalence of unmodified names, the length of vocative phrases, the
overall variety of addresses used, and the consistency of address
usage in individual dyads. In every case we have seen a notable dif-
ference between Aristophanes and Menander, with Aristophanes
preferring longer, more elaborate, and varied addresses while
Menander uses shorter, simpler, repetitive, and consistent
addresses. In every case where we have been able to compare com-
edy with the address systems of prose and tragedy, Menander has
been closer to the usage of prose and Aristophanes to that of
tragedy. Of course, Aristophanes and the tragedians do not actually
use the same vocatives, but nonetheless they agree in the general
characteristics that we have examined. In every case the facts of
literary address usage seem to be explicable only if prose and
Menander were reflecting actual tendencies of conversational
language.
It therefore seems very likely that, at least as regards forms of
address, Menander does largely use the language of conversational
Attic. In that case
Plato and Xenophon clearly reflect conversa-
tional addressusage as well, since they use the same system that
Menander does. We have seen that Plato and Xenophon display
virtually no regional or social variation in language and that their
language is certainly not at the lowest level; it followed from that
that if they were representing any type of conversational Attic, they
must have been consistently using the language of educated
citizens. We now conclude that they probably are using conversa-
tional language, at least as regards the address system, and this tells
us which level of conversational Attic we have found for Menander.
Yet there were two places in which
Menander diverged from
prose usage. The first was in the use of unmodified names (e.g., ?
S???ate?), which although frequent in Menander are less common
270 ELEANOR DICKEY

there than in Plato and Xenophon. The second was in the con-
sistency of addresses within a dyad, for Menander showed less con-
sistency thandid the prose authors. These two points are of course
closely related, for in prose the address consistently used in most
dyads is an unmodified name, and this is also true in Menander for
most dyads which show a consistent pattern. A lack of consistency
in address will inevitably mean departure from address by name
and thus will affect both sets of figures at once.
We noted that dyads with inconsistent address
usage generally
involved a slave or other
person of exceptionally low status. Such
characters are of course rare in Plato and Xenophon, or at least
they do not normally speak enough to use addresses. Perhaps these
differences in Menandrian usage point to an actual difference in the
way that slaves used addresses and in the way that they were
addressed. Menander has been shown to reflect elements of
women's language19), so it is not implausible that he might also
represent the conversational language of slaves. Such a claim can-
not be proven, but it is certainly possible and would account for the
only discrepancies in our findings.
If Menander was using the address system of conversational
language, generally that of educated citizens but perhaps also that
of slaves, what type of language was Aristophanes using? It could
be argued that he was using the address system employed by and
to slaves throughout his works, since that address system may have
had more of the variety which is so prominent in Aristophanes, but
this is unlikely. Menander's slaves do not use long, complex
addresses, and even if they are not consistent in the way they
address their masters, they still tend to draw their vocatives from
a relatively small and repetitive stock of terms, while Aristophanes'
characters seem to have a virtually unlimited vocabulary. It is
much more likely that
Aristophanes' use of addresses reflects not
the conversational usage of the lowest (or any other) level of Athe-
nian society, but rather an elaboration and humorous variation of
the conversational address system of educated citizens.
The variety of Aristophanes' language has been noted before20),

19) Bain, Female Speech.

20) E.g., Dover, Stile.
 FORMS OF ADDRESS AND CONVERSATIONAL LANGUAGE 271

but his plays are still considered an important source of conversa-

tional language. We have attempted to show that this very variety,
given the presence in other literature of a more consistent system,
indicates that Aristophanes' language is less reliable as a source of
some aspects of conversational Attic speech than is the language of
Menander or of prose. Conversational language can of course be
found in Aristophanes, but at least in some respects, Aristophanes'
conversational forms must be found by a comparison with
Menander and prose, rather than providing a standard by which
the latter two genres may be judged.

Merton College, Oxford OX1 4JD