<<

λΜνΝρΟηβμΠαΡτ ΑψΒθΨωΔφΕξΦοΓπΗε ΕκΣζΤυΘΩγΖΑψΒ ΙςΞσΚδΛλΜνΝρΟηβμ θΨωΔφΕξΦοΓπΗεEncyclopedia ΠαΡτΕκΣζΤυΘΩγΖΑψ ΙςΞσΚδΛλΜνΝρΟof ΒθΨωΔφΕξΦοΓπΗεΙςΞ ηβμΠαΡτΕκΣζΤυLanguage and σΚδΛλΜνΝρΟηβμΠαΡ ΘΩγΖΑψΒθΨωΔφ τΕκΣζΤυΘΩγΖΑψΒθΨLinguistics ΕξΦοΓπΗεΙςΞσΚδPreview ωΔφΕξΦοΓπΗεΙςΞσΚδ ΛλΜνΝρΟηβμΠαΡ ΛλΜνΝρΟηβμΠαΡτΕκ τΕκΣζΤυΘΩΑψΒθ ΣζΤυΘΩΑψΒθΨωΔφΕξ ΞσΚδΛλΜνΝρΟηβ ΦοΓπΗεΙςΞσΚδΛλΜν μΠαΡτΕκΣζΤυΘΩ ΝρΟηβμΠαΡτΕκΣζΤυ γΖΑψΒθΨωΔφΕξΑEdited by ΘΩγΖΑψΒθΨωΔφΕξΑGeorgios . Giannakis ψΒθΨΔφΕξΦοΓπΗ ψΒθΨωΔφΕξΦοΓπΗεΙ EAGLL_preview_v2.inddεΙςΞσΚδΛλΜνΝρ 1 01-05-13 15:07 Encyclopedia of Ancient Greek and

General Editor: Georgios K. Giannakis, of Associate Editors: Vit Bubenik, Memorial University of Newfoundland, St John’, Emilio Crespo, Autonomous University of Madrid, Chris Golston, California State University, Fresno, Alexandra Lianeri, Aristotle University of Thessaloniki, Silvia Luraghi, University of Pavia, and Stephanos Matthaios, Aristotle University of Thessaloniki

brill.com

• November 2013 • ISBN 978 90 04 22597 8 • Hardback • List Price EUR 850.- / US$ 1100.-

The Encyclopedia of Ancient and Linguistics (EAGLL) is a unique work that brings together the latest research from across a range of disciplines which contribute to our knowledge of Ancient Greek. It is an indispensable research tool for scholars and students of Greek, of linguistics, and of other Indo-European , as well as of Biblical . The EAGLL o�fers a systematic and comprehensive treatment of all aspects of the and study of Ancient Greek, comprising detailed descriptions of the language from -Greek to koine. It addresses linguistic aspects from several perspectives including history, structure, individual singularities, biographical references, schools of thought, technical meta-language, sociolinguistic issues, , didactics, practices, generic issues, Greek in relation to other languages, etc., and on all levels of analysis including , , , , , semantics, stylistics, etc. It also includes all the necessary background regarding the roots of Greek in Indo-European. As and when, excursions may be made to later stages of the language, e.g. Byzantine or even later.

The focus, however, will predominantly be Ancient Greek. With well over 500 entries on all aspects of Ancient Greek, this new encyclopedia aspires to become a basic research tool and the recognized reference work on the subject.

Readership: Any one with an interest in Ancient Greek language and linguistics. Classicists, linguists, historians.

λΜνΝρΟηβμΠαΡτ ΑψΒθΨωΔφΕξΦοΓπΗεEncyclopedia of Ancient Greek Language and Linguistics ΕκΣζΤυΘΩγΖΑψΒ ΙςΞσΚδΛλΜνΝρΟηβμOnline θΨωΔφΕξΦοΓπΗε ΠαΡτΕκΣζΤυΘΩγΖΑψ ΙςΞσΚδΛλΜνΝρΟ ΒθΨωΔφΕξΦοΓπΗεΙςΞ ηβμΠαΡτΕκΣζΤυEncyclopedia of brill.com σΚδΛλΜνΝρΟηβμΠαΡAncient Greek Language ΘΩγΖΑψΒθΨωΔφ • November 2013 τΕκΣζΤυΘΩγΖΑψΒθΨand Linguistics Online ΕξΦοΓπΗεΙςΞσΚδ • ISSN: 2214-448X ωΔφΕξΦοΓπΗεΙςΞσΚδ ΛλΜνΝρΟηβμΠαΡ Purchase Options: ΛλΜνΝρΟηβμΠαΡτΕκ τΕκΣζΤυΘΩΑψΒθ • Outright Purchase EUR 1275.- / US$ 1650.- ΣζΤυΘΩΑψΒθΨωΔφΕξ• Annual Subscription EUR 230.- / US$ 295.- ΞσΚδΛλΜνΝρΟηβ ΦοΓπΗεΙςΞσΚδΛλΜν μΠαΡτΕκΣζΤυΘΩ* This booklet is a preview of the major reference work Encyclopedia of Ancient Greek Language and Linguistics. ΝρΟηβμΠαΡτΕκΣζΤυ The texts in this preview might di�fer from the final publication. γΖΑψΒθΨωΔφΕξΑ ΘΩγΖΑψΒθΨωΔφΕξΑ ψΒθΨΔφΕξΦοΓπΗ ψΒθΨωΔφΕξΦοΓπΗεΙ EAGLL_preview_v2.inddεΙςΞσΚδΛλΜνΝρ 2 01-05-13 15:07 Encyclopedia of Ancient Greek Language and Linguistics

General Editor: Georgios K. Giannakis

Associate Editors: Vit Bubenik Emilio Crespo Chris Golston Alexandra Lianeri Silvia Luraghi Stephanos Matthaios

EAGLL_preview_v2.indd 3 01-05-13 15:07 of Content Ordered by Thematic Category (Some entries are included in than one category)

1. Ancient Grammatical (27 entries) 5. Greek and Other Languages (30 entries) Among others: Among others: Αllegory - Ἀλληγορία Balkan Sprachbund: Early evidence in Greek Ancient Metrics - Μέτρον Contact Through Translation Ancient Prose Rhythm Formation Of Doric Koines, The - Ἀόριστος Greek And Case - Πτῶσις Greek And (Early Contacts)

2. Historical Linguistics and 6. Sociolinguistics (32 entries) Diachronic Changes (54 entries) Among others: Among others: Accommodation Analogy Ancient Bidialectalism And Bilingualism Aphaeresis Ancient Greek Sociolinguistics And Attic Reversion Attitudes Toward Language Caland System and Greek Bilingualism In Hellenistic 7. Dialects (36 entries) Among others: 3. Lexicon (15 entries) Aeolic Dialects Among others: Arcadian Calques Arcado- Lexicon, Structure And Origin Of In Modern Dialects Terms Argolic Legal Terminology Attic Boeotian 4. Morphology (61 entries) (15 entries) Among others: 8. Theoretical approaches Abstract Among others: Action Nouns Cognitive Linguistics And Greek Activa Tantum Computational Linguistics And Greek (Morphological Aspects Of) Construction And Greek Adverbs (Morphological Aspects Of) Agent Nouns 9. Writing (17 entries) Analogy Among others: Aorist Formation Adoption of the Ionic alphabet in Attic Alphabet, Descendants Of Alphabet, Origin Of Case Syncretism (Morphological Aspects of) Cypriot

10. and Papyrology (9 entries) Among others: Derveni Papyrus Dipylon Vase Inscription Epigraphy

EAGLL_preview_v2.indd 4 01-05-13 15:07 11. Lexicography (13 entries) 15. History of Teaching (3 entries) Among others: Among others: Alphabetical Dictionaries: History Of Teaching Of Ancient Greek From Antiquity To The Byzantine Period History Of Teaching Of Ancient Greek: Concordances/Indices/Reverse Dictionaries In The Byzantine Times Bases And Dictionaries Teaching of Ancient Greek, Teaching Methods [Papyrology And Epigraphy Included] 16. History of Translation (22 entries) 12. Phonetics, Phonology and Among others: (39 entries) Arabic Tradition, Translation Among others: Byzantine phase and reception Accentuation of ancient Greek, the Anaptyxis Translation Feminism And Translation Aspiration Film Adaptation And Translation Bartoli’s 17. Literary diction, Style and Genres (19 entries) 13. Metrics (13 entries) Among others: Among others: Asianism Bridges Atticism Caesurae Choral , Diction Of Dramatic Meter , Diction Of Elegy, Diction Of 14. Syntax and semantics (117 entries) Among others: Accusative Action Nouns Activa Tantum Active Addressee Adjuncts Adpositional Phrase Adverbial Constituents Adverbs Agency and Causation Agent Nouns Agreement Anaphoric Processes Aorist Apposition Argument Clause Aspect (And Tense) Attraction (Mood, Case Etc.) Auxiliaries Beneficiary

EAGLL_preview_v2.indd 5 01-05-13 15:07 2

Abstract Nouns

Abstract nouns (ANs) are a very heterogeneous (1) (hē) krí-sis ‘separating, decision’ < kri- ‘to and diverse class. Any semantic defijinition of separate, decide’ such a class cannot describe it in its totality (cf. (2) (tò) boúl-ē- ‘purpose, will’ < boul- ‘to want, Flaux et al. 1996). However, ANs can be described wish’ by means of the functional (semantic and prag- (3) (ho) ula-g-mós ‘barking’ < ula- ‘to bark’ matic) criterion of reference, i.e., the property (4) (hē) orkhē-s-tús ‘(the) dance, art of dancing’ of linguistic signs to refer to objects in the extra- < orkhe- ‘to dance’ linguistic world. In comparison with concrete nouns (such as ‘cat’ or ‘house’), they have a lower In all these cases it is quite simple to recognize degree of reference. Even if an extensive defijini- the sufffijixes (-sis, -ma, -mós and -tús, respectively) tion of ANs cannot be given at any other level, the as separate elements which nominalize the pred- following sections will describe the category in icative content of the verbal bases (kri-, boul-, Ancient Greek (Gk.) from the of the ula- and orkhe-, with phonetic enlargements in linguistic levels which show its most signifijicant the latter three cases). See also Lallot (2008) for facets, i.e., mainly morphology, but also syntax. the of the non-simple/constructed word (“mot non simple/mot construit”) in Gk. 1. Morphological markedness [This has been shortened for the preview.] Because of their low referentiality, ANs are considered less typical members of the Germana Olga Civilleri category (compared to concrete nouns, which are highly referential). According to typologi- cal predictions, we should expect these nouns Accommodation to be marked at some level by specifijic features that underline their peculiar position in the Language accommodation (LA) is the tendency (cf. Croft 1991:67). For instance, at a of a speaker to modify linguistic behavior accord- morphological level such forms should contain ing to interlocutor characteristics. “Language morphological markers, e.g. sufffijixes, that carry Accommodation Theory” was developed in the specifijic functions (→ Derivational Morphology). 1970s (Giles 1979) in the area of social psychol- In Gk., a language with extremely rich morphol- ogy and is based on the assumption that speak- ogy, this is certainly true at least for part of ers are motivated to adjust their speech style, the category, i.e., event/state/process nominals. or accommodate it, to express their attitude Such nouns are semantically complex because to others. The motivation for accommodation their semantic-pragmatic function is not to lies in the (unconscious?) desire of speakers refer to objects (‘reference’, which is normally a to associate themselves with (positive LA), or property of nouns) but to predicate something keep themselves apart from (negative LA), given (‘predication’, which is normally a property of social groups. In order to have one of the two ; → Predicative Constituents). Hence their typologies of LA, speakers should possess dif- morphological complexity matches such kinds ferent languages and/or social and regional - of markedness. lects (→ Dialects, Classifijication Of ). In order to In most cases, as in (1)–(4) below, ANs are predict instances of LA, a model based on the deverbal nouns (DNs) that nominalize the event/ interaction of linguistic (structural diffferences state/process encoded by the verbal base (the between dialects) and sociolinguistic (degree root or the stem) from which they are derived of reciprocal prestige of interlocutors) factors by means of sufffijixes. Sufffijixation is indeed the was developed, namely the “ethnic boundary most frequent morphological device for forming model ” (Giles 1979). The model suggests that DNs in Gk. structural distance (→ Structural Linguistics And

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 1 6 4/5/2013 5:37:4301-05-13 PM 15:07 2 activa tantum 2 Greek) and sociolinguistic prestige determine tics And Dialectology) between the of the degree of LA. and that of the → Doric area. An example of positive LA is present in an archaic dedication from the (Car- Carlo Consani tledge 2000) in which two diffferent dialects, namely → Arcadian and → Laconian, belonging to distinct groups, combine in an unexpected Activa Tantum manner. Here, the use of the proper of (→ Theonyms) in Arcadian (Posoidâni) Activa tantum (‘active only’) are verbs that lack a can be understood as chosen by the author of → middle (→ mediopassive) and a passive the dedication—maybe a powerful person from (→ Passive (syntax), → Passive (morphology)) in Laconian society—to address the people of the present, aorist and stems. Activa , in order to testify to his political power tantum are typically intransitive (→ Transitivity). in a period during which was interacting Examples are: áēmi ‘blow’, baínō ‘come, go’, eîmi with some Arcadian (Consani 2012). On ‘go’, eimí ‘be’, eméō ‘vomit’, ethélō ‘want, be the other hand, the chancery of the of Lar- prepared’, geláō ‘laugh’, gēthéō ‘rejoice’, hérpō isa ’s meticulous separation of V’s decrees ‘creep’, khézō ‘ease oneself ’, ménō ‘stay’, noséō ‘be written in Koine (→ Koiné, Features Of ) from ill’, nostéō ‘return’, omikhéō ‘urinate’, ózō ‘smell’, the city law decrees in the local dialect, and the pēdáō ‘jump’, pheúgō ‘flee’, píptō ‘fall’, rhéō ‘flow’, translation into the dialect of the Macedonian steíkhō ‘go’, stílbō ‘shine, glisten’, tréō ‘tremble’, zô sovereign’s decrees (IG IX 2, 517, end 3rd . BCE), ‘live’. Activa tantum often have a middle future can be interpreted as negative LA in that the form, as e.g. gelásomai ‘I will laugh’, pēdḗsomai dialect is heralded as a symbol of city indepen- ‘I will jump’ (Attic). Schwyzer and Debrunner dence from external political power. A similar also count as activa tantum verbs of which mid- tendency indicating distance with respect to the dle forms occur with a passive meaning (but not Koine , but with greater variety in both local with (in)direct reflexive, reciprocal or intransi- and regional linguistic forms, is exhibited by tive meaning), e.g. verbs of eating and drinking another inscription from the early 2nd c. BCE such as bibrṓskō ‘eat’, dáknō ‘bite’, esthíō ‘eat’, (Tziafalias-Helly 2004–2005); in this example pínō ‘drink’, trṓgō ‘gnaw’. For activa tantum, see local forms from ’s dialect (-nthi/-nthein further Schwyzer and Debrunner (II:225–226), third pl. active/passive) appear together with Rijksbaron (2006:155). For the occurrence of the → Thessalian (-men inf. ending added to the- middle voice in the future stem, → Voice. matic verbs) and Northwestern items (toi nom. Semantically, activa tantum tend to be stative plur.). A further example of the problematic verbs , verbs of motion or verbs denoting physi- of the diffferences between spoken and cal (bodily) or mental processes. Many activa literary language is represented by the use of dif- tantum refer to events in which the subject is ferent dialects in literary works, particularly in physically or mentally afffected as a result of the plays which are characterized by greater realism event. This can be explained by the semantic compared with other literary genres, such as the unmarkedness of the active voice, i.e., the active Old Comedy (→ Comedy, Diction Of ). The socio- voice is neutral with respect to the semantic linguistic analysis of ’ plays con- feature of subject-afffectedness . For the seman- ducted by Colvin (1999) reveals an unexpected tics of activa tantum of bodily motion, see Allan absence of LA phenomena in the language of (2003:243–244). the characters from → and , Rutger Allan whose dialects were perceived as quite difffer- ent from the → Attic dialect. By reproducing the characters’ use of dialect, without any LA, the Active author achieved humor and realism: this could be a strictly literary strategy. But, according to The verbal grammatical category of voice pertains the “ethnic boundary model”, the absence of LA to the relationship between syntactic roles and could also be due to the structural and sociolin- semantic roles (agent , patient and experiencer ). guistic distance (→ Ancient Greek Sociolinguis- Ancient Greek has three morphologically distinct

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 2 7 4/5/201301-05-13 5:37:44 PM15:07 3 adjectives (morphological aspects of) 3 4 voice categories: active voice, middle voice, and 1) They have either (a) three (masc., fem, neut.) passive voice (→ Voice). The act. voice is marked or (b) two (masc./fem., neut.) ; by act. endings: -ō, -eis, -ei, etc. (→ Diathesis/Voice simply speaking, they are ‘of three endings’ (Morphology Of )). The act. voice can be viewed or ‘of two endings’; as the unmarked member in a privative oppo- 2) Masc. and neut. are inflected as either (c) 2nd sition (Garciá Gual 1970:11–12, 29–32, Ruipérez declension stems (o-stems) or () 3rd declen- 1988, Duhoux 2000:114, Allan 2003:19–30), i.e., the sion stems. act. form is neutral with respect to the semantic feature of subject-afffectedness: it does not signal Feminines, if there is a separate form, are always the absence of subject-afffectedness. inflected as 1st declension stems; however, their Act. verbs can be transitive and intransitive types difffer according to whether the masc. (and (→ Transitivity). In the prototypical act. transi- neut.) forms follow the 2nd or 3rd declension. tive event, the subject is a visible, volitional, con- Beside 2nd declension masc./neut. o-stems, Att. trolling and non-afffected agent while the object fem. forms have long ā (after e, i, r: → Attic Rever- is a visible, result-registering patient. Prototypi- sion) or ē in the nom. and acc. (and gen. and dat.) cal transitive clauses are marked with the act. sing.; beside 3rd declension masc./neut. stems, voice, e.g. ho paîs NOM anéōixe ACT tḕn thúran they have short ă in the nom. and acc. sing. ACC ‘the child opened the door’. In prototypi- (and long ā [after e, i, r ] or ē in the gen. and dat. cal transitive clauses, the subject is marked by sing.). Attic ē beside ā is an innovation of Attic- the , while the object is put in Ionic; → Ionic went even further than Attic, with the accusative. The act. voice is also used to ē also after e, i, r. The other dialects have kept code intransitive events, such as apothnēískō original ā. As a rule, simple adjectives are of ‘die’, baínō ‘go, step’, basileúō ‘be king’, geláō three endings, and compound adjectives of two ‘laugh’, eimí ‘be’, eîmi ‘go’, eruthraínō ‘be red’, endings. ménō ‘stay’, ózō ‘smell’, katheúdō ‘sleep’, píptō Examples (nom. + gen. sing.): ‘fall’, siōpáō ‘be silent’, zô ‘live’. Many of these verbs designate states (often derived from nouns (a)+(c) masc. díkaios, dikaíou—fem. dikaíā, or adjectives). A large number of these act. dikaíās—neut. díkaion, dikaíou ‘just’ intransitive verbs occur only in the act. voice masc. sophós, sophoû—fem. sophḗ, sophês— (→ Activa Tantum). neut. sophón, sophoû ‘wise’ Numerous act. verbs can be used both transi- (b)+(c) masc./fem. ádikos, adíkou—neut. tively and intransitively (so-called labile verbs ), ádikon, adíkou ‘unjust’ e.g. ágō ‘lead’ (trans.): ‘march’ (intr.), elaúnō (a)+(d) masc. takhús, takhéos—fem. takheîă, ‘drive’ (trans.): ‘drive, ride, proceed’ (intr.). More takheíās—neut. takhú, takhéos ‘swift’ examples in Kühner and Gerth (I:91–96), Smyth masc. mélās, mélanos—fem. mélaină, melaínēs— (1956:389). In some cases, the intransitive use neut. mélan, mélanos ‘dark, black’ is the result of an ellipsis of the typical object. (b)+(d) masc./fem. eugenḗs, eugenoûs—neut. An example is the elaúnō which can be eugenés, eugenoûs ‘noble’. used transitively elaúnō híppon ‘I ride a horse’ as well as intransitively (with ellipsis of the object) In → Proto-Greek, feminine forms were derived elaúnō ‘ride’. from masculine stems by means of a feminine sufffijix: *h₂ [This article has been shortened for the preview.] or (b) IE/Proto-Gk *ih₂ (becoming iă/yă) after . Thus from masc. né(w)o-s ‘new’, fem. Rutger Allan *néwo-h₂ > *néwah₂ > néā was formed; and from masc. mélas, stem mélan-, we get fem. *mélan-yă > mélaină (with any > ayn [= ain]) Adjectives (Morphological Aspects Of) (→ Indo-European Linguistic Background).

On the basis of two criteria Ancient Greek adjec- [This article has been shortened for the preview.] tives fall into four groups: Frits Waanders

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 3 8 4/5/2013 5:37:4401-05-13 PM 15:07 3 4 anaptyxis 4 Anaptyxis taining the resonant. Deviations from this norm are extremely rare and may be due to spelling Anaptyxis or is the insertion mistakes : peristiraphésthō = Att. peristrephésthō of a → vowel between two consonants. - ‘whirl round (3rd sg. pr. impv. med./pass.)’ linguistically, anaptyctic (also referred (Selinous, 475–450 BCE), ethinôn for ethnôn ‘race to by the term as svarabhakti vowels) (gen. pl.)’ (papyrus, 3rd c. BCE), galoios for gloiós develop between clusters of stop + ‘thick (masc.)’ (papyrus, 4th c. CE). (Skt. ratna ‘jewel’ > Pali ratana), sonorant + stop In → Lesbian and → Thessalian, when an r (Lat. argento ‘’ > Osc. aragetud) and sibi- was preceded by a stop and followed by the lant + stop (Eng. speed > Korean [sɨphidɨ]). Dis- → /j/ (from older prevocalic /i/), an similatory in nature, vowel epenthesis is due to a epenthetic -e- or -a- developed; subsequently, tendency to broaden the perceptual and articu- as frequently in these dialects, of latory distance between two segments (Ohala r occurred (e.g. kúrion > Thess. kûrron ‘lord’; 1992, Hall 2011). Like the related phenomena → ): of and , anaptyxis is sourced from hyper-articulated forms of speech (Blevins PN Príamos > Lesb. Pérramos (Alc. fr. 42, L.-P.) 2004). Agriánios > Lesb. Agerránios ‘the (month) Some Anc. Gk. prehistoric changes are due Agerránios’ to sporadic anaptyxis: *septmo- (cf. Lat. septi- Lagḗtria > Thess. Lageitárrai ‘to () Lag- mus) > Att.-. hébdomos, Dor. hébdemos ‘sev- etarra’ (, II BC) enth (masc.)’. An epenthetic -i- has been used * Korṓtria > Thess. Koroutárra(i) ‘to (Ennodia) to explain several developments (Vine 1999): Korotarra’ (ca. 350 BC). *kwtures > Lesb. písures ‘four’, *sp-nó- > *sipnó- > ipnós ‘oven’, * pt-n(e)-h2- > * ptnā- > pítnēmi A similar phenomenon seems to have developed ‘spread out’, *skd-n(e)-h2- > *skdnā- > skídnēmi in Mycenaean, resulting in an epenthetic -i-: ‘disperse’ (but a morphological innovation can- a-ke-ti-ri-ja / askētri( j)āi / ‘fijinisher (fem. nom. not be ruled out in the last two examples). The pl.)’ > a-ke-ti-ra2 / askētirrāi / (but this interpre- prehistoric evolution of inherited syllabic sono- tation has not gained universal approval). rants (→ Syllabic Consonants) is also anaptyctic: Lento-style articulation is probably responsi- ble for the strictly orthographic epenthetic vow- IE * Cr̥C > Gk. CarC or CraC els in → Mycenaean and → Cypriot , e.g. IE * Cr̥HC > * CrVHC or *CVrHC > Gk. CrVC̄ or Myc. ti-ri-si ‘three (dat. masc./fem.)’ for /trisí/. In CVrVC a similar fashion, vowel epenthesis occurs once IE * ClC̥ > Gk. CalC or ClaC in the fijirst Delphian hymn (128 BCE), one of the IE * ClHC̥ > * ClVHC or *CVlHC > Gk. ClVC̄ or few documents of antiquity attesting to musical CVlVC etc. notation: pétras ‘stone (gen. sg.)’ was sung as trisyllabic and thus spelled peteras (Bélis 1993). (→ syllabic consonants and → laryngeals for exam- This is paralleled by the 1971 song Annabella ples). Umbrella, in which Clifff Richard systematically Evidence for word-initial and word-internal sings umbrella as /ʌmˈbɛɹɛlə/. anaptyxis within the history of Gk. is marginal Alcorac Alonso Déniz and unsystematic: PN Βaránkhos = Βránkhos (Hippon. fr. 105.6 West), manasios = mnasíous ‘corn-measure (acc. pl.)’ (Olympia, 5th c. BCE), Att. Heremês = Hermês ‘’, Askalapiós = Ancient Bidialectalism And Asklēpiós ‘’ (Thessalia, 3rd c. BCE), Bilingualism place name Salamṓna = Salmṓnē (Olympia, 5th c. BCE). Some of these examples may be due One of the most outstanding characteristics of the to anticipatory or perseverative graphic repeti- ancient Mediterranean world is the great variety tion of the vowel . Arguably, Hom. pélethron is of languages and dialects which have coexisted a variant of pléthron ‘measure of of 100 within this vast area. In the ancient world as a feet’ with epenthetic -e-. The anaptyctic vowel whole, individual or social unilingualism is the mirrors the vocalic nucleus of the con- exception, while bilingualism is the rule. How-

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 4 9 4/5/201301-05-13 5:37:44 PM15:07 5 ancient bidialectalism andbilingualism 5 6 ever, by contrast to other regions in the Near [Miletti 2008:47]). We also fijind in his work bilin- East , the world can be characterized as gual people, for example the Scythian Skyles, a monolinguistic area (Rochette 2010:282). From born from a woman of . His mother, who (8.144 [Miletti 2008:29]) we learn that was presumably Ionian, taught him Greek lan- the were conscious of their cultural and guage and letters (4.78 [Miletti 2008:51]), which linguistic unity as Greeks: they were united by was a reason of anger for the and even common race, language, and customs. the cause of his death (4.80). We know the name Non-Greeks, on the other hand, spoke ‘barbar- of a few other bilinguals (díglōssoi), i.e., men who ian’, a language which was thought to sound knew Greek and a tongue: the Carian like the twittering of birds. Archaic and classical Mys, who during the Persian wars received from was little concerned with bilingualism Mardonios the mission to consult all the Boeo- and with contacts with other languages (Werner tian oracles (Hdt. 8.133 [Miletti 2008:56–57]) and 1983, 1989). Even during the , another Carian, Gaulites, sent by Greeks who knew —still a barbarian or ‘mid as ambassador to the Spartans (Thuc. 8.85, 2) barbarian’ tongue according to the Greeks—are (→ Greek And Carian). gives evi- few (for example , , Appian ). For dence of the knowledge of the the Romans Greek was like a mother tongue, in Athens when he refers to Artaphernes who as demonstrated for example by the → code- was sent to Sparta by the Great King and led to switching in ’s Letters to Atticus (Adams Athens as a prisoner in 425 BCE. The Athenians 2003:308–346), for the Greeks Latin remained a read the letters Artaphernes carried after having foreign language. translated them from Assyrian letters (ek tôn Assuríōn grammátōn) (4.50, 2). In ’s The Archaic and Classical Period there are some references to interpret- ers’ bilingualism, especially between Persian and Before the 5th c. BCE, the evidence of contact Greek (1.2, 17; 2.3, 17; 4.2, 18; 4.5, 34; 5.4, 4). In with other languages is scant. Only two pas- Plutarch ’s Lives we fijind other attestations of sages in the mention the linguistic diversity bilingualism: Greek and Persian (, among Trojans’ allies (2.803–805, 4.436–438). In 28, 1), Greek and barbarian tongues (- the Catalogue of the Trojans and their allies tocles, 6, 4), → Greek and Latin (, 27, 2; Cato (Il. 2.867), the are called barbaróphōnoi Maior, 12, 5). He also mentions the multilingual- ‘barbarian-speaking’. The most ancient evidence ism of Queen (Antonius, 27, 4). of bilingual people in the appears Aristophanes ’ reflect linguistic in the Homeric Hymn to (7th c. BCE?). diversity in Greece in another way, but it is The goddess says that she is the daughter of difffijicult to tell to which extent the barbarized Otreus, the king of the , adding that (incorrect) Greek of the foreigners is supposed she knows the language of the Trojans because to reflect the real talk of foreign speakers from she had a Trojan nurse (111–115). Herodotus is various countries. In the Thesmophoriazusae, the fijirst Greek author who manifests an interest staged in 411 BCE, a Scythian archer talks in in foreign languages and bilingualism (Miletti a tongue half Scythian, half Attic. According 2008). According to him, the Pharaoh Psammeti- to Willi (2003:198–225), the language of non- chus instructed and Carians to teach Greeks in Aristophanes is a literary expression of Greek to young who were intended Greek ethnocentrism, but not in an absolute to become interpreters in Egypt (Torallas Tovar way. In a passage from the Acharnians (100–107), 2010:257–258). He met their descendants during an envoy of the Great King delivers before the his travels in that country. One of them read assembly of the Athenians a sentence which and translated for him a hieroglyphic inscription nobody can understand (100: ἰαρταμαν ἐξαρξαν engraved on the walls of Cheops’ pyramid (2.125, απισσονα σατρα, iartaman exarxan apissona 6 [Miletti 2008:48]). Herodotus alludes to con- satra). Even if this verse has a humoristic char- tacts between various languages like Lydian and acter, it is possible to reconstitute its meaning Persian (1.86, 4.6) or Greek and Persian (3.38, 4; taking into account what one knows today about 140, 3) (→ Greek And Iranian). He mentions the the Old Persian at the end of the 5th c. BCE seven tongues used by the Scythians for the trade (Willi 2004). If this sentence is authentic Old Per- in the river and in the (4.24 sian (with the verb “write” as a central element),

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 5 10 4/5/2013 5:37:4401-05-13 PM 15:07 5 6 aorist 6 we can conclude that there were some bilingual that the aorist is the form normally found with speakers in Athens (Greek/Persian) able to trans- adverbs of cardinal count, which is connected late Greek into Persian or Persian into Greek. But with its denoting countable events (cf. Arm- such an assertion remains speculative. strong 1982:10), and, at the time, it may co- occur with durative expressions in order to focus [This article has been shortened for the preview.] on the duration of the state consequent to the completion of the action (cf. Napoli 2006:77–82). Bruno Rochette As demonstrated by cross-linguistic studies, “duration can be explicitly asserted in sentences with the perfective view-point” (Smith 1997:72). Aorist A Greek example is the following:

The aorist is one of the so-called temporal stems (1) tês thalássēs tês kath’ heautoùs ekrátēsan of the Greek verbal system (→ Aorist Formation, hēméras perì téssaras kaì déka → Aorist—ἀόριστος). However, its function is “They were masters of the about their own primarily aspectual (→ Aspect (And Tense)), and coasts for fourteen days” (Thuc. 1.117.1.4–5) its temporal value is limited to the indicative form. Non-indicative moods have no specifijic The perfective function of the aorist also temporal meaning, but they must be interpreted explains why, as observed in the literature, it as purely aspectual (→ Consecutio Temporum et often assumes an ingressive value, by focusing Modorum). on the beginning of the action, or a terminative The aorist indicative corresponds to a past value, by focusing on its end-point. This makes tense, although it may show a deviation from its use with telic verbs particularly frequent in past meaning. It may substitute for the present, as indicative and non-indicative moods. “a dramatic device found only in the literature Its relationship with the of telicity of the stage” (Cooper 1998:638), or with a gno- has also been considered as a possible explana- mic value (i.e., tenseless; → Gnomes), especially tion for the fact that the passive forms in -(th) in maxims , sentences, → proverbs, in order to ē- are exclusively associated with the aorist stem, convey a statement of universal validity (this which typically implies a high degree of afffected- use is more common in poetry than in prose; ness since it denotes a completed event (Allan cf. Gildersleeve 1900:109; Schwyzer and Debrun- 2003:176–177). ner 1959:260–262, Humbert 1960:145). The aorist Maria Napoli indicative may also have a function and denote a past action which precedes another past action, by conveying anteriority. Finally, already in , it may be used to Aorist—ἀόριστος express future events, especially when the speaker is emotionally involved (Duhoux 1992:385). The Greek term aóristos (ἀόριστος) is a priva- Since Brugmann (1885/1913), many schol- tive verbal built on the root of the ars have defijined the aorist as punctual or as verb horízō ‘to defijine, determine’, thus mean- non-durative , consistently with the idea that it ing ‘undefijined, undetermined’. Apart from other denotes the event without considering its dura- applications in grammar , the word referred to tion: more specifijically, it is defijined as expressing one of the tenses of the Gk. verb (→ Tense/ “une action pure et simple” (Meillet 1922:212). Aspect), the morphology of which is - More recently, it has been pointed out that ized by the morpheme -sa (1 sg.), e.g. é-lu-sa, the the aorist has an aspectual value comparable aorist corresponding to present lú-ō ‘(re)solve’. to the value proper to the cross-linguistic cat- The reason for designating this tense, which was egory of the . As is characteris- generally seen as ‘perfective ’ by ancient scholars tic of a perfective form, the Greek aorist depicts (see below), as ‘undefijined’ is somewhat surpris- the event as a single whole, i.e. as ‘global’ or ing. According to the on the Techne ‘bounded’, without taking into account the indi- ascribed to Dionysius Thrax (see in particular vidual phases of which it is made up. From this Schol. Dion. Thrax 250.26), it may be explained perspective, it is possible to account for the fact within the framework of a description of the

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 6 11 4/5/201301-05-13 5:37:44 PM15:07 7 aorist formation 7 8 Gk. verbal system of Stoic origin (→ Ancient Lin- skápson ‘fijinish digging’. But it is remarkable that, guistics: On Language). According where modern scholars call skápson an aorist to this description there are three tenses which imperative, Apollonius, like the Homeric scho- refer to the past: the perfect tense (parakeíme- liast, does not use the term ‘aorist’, even though nos, lit. ‘adjacent’) refers to the recent past, he knows it (Apoll. Dysc. Synt. III 143–146) and which has just been completed, and is opposed applies it both to indicatives like édeira ‘I flayed’ to a pluperfect (hupersuntélikos) which refers to and the subjunctive deírō (Synt. III 144.392.5); a distant past; in opposition to these two forms neither does he use the term suntelikós ‘com- of past tense defijined by their ‘quantity of past- pleted’, which would fijit well with his account ness’, the aorist is negatively qualifijied as ‘not of the perfective value of aorist forms. Without specifying’ the temporal distance separating the naming this ‘tense’, he fijinds it sufffijicient to pres- related facts from the moment of utterance— ent the perfective value as an efffect of the ‘past’ and from this point of view it is ‘undefijined’, meaning (parōikhēménon: Synt. III 102.358.9), a a-(h)óristos. In this description, the value which in his eyes attaches to all aorist stem tense is considered to be applied to actions forms, even the modal ones. almost completed but still taking place at the moment of utterance; therefore, strictly speak- [This article has been shortened for the preview.] ing, it is not a fully past tense (→ Aorist). As the aorist employs a sigmatic paradigm, Jean Lallot it lends itself to a parallel (termed ‘kinship’, sungéneia) with the other sigmatic tense of the temporal system, the future (e.g. lú-sō): since Aorist Formation anything in the future is inherently uncertain, the future as a tense is undefijined by nature. Thus, The Greek aorist exhibits a range of verbal forms the aorist is declared katà tḕn aoristían tôi mél- equaled only by the . The seman- lonti sungenḗs ‘by its indefijiniteness related to the tics of the aorist obscures the core function of its future’ (Schol. Dion. Thrax 251.9). morphology. Since aorist fijinite forms generally If we leave the “technographical” discussion refer to anterior events, the aorist is treated as and turn to the philological one, we fijind the a and historical tense. (Indeed, Diony- aorist (called suntelikós by Aristarchus ) being sius Thrax counts it among the four past tenses; opposed to the imperfect (paratatikós, lit. Aorist—ἀόριστος.) But the aorist indicative refers ‘extensive’), seen here as a past tense proper: to time other than the past in , , in numerous passages of the ancient Homeric Greek and the . scholia, the Alexandrian grammarians (above all Anterior reference, however, is not an essen- Aristarchus ) declare that one of the two tenses is tial function of the aorist from the PIE point of used (incorrectly) instead of the other, the two view, as demonstrated by the aorist’s capacity past tenses not being, normally, synonymous to express ingressive , resultative and gnomic or interchangeable. If an explanation is to be senses (see Sihler 1995:447–452; → Aorist; → Indo- given (a rare fact; the Greek reader of the critical European Linguistic Background). This range of commentary is supposed to ‘sense’ these things), temporal and aspectual reference may result this is an aspectual one: thus, Schol. Hom. Il. from the PIE aorist’s development as a punctual 11.368: . . . exenárizen· hoútō dià toû · ou gàr eté- category. In oblique moods, the aorist contrasts lesen ‘[we should read] exenárizen with with the present aspectually (the aorist is atem- [i.e., an imperfect, not the aorist exenárixen], poral and non-iterative ; the present is progres- for he has not completed [his action]’. The aorist sive or iterative ; Apollonius Dyscolus , however, is viewed here, as its Aristarchean name indi- suggests that the aorist conveys completion ). cates, as perfective , and is opposed to the imper- From the perspective of IE languages, the aspec- fect which presents the action in its extension tual distinction of the aorist has been claimed as (parátasis) and as incomplete. We see a simi- a Greek development although probably incipi- lar intuition in Apollonius Dyscolus (Synt. III ent in IE (see Sihler 1995:442fff., 510–511, Szemer- 102.358.3), who opposes the extensive value of ényi 1987:16–17). the present imperative skápte ‘go on digging’ to The Classical Greek aorist has fijinite - the terminative value of the aorist imperative tive forms with an augment , is productive in the

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 7 12 4/5/2013 5:37:4401-05-13 PM 15:07 7 8 aphaeresis 8 same range of moods as the present (imperative, ists seem to form directly from nouns as in góon subjunctive and ) and exhibits a full (‘they mourned’; not goáō) and thérmeto (‘was complement of and infijinitive forms. heated, became warm’). Additional variations Unlike the present, aorist morphology distin- include verbs with middle athematic guishes between middle and passive forms. For alongside active thematic forms (éktato and most regular verbs, the aorist generates active éktanon ‘killed’; cf. Chantraine 1973:381–383); and middle forms from the same stem, while thematic aorists with diffferent vowel grades (e.g., the passive forms are generated from a diffferent ‘to gather, assemble’: ḗgreto, égreto; agéronto; stem. From our earliest evidence, Greek devel- ēgéresthai) and lengthened vowels for metrical oped two general morphological systems for the needs (genómenos, geinómenos from gígnomai aorist: sigmatic aorists and root aorists (or sec- ‘become’). ond aorists ). The path to this picture is not quite clear. [This article has been shortened for the preview.] → offfers some evidence for aorist and perfect alternation as well as for Christensen the passive stem (see Palmer 1963:58 and 266). Early Greek evidence supports the gradual development of the s-aorist. Largely, Homeric Aphaeresis evidence reveals multiple aorist morphologies and an opportunistic form selection with a high In Ancient Greek aphaeresis, or prodelision , percentage of second aorists. While Watkins refers to the → of a short word-initial (1962:52–60) has argued that the sigmatic aorist a or e preceded by a long word-fijinal vowel. It developed in the middle voice fijirst, in Homer occurs mainly in poetry (particularly drama) and the aorist is more productive in the active (see archaic prose inscriptions. Aphaeresis is rarer Drinka 1995). The older PIE sigmatic aorist is than → or elision , two other phenomena reflected in Homer in combination with roots related to vowel contact across word boundaries that end in consonants. In this category, the (→ Prosody). is lost during athematic formation. For Aphaeresis is usually triggered by a long vowel example, élexa (‘I said’) has the middle forms or in a proclitic word (→ ), nor- elégmēn and élekto; ôrsa (‘I incited’) has middle mally the defijinite article or the negative mḗ. The ôrto, and participle órmenos. elided vowel mostly belongs to a (preposi- Homeric forms also display overlapping tions, such as es ‘to’), forms of the (estí morphologies from the perspective of Classical ‘(he) is’, éstō ‘let (it) be’), preverbs in compound Greek. Some sigmatic stems follow e/o vocaliza- verbs (apó, epí) or the → augment in the past tion instead of vocalization (e.g., edúseto tenses (e-): El. mā ’pénpoi ‘(if he) did not pro- ‘entered, put on’ and ebḗseto ‘went, walked’, nounce a sentence’, toî ’ntaûta ēgram(m)énoi ‘(in instead of edúsato and ebḗsato). Homeric sig- the penalty) written here’, Arg. mḕ ’npipaskésthō matic aorists may also present double sigmas ‘let (him) not acquire’, Ion. ḕ ’s Ermṓnossan ‘(the (convenient for metrical reasons; e.g., etélessa road) which (leads) to Hermonossa’, Lac. toì ’s ‘I completed’ and ekómissa ‘I entertained’). ásista ‘the closest (relatives)’, Selinous ḕ ’pak- Following Doric, some Homeric dental conso- oustòn ḕ ’phoratón ‘to be listened to or looked nant stems use -xa instead of -sa (e.g., hḗrpaxe upon’, Locr. ha ’piwoikía ‘the ’, mḕ ‘snatched away’). Often Homeric verbs, like ’xsthōásaijen ‘(if they) did not pay the penalty’. Vedic and some Avestan , have no augment in The following examples are taken from poetic the aorist; traditionally these verbs have been texts: khṓrei ’s tḕn naûn ‘run to the ship!’ (Aris- assumed to have injunctive or conjunctive force toph. Lys. 605), ḕ ’pò manteías ‘or by some sort of (see Hofffmann 1967). divination’ (Soph. Trach. 239), ei ’pitaxómestha Homer frequently exhibits second aorists ‘if we accept orders’ (Eur. Supp. 521), koû ’sti ‘where where Attic has a sigmatic aorist, as in peíthō is (Pyrrias)?’ (Herodas 5.9), kelētísai ’kéleuon ‘I ‘convince’ (epíthonto vs. Attic epeísanto); there ordered (her) to get on top’ (Aristoph. Vesp. 501). are also places in Homer where second aorists Aphaeresis can also occur after lexical words: coexist with s- aorists (e.g. díon ‘they feared’, and Lesb. sâma ’pì Stheneíāi ‘(I am) the monu- edíesan). Unlike , some Homeric aor- ment upon Stheneias’, aretḗ ’stin ‘the virtue is’

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 8 13 4/5/201301-05-13 5:37:45 PM15:07 9 arabic tradition, translation 9 10 (Thgn. 147), gunà ’ntì tḗnōn ‘the woman (will Hellenized before the advent of Muslim rule. have her reward) for those things’ (Theoc. Ep. Remnants of the Hellenistic scholarly tradition 20.3), euphēmía ’stō ‘let it be a sacred silence!’ still remained active at some of the ancient (Aristoph. Av. 959), ek potamoû ’panérkhomai ‘I seats of learning, e.g. in or the more return from the river’ (Anacr. fr. 40 Page PMG). recent schools of and Nisibis in Aphaeresis of the initial vowel of a con- and Gondēšāpūr in Persia (Endress 1987:402–7). tent word is less common: Aigina têi ’phaíēi ‘to Where secular school structures had faded, Aphaia’, Ion. mḕ ’lássones ‘(let) no less (than Christian monasteries and schools had become three hundred judge it)’, Att. phthiménē ’khō ‘I, key transmitters for Greek and philoso- dead, have (him dead)’, ḕ ’niautôi ‘(more) than a phy, either in the original Greek or in the form ’ (Ar. Ra. 18). of into local languages, especially In some cases the ambiguity of archaic scripts Syriac. Those remaining carriers of Hellenistic (→ Epichoric Alphabets) makes aphaeresis —Greek manuscripts, translations into indistinguishable from crasis and elision. For Syriac and the scholars who read and taught instance, Arg. TONUWALIO (sc. toû Enualíou, these texts (cf. Conrad 1999)—became the ‘of Enyalios’) may be seen as aphaeresis (toû seeds for a Greek-Arabic translation efffort that ’nualíou), crasis (tounualíou) or even elision ultimately far outgrew preceding Greek-Syriac (t’ Enualíou). Similarly, it is not clear whether translation activities, both in terms of the range Aeginetan KHOLEPHAS (sc. kaì ho eléphas ‘and and number of translated texts and the linguis- the ivory’) should be read as crasis of kai plus ho tic, philosophical and scientifijic profijiciency of its followed by aphaeresis (khō ’léphas), elision of kai (cf. Gutas 1998:136–41). and crasis of ho plus eléphas (k’ hōléphas) or just In the multilingual environment of the early crasis of the three vocalic elements (khōléphas). Islamic state, translation was a daily necessity. Such exchanges were in all probability mostly Enrique Nieto oral and fulfijilled the immediate practical needs of administrators and traders. We know little more than the of some translators active Arabic Tradition, Translation at this stage. Probably apocryphal reports credit an Umayyad prince, Ḫālid ibn Yazīd (d. 704), a Even before the rise of Islam, the Arabian Penin- grandson of the Umayyad caliph (r. 661–680), sula was exposed to the cultural heritage of the with commissioning a translation of an alchemi- Hellenized cities of and Syria , mainly cal text. Such accounts document a strong desire through the medium of caravan trade. Together to profijit from the scientifijic and technological with goods and religious , traders, travel- know-how preserved by speakers of other lan- ers and religious fijigures transmitted scraps of guages (cf. Ullmann 1978; Endress 1987:416–20). Greek philosophical and scientifijic knowledge to The political and religious shifts that accom- the Arabic-speaking population of the peninsula panied the accession of the ʿAbbāsid in (Rosenthal 1975:1–2; Berkey 2003:39–49). 749 and the foundation of Baghdad as the new The foundation of Islam in the early 6th c. CE capital in 762 laid the groundwork for a genu- was followed by a period of rapid expansion. In ine and systematic translation ‘movement’. In the 7th and 8th c., the early Islamic state took parallel with the fijirst offfijicially commissioned control of large areas of and the translations, e.g. of Aristot. Top. by the caliph formerly ruled by and al-Mahdī (r. 775–85), we hear of a budding infra- Persia. Rather than replacing existing admin- structure to support research and translation. istrative structures, the new Muslim authori- During the reign of Hārūn al-Rašīd (r. 786–809), ties frequently relied on local administrators, a library devoted to the study and transmission who continued to use their respective languages, of the scientifijic heritage, the so-called ḫizānat including Persian , Greek and (Gutas al-ḥikmah [storehouse of wisdom], was estab- 1998:17–18, 23). Only at the end of the 7th or the lished in Baghdad, possibly based on an earlier beginning of the 8th c., administrative record- caliphal library. Under the caliph al-Ma⁠ʾmūn keeping was ‘Arabized’. (r. 813–33), this institution, renamed bayt Many of these areas, especially Egypt , Syria al-ḥikmah [house of wisdom], was apparently and , had been more or less strongly expanded, but its exact size, function and

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 9 14 4/5/2013 5:37:4501-05-13 PM 15:07 9 10 argolic 10 connection to the translation movement are of neighboring Laconian , of which there is no still a matter of debate (Gutas 1998:28–104). further evidence. The sources clearly show that members of the According to Nieto Izquierdo (2009), how- ʿAbbāsid court society, including caliphs such ever, the Argolic dialects share three linguistic as al-Ma⁠ʾmūn and his successor al-Muʿtaṣim features which suggest a period of dialectal unity (r. 833–42), actively encouraged and supported between ca 1200 and 800 BCE. To begin with, translators and that this support extended for (→ synizesis) is confijined in both beyond court circles. areas to theós > thiós ‘god’ and its derivatives: Argos Thiokléos ‘of Theocles’, Epid. thiṓi ‘to the [This article has been shortened for the preview.] god’. Futures and aorists of verbs ending in -ízō and -ázō have // or // depending on the Uwe Vagelpohl phonetic context (/ss/ only if the radical has a velar stop): edikássato ‘(he) judged’, but paren- ephánixe ‘(he) showed’. Finally, unlike other Argolic West Greek dialects, the preposition poí (< potí = Att. prós ‘towards’) is only found before a dental The dialectal varieties of West Greek (→ Doric) stop: poidêsai ‘to bind to’, but potispastêra ‘thong spoken in the Argolid (northeastern Pelopon- which draws the bolt of a door’ (sg. acc.). nese ) are generally referred to as Argolic. Argolic After ca 800 BCE, Argolic split into three dia- dialects are known almost exclusively from epi- lectal areas: , Argos–Myce- graphic sources from Argos , and Tiryns nae, and Hermion. Three linguistic features in the Western Argolid, and from Epidaurus separate the fijirst two varieties: a) the aspira- and its Asclepieion, Troezen and Hermion in tion of intervocalic /s/, which is only attested in the East. Most inscriptions with dialectal fea- Argos and Mycenae after the early 5th c. BCE: tures come from Argos and from the Epidau- Arg. epoíwēhe vs. Epid. epoíēse ‘(he) made’; rian Asclepieion, and can be dated between the b) the diphthong /ew/ < /eo/, which only appears 8th c. BCE and the 1st c. CE. There are also some in Epidaurus in the ending - (ca 400 BCE): Argolic glossae in the lexicon compiled by Hesy- Epid. Damopháneus ‘of Damophanes’ vs. Argos chius (→ Dictionaries Of Dialects: From Antiq- Epikráteos ‘of Epikrates’; and c) the simplifijication uity To The Byzantine Period). of secondary -ns- between vowels and of fijinal According to Bartoněk (1972; followed by -ns, which is only attested in Epidaurus–Troezen Fernández Álvarez 1981), the Argolic dialect was (4th c. BCE): Epid. enkatheúdousa ‘to sleep’ (fem. divided into two geographical areas referred to ptc.) vs. Argos hápansan ‘whole’ (fem. sg. acc.). as Western and Eastern Argolic respectively, As was said above, Hermion difffers from both which became diffferentiated shortly after the areas in the secondary long vowels. Dorian invasion of the Peloponnese. Since East- ern Argolic shares some innovations with Corin- Enrique Nieto thian and Megarian , it is thought to be a member of the so-called → Saronic dialects (the vernacu- lar dialects spoken around the Saronic gulf ). Aspiration Bartoněk’s conclusions, accepted by most schol- ars, are based on the results of the secondary Aspiration is the friction made by the air passing (i.e. not inherited) long vowels (→ Compensa- through an open glottis (Eng. /ˈhεvən/). tory Lengthening): Argos bōlá ‘assembly’, ágonsa In Classical Greek /h/ was a almost ‘to bear’ (ptc. fem.), xḗnos ‘foreigner’, dámou (sg. restricted to word-initial position. Archaic gen.) ‘of the people’ vs. Epidaurus boulá, ágousa, alphabets used the sign <Η> (originally called xénos, dámou. Previous studies (e.g., Bechtel hêta, see Sch. D.T., p. 486, 32–35) for aspiration. 1921, Thumb-Kieckers 1932) assumed that such Although <Η> was gradually abandoned after diffferences were due to the early influence of the the 4th c. BCE, due to the extension of the Ionic Attic-Ionic koine on Eastern Argolic. Secondary script (in which <Η> was recycled with a new vowels in Hermion difffer from the rest of the value /εː/, due to the early loss of /h/ in Argolid: bōlá, ágōsa, xénos, dámō. According to Minor Ionic, see infra), Greek to Latin Bartoněk, this is due to a supposed influence and other languages show that initial aspiration

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 10 15 4/5/201301-05-13 5:37:45 PM15:07 11 assimilation 11 12 was retained until Roman times. The sign <˫>, vocalic aspiration word-internally in Classical originally a handwritten variant of <Η>, was Attic is restricted to the → interjections eîhen, introduced by Alexandrian scholars and even- euhoî, and to the tahôs ‘peacock’. tually evolved into the spiritus asper (pneûma dasú) of modern editions (i.e. <῾>). [This article has been shortened for the preview.] Historically, initial /h/ is the outcome of PIE *s and *sw-: heptá ‘seven’ (cf. Lat. septem), hístēmi Alcorac Alonso Déniz ‘I stand’ (cf. Lat. sisto), hēdús ‘pleasant (masc.)’ (cf. Lat. suavis). It may also go back to a word- initial cluster of a laryngeal + *j: pronoun hós Assimilation (< *h1jo-, cf. Lat. is). Some words with original Assimilation is the most common phonologi- *w- (→ ) unexpectedly show aspira- cal process cross-linguistically, examples from tion: hésperos ‘(of ) evening (masc./fem.)’ (cf. Lat. English including ‘gimme’ for ‘give me’ or vesper), heîma ‘garment’ (cf. Lat. vestis). Since ‘impossible’ from the prefijix ‘in-’ before [p]. The *sw- developed into voiceless /ʍ/ and later /h/, opposite process is → . Despite its initial /w/ in these forms may have been pro- prevalence, its causes and efffects are not always nounced /ʍ/ by hypercorrection . For reasons not easy to identify (Miller 2010:178–181). Assimila- well understood, aspiration was generalized to all tion of consonants may be usefully categorized words beginning with PIE *u-: húdōr ‘water’ (Ved. according to a number of parameters, of which udán-), hústeros ‘latter (masc.)’ (Ved. úttara). In the main subtypes are given below. The extent other cases h- is clearly analogical: e.g. Heraclean to which vowel assimilation occurred in Greek is hoktṓ ‘eight’, Theran and Locrian hénatos ‘ninth disputed (Sihler 1995:88–89, van Beek 2011). (masc.)’, after heptá ‘seven’ and Dor. hébdemos 1) Regressive vs. progressive assimilation. In ‘seventh (masc.)’. Initial aspiration in híppos regressive assimilation AB changes to BB (more ‘horse’ is puzzling: see however p.n. Leúkippos, common in Greek and typologically); in progres- not **Leúkhippos, with the expected outcome sive assimilation: (cf. Lat. equus). Voiceless stops become aspirated before /h/: regressive: trib-tós → trip-tós ‘rubbed’ katá + háper > katháper ‘just as’ (spellings like leg-tós → lek-tós ‘chosen, spoken’ kathháper with < thh > in Attic inscriptions are progressive: *thárs-os >thárr-os ‘boldness, daring’ rare), apò + hairéō > aphairéō ‘I take away from’, (Early Att. > Class. Att.). ouk + Helénēs > oukh Helénēs ‘not of ’. Aspi- 2) Place vs. manner vs. laryngeal setting. Place ration is written in compounds in early inscrip- and often assimilate, as tions after other consonants and vowels: Att. does laryngeal setting (state of the glottis: aspira- párhedros ‘assessor’, proshēkétō ‘belong to (3 sg. tion and voicing in Greek): imp.)’, ahṓrios ‘unseasonable’ (masc.). /h/ does not block crasis or : kaì hoi > khoi ‘and the (masc. place: en-bállō > em-bállō ‘I throw, put in’ [labial] pl.)’, tà hátera > thátera ‘the rest (nom./acc. pl.)’, manner: gé-graph-mai > gé-gram-mai ‘I have been apò hoû > aph’ hoû ‘after’. In addition, /h/ does not written’ [nasal] create heavy quantity for a preceding syllable: see lar. setting: plék-dēn > plég-dēn ‘entwined, plaited’ the beginning of the hexameter hímeros hairḗsei [voice] ‘desire will seize (him)’ on → Nestor’s Cup. e-plék-thēn > e-plékh-thēn ‘I was entwined’ [aspira- tion] Aspiration from inherited intervocalic *s, which eventually disappeared in later Greek, was still These assimilations may be seen in the regular preserved in → Mycenaean. The special syllabo- alternations of root-fijinal labial and velar stops gram represents /ha/: pa-we-a₂ /phárweha / before following dentals and /m/: ‘piece of cloth (nom./acc. pl.)’, me-zo-a₂ /mézoha/ ‘bigger (nom./acc. pl.)’. In other cases aspiration gráph-ō ‘I scratch, write’ gráb-dēn ‘scraping, graz- is presupposed by orthographic : e-e-si / ing’ gégram-mai (1.sg.pf.mid./pass.) plék-ō ‘I weave’ plég-dēn ‘entwined, plaited’ pé-pleg- éhensi / ‘they are’, e-ke-e /ékhehen/ ‘have’ (inf.), mai /pépleŋmai/ or /péplegmai/ (1.sg.pf.mid./ te-qa-i /thḗgwāhi / ‘at ’ (cf. Class. Athḗnēsi pass) ‘at Athens’, with analogically restored /s/). Inter-

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 11 16 4/5/2013 5:37:4501-05-13 PM 15:07 11 12 attic reversion 12 Although two successive stops agree in aspira- Att. Ion. other dialects tion (plék-ō, eplékh-thēn 1 sg. aor. pass.), if they have the same place of articulation the fijirst is timḗ timḗ timā ́ ‘honor’ conventionally written as the unaspirated stop, neā ́ neḗ neā ́ ‘new’ e.g. Sapphō <Σαπφώ>, not *Saphphō <Σαφφώ> oikíā oikíē oikíā ‘house’ (on the phonetics of these clusters see Lejeune khṓrā khṓrē khṓrā ‘country’ 1972:69, Threatte 1980:541–546 and 570–571 with bibliography). Inherited *ā raised to [ē] through the inter- Not all combinations are so regular. Contigu- mediate value [ǣ], as attested in Cycladic Ion. ous dental stops, for instance, do not assimilate inscriptions from the 6th c. BCE, which maintain as simply as a labial or velar before a dental, a three-way distinction between etymological *ē but surface instead as a cluster of /s/ + dental: (written as Ε), etymological *ā (Η), and a new underlying oîd + tha (2.sg.pf.act. of oîda ‘I know’), [ā] (A) arising from → compensatory lengthen- presumably assimilated to *oit-tha, is realized ing (Buck 1955:90, Gates 1976:190). as oîs-tha; diachronically, this cluster reflects an Two possible scenarios for the Att. develop- → epenthesis (or afffrication) of PIE date (PIE *TT ment have been proposed. One holds that the > *TST > Greek /st/), e.g. PIE *wid-tós ‘known’ development of *ā to [ǣ] was never fully com- > Greek is-tós (Mayrhofer 1986:110–112, Sihler pleted after e, i, and r (Schwyzer et al. 1939). The 1995:201–203). other proposes that the development of *ā to 3) The degree to which the assimilation [ǣ] was a P(roto)-Att.-Ion. (Bran- occurs is also to be distinguished, ‘partial’ vs. denstein 1954, Gates 1976, Lejeune 1972, inter ‘total’ assimilation. Labial-fijinal stems before alia). Under this analysis, Att. later reversed this, /m/ show total assimilation, since the labial seg- taking [ǣ] back to [ɑ̄] in certain environments ment becomes identical to the following /m/: (d and h below). Peters (1980) proposes the gráph-ō, gégram-mai. On the other hand, velar- following relative (slightly emended fijinal stems show a partial assimilation before by Samuels 2008): /m/, assimilating in voicing and perhaps nasality but not in place: plék-ō, pépleg-mai /pépleŋmai/ ̵ or /péplegmai/. See → Greek . a) Unconditioned shift [ɑ] > [ǣ] (PAtt.-Ion.) 4) Finally, we should distinguish whether the b) [eǣ] > [ǣ] sounds are contiguous or separated by some c) Dissimilation [ǣ] . . . [ ǣ] > ̵ material, ‘contact’ vs. ‘distance’ (or ‘harmonic’) [ę] . . . [ǣ] ̵ assimilation. Contact assimilation is very fre- d) Reversion: rhotic [ǣ] > [ɑ] / r __ quent and all of the preceding examples are lowering cases in point. Distance assimilation is excep- e) Contraction [eɑ] > [ǣ] tional in Ancient Greek (p.n. ‘Mekakléios’ for f ) Loss of [w] > ø ̵ ̵ ̵ ‘Megakléios’, IG II2 8491). Distance dissimilation, g) Quantitative [ę] > [e] / __ {ǣ, ɑ, ǫ} on the other hand, is more common (cf. → Grass- metathesis ̵ mann’s Law). h) Reversion: [ǣ] > [ɑ] / {i, e} __ dissimilation ̵ [This article has been shortened for the preview.] i) Merger [ǣ] > [ę]

Jesse Lundquist Phonetically, a reversion could be explained if *ā did shift to [ǣ] in all positions in Att., but Attic Reversion the pronunciation of the new [ǣ] remained a bit more open after e, i, and r. The lowering Attic-Ionic difffered from other dialect groups efffect of rhotics on a following vowel is well in its treatment of PIE *ā. While other dialects documented (Walsh Dickey 1997). The efffect typically retained * ā as [ɑ̄] (written as α), in of [i] and [e] on [ǣ] can be explained as → dis- Att.-Ion. *ā merged with inherited *ē (written similation (Szemerényi , 1987 [1968]), particularly as η), as seen with timḗ below. In Att., however, given that in other cases in Greek where two some cases of expected [ē] after [e, i, r] surface front vowels came into → hiatus, → contractions as [ɑ],̵ as seen with neā,́ oikíā, and khṓrā : or other avoidance strategies occurred. Other

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 12 17 4/5/201301-05-13 5:37:45 PM15:07 13 augment 13 14 dialects exhibit either rhotic-induced lower- Types of Augment ing or dissimilation , while only Att. shows both. Examples include occasional sub- The augment is of two types: (1) the (original) stitutions of e for expected i after r in Aeol. and syllabic, “qualitative” (also “”) augment, a few scattered examples in Sicilian Dor. and El. added to roots beginning with a , of (Buck 1955:25–26); e also lowered to a before r in the shape e- (a- in Sanskrit , e- in Armenian), NW. Gk (Buck 1955:22). sometimes ē- (ā- in Sanskrit), if the root began with a digamma in Greek, or a resonant in - Bridget Samuels skrit: Homeric ēeídei from *ēweid-‘knew’; Sanskrit āvar ‘covered’ from root *wer ‘cover’; and (2) the “quantitative” or “temporal” augment, which is Augment a feature of Greek only, whereby augment and root-initial (short) vowel are contracted, giving The augment is an inflectional verbal prefijix, a long vowel: a > ē (akoúei ‘he hears’ > ḗkouse ‘he associated with past tense, occurring (in Greek) heard’), ai > ēi (aídō ‘I sing’ > êidon ‘I sang’), e > ē, only in the indicative of the aorist, imperfect i > ī, o > ō, etc. If the stem already begins with a and pluperfect. Although attempts have been long vowel, a “long” diphthong or ou, there is no made to argue for its presence in other branches change (Smyth 1956:146, §§ 435,436: Mastron- such as Slavic , Germanic (e.g. Gothic iddja ‘I arde 1993:114). went’, as the reduplicated imperfect of yā ‘go’) Certain ancient parallel forms suggest that, and Hittite (ēsun ‘was’ from *e-Hes-m) (Szemeré- in the mother language, or a dialectal variant nyi (1996:297–299), the data are few, weak, and thereof, the augment contracted with a root- unconvincing. Its limited distribution to a few initial vowel or laryngeal (and possibly s (Palmer members of the Indo-European language fam- 1980:294), the so-called “long augment”): ily (Greek, Indo- , Iranian, Armenian and *h₁e-h₁es > *ēs ‘was’ in (Homeric) êa ‘I was’, Vedic Phrygian ) suggests a late common innovation āsaḿ ‘I was’ (→ Laryngeal Changes). Sihler sug- within this group of languages. The reason for gests that it was on this pattern that the tempo- this restricted occurrence is not known. Sihler ral augment developed (1995:485). (1995:485) hints at some sort of aspectual sig- Exceptional or irregular cases of the augment nifijicance, an insight supported by evidence from usually involve roots beginning with r, s, or w, or both Ancient Greek and Old Indic : in Old Persian a combination thereof. For instance, if the root and Avestan optatives the augment is used to begins with r, the r is doubled after the augment: mark repeated or habitual past action. In Greek, rīptṓ ‘throw’ > érrīpton ‘threw’; sometimes the iteratives in -sk- both in Homer and Herodotus double rr is due to assimilation of wr (Homeric usually have no augment, suggesting some type érrēxa ‘did’ < wreg- ‘do’), or sr (érreon ‘flowed’, of aspectual incompatibility. cf. Skt. sravati ‘flows’). Since digamma disap- The augment was, in origin, an unaccented, peared early in the history of the language, many independent word (*h₁e-) of adverbial or deictic augmented forms show no trace of it, as ṓikoun nature, meaning ‘yonder’, or ‘there’, extended from oikéō ‘dwell’ (< (w)oíkos ‘household’). There metaphorically to refer to distance in time, are cases of “double augments”, such as from rather than space. When added to a form already the verb méllō ‘be destined to’, with imperfect characterized for tense by “secondary endings ”, ḗmellon as well as émellon, aorist ēméllēsa as well it clarifijied or reinforced the notion of the ‘past’ as eméllēsa (Mastronarde 1993:132, nt.2). tense: é-lūon ‘was loosening’ (impf.), é-lūsa In Greek and Sanskrit, the augment is ‘loosed’ (aor.), e-lelúkē ‘had loosed’ (pluperf.). accented: é-pheron ‘carried’, Skt. á-bharam. In Without the augment, forms were ambiguous com-pound verbs, that is verbs composed of and could be variously interpreted, as modals a prefijix and a root, both in Greek and San- (injunctives , in the oldest usage), or with past, skrit the augment occurs closest to the verbal timeless, or even (occasionally) present refer- root, between the prefijix(es) and the root, and ence (for the injunctive see → Indo-European bears the accent (which may not precede it): Linguistic Background). prosbállō ‘I attack’ > pros-é-ballon ‘I attacked’, kat-é-graphon ‘I was writing down’, sun-é-legon

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 13 18 4/5/2013 5:37:4501-05-13 PM 15:07 13 14 bartoli’s law 14 ‘collected’, Sanskrit abhi-á-gacchat > abhyágac- e.g. agorḗ ‘assembly’ < ageírō ‘I gather’ or molpḗ chat ‘approached’. ‘dance’ < mélpomai ‘I dance’ (see Risch 1974:10 for examples; → Action Nouns). Bartoli (1930:32) [This article has been shortened for the preview.] notes that a large percentage of tomḗ nouns were disyllabic and thus were not susceptible Sarah Rose to Bartoli’s Law, and therefore, he claims that oxytone accentuation spread to the few tomḗ nouns whose accent would have been retracted Bartoli’s Law by Bartoli’s Law, such as agorḗ. The chronology and dialectal distribution of ‘Bartoli’s Law’ (Bartoli 1930) describes the pho- Bartoli’s Law is controversial, in part because our nological process in Greek by which words knowledge of the accentuation of dialects other that were originally oxytone (stressed on their than Attic , Ionic and Lesbian (which has gen- fijinal syllable) became paroxytone (stressed on eralized recessive accentuation) is limited (see their penultimate syllable) if they ended in a Buck 1955:85, Probert 2006:70fff. for discussion). sequence of a light followed by a heavy syllable: Bartoli (1930:34) regarded this sound change as e.g. *thugatḗr > thugátēr ‘daughter’, *eretḗs > -dialectal. Kiparsky (1967:77), on the other erétēs ‘rower’. Note that word-fijinal con- hand, proposed that Bartoli’s Law applied in taining a short vowel followed by a consonant Attic only, in an attempt to unify it with → Ven- (e.g. the masculine nominative singular ending dryes’ Law, which is uncontroversially limited to -os) count as light for Bartoli’s Law: e.g. khtha- Attic. While there is little doubt that it applied malós ‘on the ground’ rather than *khthamálos. in Attic, it is probable that Bartoli’s Law also Words that contained fewer than three sylla- applied in Ionic; there is no example of a lexical bles were unafffected by Bartoli’s Law: e.g. patḗr item that is afffected by Bartoli’s Law in Attic but ‘father’ instead of *pátēr. The original analysis not in Ionic: e.g. thugátēr is the only version of of the Greek material was proposed by Matteo this word attested in Ionic, while we never fijind Bartoli (1873–1946). *thugatḗr. Intra-paradigmatic alternations resulting from The phonetic motivation for Bartoli’s Law Bartoli’s Law are normally eliminated by → anal- remains unclear, but it is one of several accent ogy ; thus, alongside the nominative singular retraction processes that is sensitive to syllable erétēs we fijind a nominative erétai, weight, alongside → Wheeler’s Law and Ven- which has generalized Bartoli’s Law accentua- dryes’ Law. tion instead of preserving the expected *eretaí (word-fijinal -ai and -oi are treated as short by Steven Faulkner the rules of accentuation, cf. Probert 2006:61). The notable exception to this tendency is the word for ‘daughter’, which faithfully preserves Causative Formation the alternations arising from Bartoli’s Law: hence nominative singular thugátēr beside accu- Causatives are verbs which refer to a causal sative singular thugatéra, which displays the relation between two events, i.e., verbs mean- expected accentuation without application of ing ‘cause to V₀’, ‘make V₀’, where V₀ stands for Bartoli’s Law. the non-causative (anticausative) member of the While numerous exceptions to Bartoli’s Law opposition. have been observed (see Bartoli 1930:29fff. for Ancient Greek has no specifijic causative or further discussion), some of these violations anticausative morphemes. However, there are belong to certain morphological categories that several morphological means to express caus- systematically do not obey Bartoli’s Law, but ative meaning (for monographic studies of this rather have created unifijied patterns of accentua- category see Kuehne (1882) and Hildebrand tion that are not sensitive to → syllable weight (1889)). or the historical placement of the accent. One (i) The commonest formal type of causative of these categories consist of fijirst-declension opposition is associated with → diathesis (active/ o-grade deverbative action or result nouns (i.e., middle voice distinction): causative members of the “tomḗ” type), which are consistently oxytone: the opposition take active morphemes, while

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 14 19 4/5/201301-05-13 5:37:45 PM15:07 15 code-mixing 15 16 anti-causatives (sometimes also called, quite separate two language systems, which gener- infelicitously, ‘pseudo-reflexives’ or ‘pseudo- ally overlap; it is not motivated by any evident passives’) are inflected in the middle; see e.g. communicative function. Structural phenomena Rijksbaron (2002:151fff.), Allan (2003:2, 60fff., 82fff. of CM have been researched extensively in the et passim). For some such pairs in the → pres- last decade, which has provided a classifijica- ent tense system, the diathesis opposition is tion of typologies of CM processes, namely the accompanied by the alternation of root vowel following three: 1) insertion of linguistic items (o in causatives, e in non-causatives), which from one language into the structure of another; corresponds to the Common Indo-European 2) alternation between structures from diffferent present causative with the sufffijix *-ey e/o- and systems; 3) ‘congruent lexicalization ’ of lexical o-grade in the root (cf. Vedic pāt-áya-ti ‘makes items from diffferent languages into a shared ’ < *pot-eye-ti etc.), as in the case of phobéō grammatical structure (Muysken 2000). (phobô) ‘terrify’—phébomai ‘panic, flee in terror’ Ancient Greek texts evidence cases of CM (cf. ós te kaì álkimon ándra phobeî ‘who terri- which are due to contact among various dialects fijies even the warlike man . . . ’ (Hom.Il . 16.689)— and between the Koine and dialects (→ Language allà kaì autoì hup’ Argeíoisi phébonto ‘but they Contact), and in cases where it was used abroad, themselves were running in fear from the with other languages in highly bilingual areas, Argives’ (Hom. Il. 11.121)), see Lavidas (2009:65fff.). such as and → . Examples of This morphological type, still attested in (Old) CM can be found in areas of . In Germanic and Slavic and very productive in digraphic inscriptions of Kaphizin ( ), the Indo-Iranian , virtually disappeared in Greek (see syllabic a- to-i we-te-i (apò tṓi wétei ‘from the/ Brugmann 1913:360fff.; Marguliés 1930:87fff.; Schw- this year’) formula appears with the preposition yzer 1950:222; Tucker 1990:138fff.; Sihler 1995:504); apó in the phonetic form of the Koine and with according to Tucker (1990:143), even in the case the Cypriot flexional case. Interference between of the handbook example phobéō, the causative the Koine and the Cypriot dialect also appears opposition exists between phobéō and phobéomai in the same inscriptions and produces mixed ‘panic’, not between phobéō and phébomai. Very artifijicial forms such as u-na-po-re-i (unaphorêi) few are also examples of causatives associated ‘offfers’, which results from a of the pre- with other present types, such as nasal, redu- verb of the dialect (on-, un- with apocope) and of plicated or -skō presents (Marguliés 1930:98fff.; the preverb of the Koine (aná). Many interesting Schwyzer 1950:222). Thus, for most present examples are found in → Thessalian. In Larisa causative oppositions, the active morphology is the inscription IG IX, 2, 516 (end 3rd c. BCE), the only marker of the causative meaning, cf. presents a case for CM in the syntagm tàn dè élpō ‘cause to hope, give hope’—élpomai ‘hope’, állan ‘but the other one’ (dè replaces the dia- koimáō ‘make sleep’—koimáomai ‘sleep’, komízō lect particle má), however, in a dialect phonetic ‘carry’—komízomai ‘travel’. context. A Hellenistic inscription from features the articifijial form oidenós (= oudenós [This article has been shortened for the preview.] ‘of no one’), which derives from the incorrect application of the conversion rule according to Leonid Kulikov which Koine -ou (→ genitive morpheme) corre- sponds to Thessalian -oi. In the honorary decree (Tziafalias-Helly 2004–2005) from Larisa (early Code-Mixing 2nd c. BCE), dialectal forms are found above all in nominal and verbal morphology. There are Code-mixing (CM), which is also referred to as also many mixed forms, for example, opeideí intrasentential → code-switching, is a phenom- ‘for’, which results from a fusion of hópei (dia- enon whereby elements belonging to diffferent lect) and epeidḗ (Koine) or hupárkhonsa ‘exist- language systems or sub-systems of the same ing-fem.’ with the feminine present → participle language are included in a single utterance. morpheme, typical of the dialect, added to a This phenomenon, similar to code-switching, lexical item of the Koine. Further examples of is indicative of individual bi- or plurilingual- CM, due to the presence of the Koine and resi- ism. CM appears largely independent of speaker dues of ancient Doric dialects as well as Latin , intent and is caused by the difffijiculty of keeping are documented during late antiquity in Sicily .

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 15 20 4/5/2013 5:37:4501-05-13 PM 15:07 15 16 code-switching 16 CM is also seen in an inscription on a lead cross several code alternations in correspondence from Syracuse (5th/6th c. CE) with the following with the provenance of the proxenos who was syntagm atâs, hō theós, tîs doúlis (‘of (your) ser- to be honored. The second type is found in vant, o Lord’): in the fijirst word, a reduced form of numerous inscriptions that alternate between the diphthong aw-, in accordance with the - → Greek and Latin which come from Greece cism of the entire text, co-exists with the → Doric as well as from other Mediterranean countries, dialectal form of the morpheme. and in inscriptions from that alternate between Greek and various local languages, e.g. Carlo Consani Pisidian and Phrygian (→ Greek And Phrygian). Given the presence of diverse ethnic groups and their geographical distribution, the island of Code-Switching Cyprus has supplied several texts that either use the two varieties of Greek which were difffused “The juxtaposition within the same speech on the island (Cypriot and Koine; Cypriot) and exchange of passages of speech belonging to two Phoenician or Egyptian (→ Greek And Egyptian) diffferent grammatical systems or subsystems” or → Eteocypriot. (Gumperz 1982:59) or code-switching (), with- out taking into consideration the degree of [This article has been shortened for the preview.] diversity of the language systems involved, is the outcome of → language contact and bilingual- Carlo Consani ism. The analysis of a vast repertory of ancient Greek texts provides two types of CS: the fijirst pertains to the language domain of Greece and Cohesion concerns varieties of Greek dialects and their relation to the Hellenistic → Koine; the second The concept of cohesion is closely tied up with type is a result of the external contacts that the that of → coherence . The latter refers to the Greek language had during the course of its long property of a discourse ‘hanging together’ (from history with other languages of the Mediterra- Lat. cohaerēre ‘to stick together’), which may nean basin. Language repertoires which exhibit remain implicit. Cohesion, on the other hand, CS phenomena belonging to the fijirst type are can be regarded as the visible ‘glue’ between the characterized by (i.e., ‘bidialectism ’ parts of a discourse. In other words, cohesion or the use of two functional varieties of the is the explicit linguistic marking of discourse same language, vs. bilingualism which involves coherence. On this view, coherence primarily is two diffferent languages). In fact, the Hellenistic a cognitive phenomenon, whereas cohesion is Koine is a proper standard language, which was linguistic. promoted and difffused by the Macedonian mon- The landmark work on cohesion in English archy at the supra-regional and international is Halliday and Hasan (1976), which defijines the levels, assuming thus a privileged position in concept as the possibilities in a language for mak- relation to various ancient dialects connected ing texts ‘hang together’. The authors argue that with individual city-states. On the other hand, there is cohesion whenever the interpretation the second type of CS phenomena is character- of an element in a discourse is dependent on ized by bi- or plurilingualism, not necessarily by that of another element. Thus, it involves the diglossia. explicit linguistic marking of relations between Examples of the fijirst type are encountered parts of a discourse, as expressed through both in two long inscriptions that alternate between grammar and vocabulary. Such explicit marking Thessalian and the Koine (IG IX 2, 517) and helps hearers/readers to understand a discourse. between → Boeotian and the Koine (IG VII, 3172). According to Halliday and Hasan, the diffferent The one is characterized by a noticeable capac- expressions of cohesion in English are reference , ity to maintain the two separate codes distinct, substitution , ellipsis , conjunction and lexical while the other shows frequent instances of cohesion . → code-mixing. An interesting instance of CS is Halliday and Hasan’s description of cohesion found in the proxeny decrees from Olus (IC I, has been criticized (e.g. Carrell 1982, Sanders XXII, 4a), dated to the 3rd c. BCE, which display et al. 1992, Sanders and Pander Maat 2006) for

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 16 21 4/5/201301-05-13 5:37:46 PM15:07 17 compensatory lengthening 17 18 treating the concept as a necessary condition (usually, adjacent) segment lengthens to com- for the connectedness of a discourse. In Carrell’s pensate for its loss. Ancient Greek manifests view (1982:486), cohesion is not the cause, but three basic instances of CL named 1st, 2nd, and the efffect of coherence. Similarly, Sanders et 3rd, owing to the chronological order of their al. (1992:2–3) point out that cohesive elements appearance. Samuels (2006) argues that the 1st are “important though not necessary features of CL was a Proto-Attic-Ionic change (early 1st mil- discourse; they are linguistic markers, expressing lennium BCE), while the 2nd CL occurred before the underlying conceptual relations that are of a the Attic-Ionic dialect split (ca. 8th century BCE). cognitive nature”. These authors present coher- The 3rd CL happened sometime within the fijirst ence and cohesion as alternative approaches, millennium (Rau 2010:178). but in fact (the students of) both concepts look The 1st CL refers to a set of processes whereby at diffferent phenomena, and can therefore also *h derived from Proto-Greek *s or *j (also sym- be seen as complementary. Investigating cohe- bolized by *y) deleted in consonant clusters with sion means focusing on the linguistic reflections subsequent lengthening of the preceding vowel. of coherence. Tanskanen (2006:7), for instance, The clusters implicated were (Crist 2001:76): adopts this milder view, assuming that cohesion contributes to coherence. (a) *hC clusters: *hm *hn *hl *hr *hw Regarding Ancient Greek, research on cohe- (b) *Ch clusters: *mh *nh *rh *lh *wh sion is most elaborately undertaken in Bakker (c) *nj *rj *wj after *u, *i and *e and Wakker (2009), a collection of essays dealing with several cohesive devices in the language. Notably, vowel lengthening occurred every- The editors explicitly adopt the basic tenets of where save the → Aeolic dialectsLesbian and Halliday and Hasan’s framework, while at the → Thessalian. There, the segment afffected by same time recognizing that a discourse may lengthening was a consonant either preceding or sometimes display coherence without explicit following the deleted segment (Buck 1955:61–62, marking. Some in that volume discuss Ingria 1980:476–478). anaphoric pronouns, complement clauses or particles, all of which are elements that may (1) Prehistoric Lesb./ Att. mark the familiarity of information or help to Thess. structure the discourse. Others focus on the cohesive function of diffferent tenses. Although *éstelsa éstella ésteila ‘set in order, lexical cohesion is sometimes touched upon as arrange-aor.’ well, there does not yet exist any full-fledged *phthérjō phthérrō phtheírō ‘destroy- work on lexical cohesion in Greek. pres.’ Without explicit reference to the concept of *selásnā selánnā selḗnē ‘’ cohesion, however, there are many other works *esmí émmi eimí ‘(I) am’ dealing with grammatical cohesive devices in Greek. Examples are Lallot et al. (2011) on difffer- Note that the orthographic <ει> (here ei) and <ου> ent functions of the → historical present , and the (here ou)—known as spurious —that great number of publications on particles, such were produced through CL (and → contraction) as the monographs by Bäumlein (1861), Dennis- in Attic-Ionic were not pronounced as diph- ton (1954) and Hartung (1832–1833), to mention thongs but as long ẹː/ọː with a tongue position just a few. These works suggest that a wide range higher than and distinct from the original long of functions is performed by the grammatical mid vowels ęː/ǫː (Samuels 2006:2, → Phonetics). features in question. The 2nd CL involves vowel lengthening after the deletion of the consonant n before s, e.g. mélan-s Annemieke Drummen > mélās ‘black’, pán-sa > pâsa ‘all-nom.fem.sg.’. A related process is the loss of n + t,d,th clusters before an s in with subsequent CL Compensatory Lengthening of the previous vowel, e.g. pant-si > pâsi ‘all- dat.pl.’, spend-sō > speísō ‘shall make ’, Compensatory lengthening (CL) is the process penth-somai > peísomai ‘shall sufffer’. However, whereby a segment deletes and a neighboring the deletion of n alone before the si of the dative

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 17 22 4/5/2013 5:37:4601-05-13 PM 15:07 17 18 conjunction reduction 18 plural causes no lengthening, hence, daimon-si > (2) háma dè têi hēmérāi têi pólei prosékeito oúsēi daímosi and not daímōsi ‘the gods-dat.pl.’ (Good- ou megálēi kaì Ø haireî win 1900:20–21). ‘At daybreak he assaulted the , which is not The 3rd CL afffected the digamma w. While a large one, and took (it)’. (Thuc. 7.29); preserved in Mycenaean , w was gradually lost from the onset in the dialects. An onset is the (3) hṑs tṓ g’antibíoisi makhessaménō epéessin position a consonant holds before a tautosyllabic anstḗtēn, Ø lûsan d’agorḕn parà nēusìn Akhaiôn vowel, e.g. in Mycenaean wiriza ‘root’ all of w, r ‘So when the two had made an end of contend- and z are onsets to their corresponding syllables. ing with violent words, they rose, and broke up The deletion of onset w had no efffect when pre- the gathering beside the ships of the ’. ceded by an open syllable or when word-initial, (Il. 1.304–305) e.g. wánax > ánax ‘lord’. It was accompanied by CL, however, if w was post-consonantal. This was In example (2), the two verbs prosékeito and true for Ionic , but not for Attic or Aeolic where haireî share the same direct object, which is no CL arose (Steriade 1982:118). encoded in the fijirst clause through the → dative noun phrase têi pólei (it is indicated as Ø in the (2) Ion. Att. second). Note that the verb hairéō takes the → accusative: this shows that the dative in the odwós oudós odós ‘threshold’ preceding clause must be taken as a real direct xénwos xeînos xénos ‘stranger’ object (Gk. has transitive verbs that take non- accusative objects, Luraghi 2010). Conjunction Finally, it is signifijicant that in some instances reduction afffects constituents which have the of the 2nd and 3rd CL, e.g. émensa > émeina same grammatical relation, and is not sensitive ‘I remained’, stenwós > steinós ‘narrow’, no direct to morphological coding. If one now considers adjacency between the lengthened and deleted the categories expressed by the verbs, another segments existed in the original form; rather, a type of reduction emerges: while the form pros- segment intervened and yet CL ensued. This fact ékeito is a past tense, the form haireî is a present. has been difffijicult to analyze and has stirred inter- This type of reduction, whereby a present fol- esting theoretical discussion. See for instance lows a past or , has been described in Steriade (1982) and Hayes (1989). Kiparsky (1968), where it is argued to be a feature of Proto-Indo-European . Nina Topintzi [This article has been shortened for the preview.] Conjunction Reduction Silvia Luraghi

Conjunction reduction, or → coordination reduc- tion (Harris Delisle 1978), occurs when some Contraction common feature of two coordinated sentences or clauses, which is overtly encoded in the fijirst, Contraction is the coalescence of two adjacent is not repeated in the second. Often, the defijini- vowels into a long vowel or a diphthong . It is one tion of conjunction reduction involves overt vs. of the possible strategies for eliminating hiatus null realization of an argument, as in the case of (→ Diphthongization; → Synizesis). In Ancient the → subject in an English sentence like (1): Greek contractions ensued from loss of inter- vocalic /s/, /j/ and /w/. The term ‘contraction’ (1) I eat and drink. refers to coalescence of vowels word-internally. In Gk. conjunction reduction applies both to the For the same phenomenon in word junctures, subject and to the → direct object. In addition, see → Crasis. grammatical categories can also be subject to Similar vowels coalesce into the correspond- conjunction reduction. The following examples ing long vowel: *kréwaa > kréā ‘meat (nom./ contain various types of conjunction reduction: acc. pl.)’, *pólii > Ion. pólī ‘city (dat. sg.)’, basilêes ‘kings (nom. pl.)’ > Old Att. basilês, *dēlóōsi > dēlôsi ‘to show (3rd pl. subj.)’. e + e and o + o become close-mid vowels /eː/ and /oː/ (spelled

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 18 23 4/5/201301-05-13 5:37:46 PM15:07 19 cypro-minoan syllabary 19 20 ei and ou) in Attic-Ionic and the dialects of the of the second millennium BCE. These syllabaries so-called mitior (→ Doric): *tréjes > *trées seem to be linked in some way with the Minoan > treîs ‘three’, thematic gen. sg. *-o( j)o > -ou. On scripts, which were used mainly on the island the other hand, → Aeolic, → Arcadian and the of . The Cypriot ‘Greek’ (or ‘classic’) syl- dialects of the so-called Doris severior exhibit labaries (→ ) of the fijirst millen- open mid-vowels /εː/ and /ɔː/: três ‘three’, the- nium BCE clearly derive from these Bronze-Age matic gen. sg. -ō. → Analogy may explain some Cypro-Minoan scripts. The main characteristics uncontracted forms: Hērakléēs ‘’ for of the Cypro-Minoan corpuses are the follow- Hēraklês (after gen. Herakléous), Arg. grophées ing (after Olivier 2007; since then, four short ‘secretary (nom. pl.)’ for gropheîs (after gen. sg. Cypro-Minoan (= CM) inscriptions have been grophéos). All things being equal, the more two published: Cadogan et al. 2009): vowels are similar, the more likely they are to Cypro-Minoan 0: One clay tablet found in coalesce: *histáāsi > histâsi ‘to set up (3rd pl. (Cyprus ); 23 signs in total; 20 or 21 dif- pres.)’ vs. tithéasi ‘to place (3rd pl. pres.)’, didóasi ferent signs; the text is too short to venture an ‘to give (3rd pl. pres.)’; pléete > pleîte ‘to sail (2nd estimation of its signary’s total number of signs; pl. pres.)’ vs. pléomen ‘to sail (1st pl. pres.)’. dated not later than 1525/1500–1425/1415 BCE. Contraction of dissimilar vowels is governed Eight signs are common to CM 0 and CM 1–3. by various factors. In → Attic ĕ + ă and ă + ĕ This script is frequently considered as an archaic behave diffferently: génea > génē ‘race (nom./ form of the other Cypro-Minoan syllabaries. acc. pl.)’, but *eníkae > eníkā ‘to win (3rd sg. Cypro-Minoan 1: 204 inscriptions found in imperf.)’. However, ă + ŏ /ō and ŏ/ō + ă always Cyprus (whole island; written on clay, ivory, contract into ō. In other dialects the length of a metal, stone, glass); 1079 syllabograms in total; plays a major role. When contracted, ă + ĕ and ĕ 72 diffferent syllabograms—ca 77 according to the + ă yield ē, and ă + ŏ and ŏ + ă yield ō in Doric: Mackay formula ; theoretically dated 1600/1575– *eníkae > eníkē ‘to win (3rd sg. imperf.)’, basiléa 850, but more likely to 15th/14th c.–950 BCE. > Arg. basilê ‘king (acc. sg.)’, p.n. Sawo- > Sō-, There are 45 and 41 CM 1 syllabograms common Delph. *awóa > aô ‘dawn (acc. sg.)’. However, ā + to the CM 2 and CM 3 respectively. ĕ or ŏ / ō and ŏ + ā and ĕ + ā become ā : *āwélios Cypro-Minoan 2: Three clay tablets found in > hálios ‘sun’, gen. sg. of masc. in -ās -āo > -ā, gen. Enkomi (Cyprus); 1369 syllabograms in total; pl. -ā́ōn > - ân, Rhod. Boādrómios > Bādrómios 61 diffferent syllabograms—ca 64 according to (month’s name), Selinuntian kréā > krâ ‘meat the Mackay formula; dated no later than 1190— (nom./acc. pl.)’. When contracted, ŏ + ĕ become 1125/1100 and 1125/1100–1050 BCE; 34 syllabo- /ɔː/ or /oː/: *dóhelos > doûlos or dôlos ‘slave’. grams are common to the CM 2 and CM 3. Contraction of ĕ + ŏ is mainly Attic: philéomen > Cypro-Minoan 3: 8 clay tablets found in Ugarit , philoûmen ‘to (1st pl. pres.)’. modern Ras Shamra (Syria ); 253 syllabograms in The contraction of a, e, o + i/u yields a → diph- total; 50 diffferent syllabograms—ca 62 accord- thong: *h₁es(s)i > *ehi > eî ‘to be (2nd sg. pres.)’, ing to the Mackay formula; dated 1320–1190 and *h₁su > * ehu > eû ‘well’. a or o + diphthong 1190–1125/1100 BCE. CM 3 is generally supposed become a long diphthong: *erōtáei > Dor. erōtêi, to be a variant of other CM scripts, but there are Att.-Ion. erōtâi ‘to ask (3rd sg. pres.)’. However, good arguments for considering it as a distinct e or o + ei or oi is always -ei and -oi: *philéei > system. 9 and 16 CM 3 syllabograms are absent phileî ‘to love (3rd sg. pres.)’, *philéoi > philoî ‘to from CM 1 and CM 2 respectively. Since CM 1 and love (3rd sg. opt.)’. CM 2 corpora are four and fijive times larger than CM 3, these absences seem highly signifijicant. It [This article has been shortened for the preview.] is especially impressive that no less than seven CM 3 syllabograms are totally unknown in both Alcorac Alonso Déniz CM 1 and CM 2.

[This article has been shortened for the preview.] Cypro-Minoan Syllabary Yves Duhoux Cypro-Minoan is the modern name given to sev- eral diffferent but related Cypriot syllabic scripts

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 19 24 4/5/2013 5:37:4601-05-13 PM 15:07 19 20 derveni papyrus 20 Derveni Papyrus informal and careless, its most prominent char- acteristic being the very frequent use of asyn- A carbonized papyrus-roll found in 1962 near deton. Irrespective of the author’s identity or Derveni , a site about 10 km northwest of Thessa- his/her real intent, the main interest of the work loniki , Greece, among the debris from a funeral lies in the fact that its discovery in the Derveni pyre that had been strewn over the slabs cover- papyrus reveals a hitherto unsuspected though ing a tomb. The tomb has been dated to the end insolubly puzzling offfshoot of the 5th c. BCE of the 4th or the beginning of the 3rd c. BCE, the Parmenidean revolution in . terminus ante quem for the papyrus, whose script Theokritos Kouremenos can be cautiously dated to between 340 and 320 BCE (Kouremenos et al. 2006:8–9). The papyrus, the oldest Greek manuscript Desideratives found to date, preserves substantial parts of a hitherto unknown prose work by an unidentifiji- Desideratives are secondary derivatives from able author. It is a detailed and fanciful alleg- verbal and nominal roots displaying stems con- oresis of a hexametric poem ascribed to the sisting of two thematic formants (i) -i-á and (ii) mythical singer and . The poem, -s-ei. Both are documented already in Homer a , is interpreted by the author as a (Iliad): glauk-i-á-ei ‘glares fijiercely’ (of a ), coded cosmology built on the same general prin- op-s-eí-ō ‘I wish to see’ (based on the sigmatic ciples as the physical advanced in the future óp-s-omai ‘I will see’). Their lists are avail- wake of of Elea by of able in Schwyzer (1939–1950:732 and 789). Acragas, of Clazomenae, the atom- (i) A closer look at those formed by -i-á allows ists, and of (Kouremenos us to distinguish two closely related sub-cate- et al. 2006:28–45). Such a work can be plausibly gories: assumed to have been composed in the 2nd half of the 5th or at the beginning of the 4th c. BCE. – those expressing strong desire or emotions We have no way of knowing if it continued in (Schwyzer’s “krankhafter Trieb”) one or more rolls or what the purpose of the – those describing bodily conditions: allegoresis might have been. The author’s Greek belongs to the Ionic-Attic dialectal group. In the (i.a) surviving fragments of the work some prominent Ionic features are unexceptionally preferred over their Attic alternatives, but there is a remark- khēzēt-i-á-ō ‘to be eager khéz-ō ‘to ease oneself ’ able inconsistency with regard to other features, to ease oneself ’ occasionally even in the same line or sentence, binēt-i-á-ō ‘to want to biné-ō ‘copulate’ whereas, on the other hand, there are quite a copulate’ few instances of unexceptional adherence to ōnēt-i-á-ō ‘to be eager ōnēt-ós ‘bought’ Attic usage (see Kouremenos et al. 2006:11–14). to buy’ Among the prominent → Ionic features consis- tently present in the surviving parts of the text (i.b) are: -ss- for Att. -tt-; gínesthai/ginṓskein for Att. gígnesthai ‘become’ /gignṓskein ‘know’. But we ophthalm-i-á-ō ‘to have ophthalm-í-a fijind -ē and -ā after r, e and i (cf. → Attic Rever- diseased eyes’ ‘ophthalmia’ sion); uncontracted as well as contracted -ea; ōkhr-(i)-á-ō ‘to be pale’ ōkhr-ó-s ‘pale’ eṓn and ṓn (masc. ptc. ‘being’); o -stem dat. pl. eruthr-i-á-ō ‘to be apt to eruthr-ós ‘red’ -oisin and -ois; ā -stem gen. pl. -eōn, possibly blush’ -ōn, too; ḗn occurs only once, but the Att. eán lith-i-á-ō ‘to sufffer from líth-os ‘stone’ ‘if ’ might occur in an unplaced fragment. Nota- stone’ ble unexceptional deviations from strict Ionic ilingiáō ‘to be(come) ílingos ‘vertigo’ include: mónos, not moûnos ‘alone’; héneken, not dizzy’ heíneken ‘on account of ’; aeí, not aieí ‘always’; splēn-i-á-ō ‘to have an splḗn ‘spleen’ adelphḗ, not adelpheḗ ‘sister’; thaumázein, not enlarged spleen’ thōmázein ‘wonder, marvel’. The author’s style is

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 20 25 4/5/201301-05-13 5:37:46 PM15:07 21 eteocretan 21 22 There are also formations in -a (phon-á-ō ‘to The six undisputed Eteocretan inscriptions be murderous’ versus phon-eú-ō ‘to kill’) or the date between ca 650 (?) and the 3rd or 2nd formants -i-á can be added to the stem of the c. BCE. All of them are fragmentary and their sigmatic future (klau-s-i-á-ō ‘to desire to weep’) texts are short—their grand total amounts just as in tò thúrion klausiâi ‘the door is like to weep’ 422 letters, less than approximately six printed (Aristophanes ). lines of today (!). Moreover, their internal analy- Some examples for (ii) include dra-s-eí-ei ‘he sis is greatly hindered by their scarce use of is going to do’ () versus drá-s-ei ‘he dividers. The best-identifijied word is the Eteo- will do’; -s-eí-ei ‘he is ready to laugh’ (Plato ) cretan form of the name of Praisos, which is versus gelá-s-e-tai ‘he will laugh’. In the later spelled -phraiso-. A sequence -komn- is written language this formation (called ‘Doric future ’) three times. It also appears in the Hellenized was understood as possessing simple future time month’s name Komnokários in the Eteocretan reference: klau-s-eí-tai ‘he will weep’ (next to the city of , but we are in no position to assess ordinary sigmatic future klaú-s-e-tai). Outside its meaning. The same is true for autonomous Attic and Ionic this formation is found above sequences like barze, et, inai, lmo or men. The all in → Doric dialects: ‘strict’ Doric (Crete ) Eteocretan spelling rules opted for the omission prak-s-í-omen ‘we will do’ (Attic prák-s-o-men), of many vowels, as shown for instance by the speu-s-í-ō ‘I will hurry’ (Attic speu-s-é-ō < *speud- sequence -stnmt- (division unknown). It is, then, s-é-ō), ‘mild’ Doric ( ) apo-dō-s-eû-nti ‘they no wonder that the identifijication of the Eteocre- will give back’ (Attic dō-s-oû-si < * dō-s-é-o-nti) tan language has proved to be an extremely difffiji- with regressive (eo > ) and progressive height cult task. Several solutions have been proposed, → dissimilation (eo > eu), respectively, see Bube- such as, for instance, Balto-Finnish , Greek, Hit- nik (1983:65–8). tite , an independent Indo-European language akin to Venetic , a mixed idiom, Phrygian or [This article has been shortened for the preview.] Semitic . None of these proposals has succeeded to convince the scholarly community. What Vit Bubenik seems reasonably certain is that although the Eteocretan inscriptions are written in the their language is clearly not Greek. Eteocretan [This article has been shortened for the preview.] ‘Eteocretans’, Eteókrētes, is the name given by the ancient Greeks to a group of Cretan people. Yves Duhoux The meaning of the Greek word is ‘true Cretans’, implying that the Eteocretans were supposed to be the oldest inhabitants of the island. The Formulas fijirst mention of the word ‘Eteocretans’ that we know appears in Homer (Od. 19.175). The Greek The term ‘formula’ is broadly used to denote the tradition adds that they were , viz. verbal building blocks of ancient Greek poetry, people speaking a language other than Greek. primarily of the epic verse Eteocretans were supposed to dwell in the south- form. Formulas are the standardized phrases ern part of Crete, especially in Praisos , ca 20 km that were ‘stitched’ together—to use an ancient southwards of Sitia. metaphor—by epic poets to describe the typical In 1884, an inscription in Greek letters but writ- characters and objects and to narrate the typi- ten in a non-Hellenic language was unearthed cal actions of heroic epic: ‘rosy-fijingered Dawn’; in the ruins of the ancient city of Praisos. Four ‘dark-prowed ships’; ‘standing opposite him he other similar texts were discovered in the same spoke . . .’; ‘he fell with a thud and his armor place, and still later a sixth one was found in rattled about him’. Concentrated study of the the small town of Dreros (near , in the essential nature of the formula was the key to gulf of Mirabello; this damaged inscription could determining that the Iliad and the , and, perhaps be bilingual). This proves that the Eteo- by extension, all early Greek epic verse, were cretans were actually not concentrated in the fundamentally oral in nature: composed by a southern part of Crete, but in its oriental half. poet who did not know how to write, performed

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 21 26 4/5/2013 5:37:4601-05-13 PM 15:07 21 22 glides 22 for an audience that did not know how to read, /i/ and /u/ and a following mid or low → vowel, and transmitted for through an cf. French plier /pliˈ(j)e/. Glides have no pho- oral rather than manuscript tradition. Milman nological status in Ancient Greek and they are Parry , who was at the center of this concentrated not usually represented in writing. However, study, defijined the formula in his 1928 thesis as glide-notation appears sometimes in early Greek “an expression regularly used, under the same inscriptions. metrical conditions, to express an essential idea” A [ j] glide, spelled with iôta , is found in the (Parry 1971:13 and 272). The story of the evolution Ion. p.n. Diiophánēs and in Sicyonian Sekuwṓniios of his research offfers an excellent prism through ‘ (gen.) ’. Such spellings typically occur in which to examine the of the Homeric Greek regions like Argos (→ Argolic), Tiryns and formula. → : Arg. Poliiádi ‘(Athena) Polias (dat. It had been recognized, even from the earli- sg.)’, Pamph. diiá ‘through’, hiiaroîsi ‘the holy est period, that there was something unique ones (dat.)’. Similar spellings also occur between about the style of Homeric epic . The language a → diphthong in /j/ and a following vowel: Arg. was notoriously repetitive. Highly ornamental Athanaíias ‘Athena (gen.)’, Karneíias ‘Karneia epithets were attached to all the main charac- (gen.)’, Tirynthian aliiaíian ‘assembly (acc.)’. ters: ‘swift-footed ’ occurs 33 times in the A [w] glide, written with waû , appears Iliad; ‘much-sufffering ’ occurs 37 times between /u/ and a following vowel: El. dúwo ‘two’, in the Odyssey. Half- and whole-verse formu- kathúwēn ‘to sacrifijice’, Sicyonian Sekuwṓniios ‘of described the most common actions: ‘So he Sicyon’. In Euboean , where was pronounced spoke, and all of them were stricken with silence’ /y/ or /yː/, the in dúwe, dúwo ‘two’ probably occurs 10 times in the Iliad, 5 times in the Odys- stands for a glide [ɥ]. It is also attested between a sey; ‘They put their hands to the good things diphthong /ew/ and a following vowel in the Cor. that lay ready before them’ occurs 3 times in the p. n. Eúwarkhos and in Cypr. ka-te-se-ke-u-wa-se / Iliad, 11 times in the Odyssey. Entire speeches kateskeúwase/ ‘he built (aor.)’. In late → Laconian were repeated almost verbatim: ’s appears for after a diphthong /ew/: promise of rewards to Achilles in Iliad 9.122–57 p.n. Eubámeros = Att. Euḗmeros, p.n. Eubáberos and again in 9.264–99. Indeed not just words (= euáeros ‘with good air’ < *eu + áweros). Cret. and phrases but entire scenes were very stereo- tawûros ‘bull’, Att. awutár ‘but’, and Eub. awutós typical in nature, with close verbal and struc- ‘himself (nom.)’ cannot represent bisyllabic tural similarities, especially scenes that narrated [awu] with a glide, but are likely to reflect a com- frequently occurring activities in the epics: arm- bination of the two possible archaic spellings of ing for battle (Il. 3.328–38; 11.15–46; 16.130–44; the diphthong, tawros and taûros. It has been 19.364–91); preparation of feasts (Od. 1.136–40; suggested that a [w] glide may have appeared 4.52–6; 7.172–6; [10.368–72]; 15.135–9; 17.91–5); as between /a/ and /o/ in Corinthian p.n. Tlasíawo well as sacrifijice, libation, dressing, bathing, bed- ‘Tlasias (gen. sg.)’, Potedáwōn ‘Poseidon’, but the preparation, departure by ship, arrival by ship, former is probably analogical, based on the gen. decision-making, and so forth. Homer could sg. of the compound p.n. in -lawo (= Ion./Att. draw upon a rich inherited tradition of poetry -leō) and the latter, on Paiáwōn ‘’. that provided for him the very words (epithets, formulas, type-scenes) that were the building [This article has been shortened for the preview.] blocks of epic verse. Alcorac Alonso Déniz [This article has been shortened for the preview.] Steve Reece Greek Loanwords In Slavic

There are about 150 loanwords in the Old Church Glides Slavonic corpus of the texts of the 9th–11th c. CE. Prominent among them are (i) theological terms, Although the term ‘glide’ can be used as a syn- (ii) names of professions, and those reflecting onym of → semivowel, in a narrower sense it (iii) biblical realia. After the 11th c. Greek borrow- describes the transitional semivowel between ings and loan translations appear in hundreds, in

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 22 27 4/5/201301-05-13 5:37:46 PM15:07 23 heteroclitics 23 24 correspondence with the needs of the fast devel- Heteroclitics oping ecclesiastic and secular literature. Most of the texts are translations A combination of r- and n- stems is found in into the Macedonian Slav dialect made most a small group of Indo-European neuter nouns likely during the 863–865 by two bilin- called heteroclitics. The r-stem appears only gual Greek missionaries, Constantine (Cyril as a in the nom./acc., the n-stem elsewhere, as in monk) and his brother Methodius , natives of the Latin femur ‘thigh’, gen. feminis (without any city of Thessaloniki . Their work was continued in other IE languages) and well known by their disciples after 885 in and Cen- examples such as Greek údōr ‘water’, gen. tral (Moravia and Pannonia ). údat-os from *údn̥t-os, with syllabic -n̥- The extant corpus of the 9th–11th c. includes from PIE *u̯ód-r̥, gen. *u̯ód-n-os. The insertion biblical translations, homilies, lives of saints, of -t- must have taken place during the pre- and others. Their somewhat articfiji- historical period. cial language was intended from the very begin- Their exhaustive list can be found in Schwyzer ning for liturgical purposes and the translators (1939–1950:517–21). Various proposals regard- behaved as fijidi interpretes of the Greek originals ing their origin and distribution are available in (very much as the earlier translators into Arme- earlier studies by Petersson (1922), Benveniste nian and Syriac , cf. Brock 2001). In their transla- (1935), Shields (1979), and more recently by tional technique (verbum e verbo ‘word by word’) Friedman (1999). The following list of cognates they reproduced many morphosyntactic gram- in Hittite , Sanskrit , Greek and Latin presents the matical categories of the New Testament and words with heteroclitic cognates in three or at Greek (such as the progressive aspect least two daughter languages, i.e. those with the ópou ên o Iōánnēs baptízōn > idě že bě highest probability of IE ancestry. Iōannŭ krestę ‘where Ioannes was baptizing’, and The evidence of other IE languages allows us to absolute constructions ), they rendered consis- reconstruct more heteroclitics for PIE, but their tently Greek particles (gár > bo ‘for, because’, dé > Greek cognates are not heteroclitic; some are ath- že ‘and, now, but’) and adhered as much as pos- ematic r-stems and some are thematized (ro- and sible to the Greek word order. no- formations, → , Stem Forma- tion). For instance, the PIE heteroclitic *sw̯óp- (i) Theological terms r/n- ‘sleep’ can be reconstructed on the basis of Gk. úpar and úp-no-s (thematized), Ved. anŭngelŭ/angelŭ < ángelos ‘angel’ sváp-na-s and Hitt. suppar-iya-. There is another ijerei < (h) iereús ‘priest’ PIE expression for ‘excrement’ reconstructible as olokavŭtōma < olokaútōma ‘holocaust’ heteroclitic based on the exclusive evidence of skanŭdalŭ < skándalon ‘scandal’ Skt. śákr̥t, śakn-ás, but its Gk. kóp-ro-s is (ii) Professions thematized. The reconstruction of the PIE word for ‘feather, wing’ rests on the exclusive evidence arkhitektonŭ < arkhitéktōn ‘architect’ of Hitt. pattar, paddan-as, while there are ro- and arkhitriklinŭ < arkhitríklinos ‘table master’ no- formations elsewhere: Skt. pát-ra-m, Gk. pte- gnaphei < gnapheús ‘a fuller’ ró-n, OIr. én ‘bird’. There is no direct evidence (iii) Biblical realia for heteroclisis in one of the PIE reconstructions for ‘hand’, *méH-r/n (Mallory & Adams 2006:181), akridŭ < akrίs ‘locust’ based on cognates in Lat. manus, Gk. márē ‘hand’ aromatŭ < árōma ‘perfume’ (if indeed related), Hitt. māniaḫḫ- ‘hand over’ kitŭ < kétos ‘sea monster, whale’ and Alb. marr ‘take, grasp’ (< *mar-n-(y)e/o-), cf. Hom. márptō ‘catch, seize’. [This article has been shortened for the preview.] [This article has been shortened for the preview.] Vit Bubenik Vit Bubenik

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 23 28 4/5/2013 5:37:4601-05-13 PM 15:07 23 24 internal reconstruction 24 Table 1: Heteroclitics Hittite Sanskrit Greek Latin PIE ‘’ nom. yákrt̥ ȇpar iecur *iēku̯ ̯-r/n- gen. yakn-ás ḗpat-os iecin- ‘blood’ nom. ēsḫar ásrk̥ êar, (e)îar aser, assyr *esh₂-r/n- ‘excrement’ nom. sakkar skôr *sóḱ-r/n- gen. sakn-as skat-ós ‘udder’ nom. (udne ?) ūdhar oûthar uber *uHdh-er/n- gen. udhn-ás oúthat-os ‘water’ nom. wātar (udán) údōr *u̯ód-r/n- gen. witen-as udn-ás údat-os

Internal Reconstruction of language change, e.g. the frequency of sound changes attested in languages in generalː the 1. Introductory remarks alternation in Old Church Slavonic nom. bogъ : voc. bože ‘god’ could in principle be reduced Reconstruction implies the reversion of lan- either to an original paradigm *bogŭ : boge or guage history and the reduction of diversity to *božŭ : bože. Apart from our historical knowl- identity. While external reconstruction based edge, the former is advisable as assimilation on the → comparative method uses diffferent, of stops to the palatal quality of a following but genetically related languages or dialects to vowel and subsequent afffricatization and frica- reconstruct an earlier stage from which the sub- tivization is a common sound change. Narrowed sequent forms of the daughter languages have down to the individual language, this typologi- evolved, internal reconstruction (IR) seeks to cally informed way of reasoning could be called reduce alternations attested in one language to the “language consistency principle ”: IR should a uniform previous pattern by defijining them as aim at reconstructing features consistent with the result of a split development. the structural characteristics of the language and Beside the comparative method (CM) as the its cognate languages. most important tool of reconstruction in histori- [ ] cal linguistics , IR was used successfully already This article has been shortened for the preview. towards the end of the 19th c., most famously by Daniel Kölligan F. de Saussure in his reconstruction of the Proto- Indo-European laryngeals (de Saussure 1879, see below; → Laryngeal Changes). While CM inter- Labiovelars prets data from diffferent languages or dialects as indicators of historical developments, IR uses Myc. inscriptions show that the three labiovelars data from a single language to make assertions k w, g w, and g wh inherited from Proto-Indo- about an earlier state of afffairs. The crucial con- European were still distinct from other velars cept involved is that of alternation, usually pat- and from labials in most environments during terns of distribution on the phonological and the second millennium BCE. They were rep- morphological level which are interpreted as the resented with the letter qoppa, written Ϙ. A result of language change . Like CM, IR relies on notable exception was when adjacent to [u]; in the notion of the regularity of language change such instances, the labiovelar unrounded, as it leading to a rule-based set of alternations that also did before [ j]. Both of these developments it seeks to reduce to a single pattern (phoneme, took place in Proto-Greek (→ Proto-Greek And morpheme, etc.) in the prehistory of the lan- Common Greek) (Stephens & Woodard 1986, guage. Both methods rely on basic principles Woodard 1997). such as Occam’s razor (prefer the simplest analy- sis possible) and takes into account the typology

EAGLL_preview_v2.inddEAGLL_sample_1-25.indd 24 29 4/5/201301-05-13 5:37:46 PM15:07 25 merger 25 son or entity (divinity, animal, seldom object) the Myc. Later Gk. PIE speaker is talking to. Its function is to establish qe-to-ro téttares/ *kwetwor ‘four’ and identify the class within a given speech act, téssares e.g. Sṓkrates ‘oh !’. In other words, the qo-u-ko-ro boukólos *gwou-k w ol- ‘cowherd’ vocative functions as a 2nd person marker on e-u-ke-to eúkhetoi *wegwh ‘(s)he proclaims’ nouns, i.e., it is a 2nd person deictic form (→ péssō *pekw-yo ‘I ripen’ (Including 1st and 2nd Person)), since it realizes its reference by linking to the extra-linguistic con- Later, the → labiovelars palatalized or labial- text in which the speech act takes place. ized according to environment. Before [i], the In the Stoic tradition , the term for the voca- development of *kw in all dialects was a dental, tive case is prosagoreutikón , from prosagoreúō while *gw and *gwh yielded labials; one notable ‘to greet’ (Belardi and Cipriano 1990). From exception is kís ‘who?’ from *kwis in Thess. (see the Alexandrian grammarian Dionysius Thrax Stephens & Woodard 1986). (2nd–1st c. BCE) on, the standard term among Greek grammarians is klētikḗ ptôsis ( ), from kaléō tís *kwis ‘who?’ ‘to call’. bíos *gwih3os ‘life’ óphis *ogwhis ‘’ 2. Formal Characteristics

[This article has been shortened for the preview.] The vocative displays specifijic forms only in some masculine and feminine nouns when singular. Bridget Samuels Otherwise, in the dual, plural, as well as in the neuter nouns, it always has the same form as the → nominative. This case syncretism took place at Merger an early stage of Proto-Indo-European : in PIE the vocative plural had the same form as the nomi- Several mergers took place in the consonantal native except for . Vocative forms were system between PIE and the Classical period, atonic except when occurring at the beginning while later mergers transformed the vowel sys- of an utterance or at the beginning of a verse. tem as it developed into Koine. In that case their initial syllable was stressed (cf. Ved. voc. pl. pítaras/ pitaras ‘oh fathers!’ vs [This article has been shortened for the preview.] Ved. nom. pl. pitáras ‘fathers’ (Sihler 1995:250). This prosodic diffferentiation, however, is lost in Bridget Samuels Greek.

[This article has been shortened for the preview.] Vocative Margherita Donati 1. Generalities

The vocative is the (→ Case (Including Syncretism)) used to address the per-

EAGLL_sample_1-25.inddEAGLL_preview_v2.indd 25 30 4/5/2013 5:37:4601-05-13 PM 15:07 25 Contact

If you would like to get in touch with the Brill BRILL publishing team please feel free to contact us. P.O. Box 9000 2300 PA Leiden Jennifer Pavelko, Acquisitions Editor The [email protected] VAT number NL 001539243 B01 Commercial Register Leiden 28000012 For questions about our marketing, please contact Heleen Palmen, Marketing Manager at [email protected].

Where to Order

Journal Orders Worldwide Online Products & Primary Source Collections & Book Orders outside the Americas [email protected] (outside the Americas) [email protected] (in the Americas) BRILL c/o Turpin Distribution Or contact your Library Supplier Stratton Business Park Pegasus Drive For sales information and contact details of our sales Biggleswade representatives, please visit brill.com/services/trade Bedfordshire SG1 8 8TQ For General Order Information and Terms T +44 (0) 1767 604-954 and Conditions please go to F +44 (0) 1767 601-640 [email protected] brill.com

brillonline.com Book Orders in the Americas

BRILL c/o Books International P.O. Box 605 Brill’s VAT registration number is NL8088.46.243 B01 Herndon, VA 2017 2-0605 USA T (800) 337-9255 (toll free, US & only) T +1 (703) 661 -15 85 F +1 (703) 661 -15 01 [email protected] Printed in the Netherlands

All given prices are subject to change without prior notice. Prices do not include VAT (applicable only to residents of the Netherlands and residents of other EU member states without a VAT registration number). Prices do not include shipping & handling. Customers in Canada, USA and Mexico will be charged in US dollars. Please note that due to fluctuations in the exchange rate, the US dollar amounts charged to credit card holders may vary slightly from the prices advertised.

EAGLL_preview_v2.indd 31 01-05-13 15:07 Encyclopedia of Ancient Greek Language and Linguistics

General Editor: Georgios K. Giannakis, Aristotle University of Thessaloniki

Associate Editors: Vit Bubenik, Memorial University of Newfoundland, St John’s, Emilio Crespo, Autonomous University of Madrid, Chris Golston, California State University, Fresno, Alexandra Lianeri, Aristotle University of Thessaloniki, Silvia Luraghi, University of Pavia, and Stephanos Matthaios, Aristotle University of Thessaloniki

brill.com

EAGLL_preview_v2.indd 32 01-05-13 15:07