Greek in Italy

Project Research Blog

Leave a comment

The homecoming of an Odyssey

In the final lines of Hesiod’s Theogony (1011-1016), we are told that Odysseus had two (or three) sons by Circe: Agrios and Latinos (and perhaps Telegonos), who ruled the Tyrsenians (a people in Italy [with the ethnic suffix -ηνός commonly found in Asia Minor and the Levant], if not actually the Etruscans who were later referred to as Tyrsenians/Tyrrenians). The parentage of this Latinus is problematic. Virgil does (Aeneid 12.164) and does not (Aeneid 7.47) present Circe as his mother. Hyginus (Fabulae 127) has Telemachus as his father.

One point of interest, although for another day, is the quantity of the i vowel in the suffix (assuming that it is a suffix): Λατῖνος. This is not the Greek suffix used to form adjectives (of material), such as λίθινος ‘made of stone’.

That is all by way of introducing a post on the Odyssey of Homer, not least because the Teubner edition by M.L. West has recently been published, but mainly because nearly two weeks ago I obtained a copy of the Odyssey as edited by Arthur Platt (Cambridge: Cambridge University Press, 1892), one of the ‘Cambridge Homer’ volumes.

‘The principle on which this edition is made is that of going back as far as is reasonably possible to the original language of Homer’ (Introduction, vii). For Platt, ‘original language’ means that digammas abound and contractions are resolved (infinitives in -εμεν from -ειν before Bucolic diaeresis; aorist 2 infinitives in -έεν from -έειν), among other instances of antiquing. This Odyssey has the complete Vulgate text in order (V 1-42, VIII 266-369, XI 568-629, and XXIII 297-end of XXIV, ‘the wretched conclusion’, are present, but bracketed). This Homer is thoroughly Ionic.

Here is the beginning of Book VIII (lines 11-25 featured in last year’s Part 1B exam and I, at least, have set them for a supervision later this term).

Platt Od.8.1-53

All this is far less radical than the ‘Aeolic Homer’ of Augustus Fick (Ilias, Odysseia). Here is the ‘same’ passage, now a third part (γ) in Part III, the (second) Return of Odysseus.

Fick Od.8.1-28 III nostos Odysseos G.1-27

Digammas abound, but also the original long a-s for the eta of Ionic. Gemination is found for compensatory lengthening. Rosy-fingered Dawn is squarely Aeolic βροδοδάκτυλος Αὔως (line 1) and Odysseus has dual, not plural shoulders (line 19). I shall say no more, but leave readers and my supervisees to consider other points of linguistic interest in these restorations with the help of these visual aids.

For Fick, the Odyssey, or rather Part II, the Revenge of Odysseus, ended thus (cf. Od.23.296, the τέλος): ἀσπάσιοι λέκτροιο παλαίω θέσμον ἴκοντο (‘Glad they approached the assemblage of the old bed.’). (θεσμός was one of my first words for the Cambridge Greek Lexicon Project…).

This volume is no insignificant second-hand outdated book. For starters, I have had in mind to obtain a copy of Platt’s Odyssey ever since my Latin-Greek-Ancient History school master gave me his copy of Platt’s Iliad before I came up to Cambridge (a casket copy, if you will). Therein, Platt simply refers to his Odyssey for an explanation of his approach. Second, the marginal reference to Munro’s Homeric Grammar beside line 48 is not the only annotation.

Platt Odyssey

This copy was bought from W. Heffer & Sons of Cambridge and belonged to none other than the Classical philosopher F.M. Cornford. Complete with a Heffer’s book mark, this Odyssey, bought online from Wigtown, has indeed come home. Not bad for £8 + P&P!


Leave a comment

The Palermo Stone-cutters

A paper that I co-wrote with my colleague, Moreed Arbabzadah, will appear any day now in the next issue of Zeitschrift für Papyrologie und Epigraphik (ZPE or ‘Zippie’): ‘New and Old Interpretations of the Stone-cutters Bilingual Inscription (IG XIV 297 = CIL X 7296) from Palermo’, ZPE 205 (2018) 145–150.

The inscription, depicted below, from Palermo, Sicily shows Greek on the left and Latin on the right and is a typical ‘bilingual bi-version’ (two versions in different languages of the same content). Epigraphic services for sanctuaries and public buildings are advertised in ‘both languages’.

Palermo Stone Cutters

There has been a great deal of discussion about oddities in the Greek and in the Latin alike and what they tell us about the primary language of these stone cutters: was it Greek that prompted odd Latin, Latin that prompted odd Greek, or another language that prompted oddities in the Greek and the Latin alike?

In a ‘work-in-progress’ seminar last February, Moreed suggested that the odd use of cum (here spelled qum) with a genitive (not an ablative) in qum operum publicorum (last two lines  on the right) could be explained as ‘Latin-Latin’ (my term) without recourse to seeing it as the result of interference from Greek (‘Greek-Latin’, my term). The Greek text would then be a translation of the Latin, not vice versa.

I asked about the phrase aidibus sacreis ‘sacred houses’ (three lines up on the right: Classical Latin aedibus sacris), which seemed unproblematic, and its Greek counterpart ναοῖς ἱεροῖς ‘sacred sanctuaries’ (three lines up on the left), which did seem distinctly odd: either ναοῖς or ἱεροῖς alone would adequately reflect aidibus sacreis.

As far as I have found, the various scholars who have discussed this bilingual inscription have not commented on these counterpart phrases.

I suggested that ναοῖς ἱεροῖς was a ‘calque‘ of aidibus sacreis, an element-by-element translation of a phrase from another language and, in this case, one that results in odd Greek and so betrays its origins. Although the general word aedes ‘house’ needs some clarification, neither ναός nor ἱερόν (‘sanctuary’) does. In other words, the Greek text must be a translation of the Latin, not vice versa.

My chief contribution to the paper was to lay the foundations for Moreed’s Latin explanation of the use of cum (oddly with a genitive) by opening up a new argument from this curious Greek phrase for the primacy of the Latin text over the Greek (pp. 145-146). That paves the way for parallels for cum with a genitive in the context of ellipse of a familiar ablative (pp. 147-149). That phenomenon is then along the lines of English ‘I am going to St Paul’s’, in which a genitive ‘St Paul’s’ seems to be the accusative of the goal of motion after the verb, while an accusative, ‘Cathedral’, is readily understood.

To paraphrase A.N. Whitehead, it might seem nowadays that Latin philology is ‘a series of footnotes’ to J.N. Adams. This paper is indeed one such footnote, but, we hope, one that furthers the study of this inscription, of Greek and Latin bilingualism, and of Greek in Italy.

A PDF offprint/Sonderdrucke/separatum of the paper is available on request: please e-mail.

Leave a comment

More on the dangers of diglossia

I am very pleased to see that the Bellum Catilinae of Sallust is back on the schedule of Latin texts for Cambridge undergraduates and I look forward to drawing on it in Linguistics and Philology teaching once this term is underway. (The companion site for a recent edition is a very welcome resource.)

One of Sallust’s purple passages is his ‘portait of a lady’, Sempronia (25). She is largely incidental, it seems, to the work as a whole and to Catiline’s conspiracy. She did provide a venue later on (40.5: ille eos in domum D. Bruti perducit, quod foro propinqua erat neque aliena consili propter Semproniam; nam tum Brutus ab Roma aberat. ‘He led them into the house of Decimus Brutus, because it was next to the forum and not unfamiliar with his plan, because of Sempronia; for at that time Brutus was away from Rome). However, the conspiracy was betrayed by another woman, who gets no such portrait: Fulvia (23.3-4, 26.3, and 28.2). Still, Sempronia has a prominent place in the history of the end of the Roman Republic, as the mother of Decimus Junius ‘Et tu, Brute’ Brutus (cf. 40.5 again).

sed in eis erat Sempronia, quae multa saepe uirilis audaciae facinora commiserat. haec mulier genere atque forma, praeterea uiro liberis satis fortunata fuit; litteris Graecis et Latinis docta, psallere, saltare elegantius quam necesse est probae, multa alia, quae instrumenta luxuriae sunt. sed ei cariora semper omnia quam decus atque pudicitia fuit; pecuniae an famae minus parceret haud facile discerneres; lubido sic accensa ut saepius peteret uiros quam peteretur. sed ea saepe antehac fidem prodiderat, creditum abiurauerat, caedis conscia fuerat: luxuria atque inopia praeceps abierat. uerum ingenium eius haud absurdum: posse uersus facere, iocum mouere, sermone uti uel modesto, uel molli, uel procaci; prorsus multae facetiae multusque lepos inerat.

Now among these women was Sempronia, who had often committed many crimes of masculine daring. In birth and beauty, in her husband also and children, she was abundantly favoured by fortune; well read in the literature of Greece and Rome, able to play the lyre and dance more skilfully than an honest woman need, and having many other accomplishments which minister to voluptuousness. But there was nothing which she held so cheap as modesty and chastity; you could not easily say whether she was less sparing of her money or her honour; her desires were so ardent that she sought men more often than she was sought by them. Even before the time of the conspiracy she had often broken her word, repudiated her debts, been privy to murder; poverty and extravagance combined had driven her headlong. Nevertheless, she was a woman of no mean endowments; she could write verses, bandy jests, and use language which was modest, or tender, or wanton; in fine, she possessed a high degree of wit and of charm. (tr. J.C. Rolfe)

So, her linguistic attainments are reported with some ambivalence. In her defence, she could write verse. However, Sallust slips easily from his comment that she was ‘learned’ (docta) in Greek and Latin literature, not only to lyre-playing and dancing, but to greater ability there than an honest woman needs or should have: psallere, saltare elegantius quam necesse est probae. The infinitive psallere ‘to play the lyre’ is one of a handful of Greek loanwords in Sallust’s Bellum Catilinae and neatly associates Sempronia’s Greek learning with her conduct unbecoming a Roman matron and downright criminal.

(Sallust’s other Greek loanwords are (as far as I know at present): camera [καμάρα] ‘a vaulted roof’ in 55.4; machinor ‘machinate’ in 18.7 and 48.7 (a derivative within Latin from machina, a borrowing from ‘Doric’ μᾱχανά̄ and early enough to show vowel-weaking of unstressed <a> to <i>; then, tetrarches ‘tetrarch’ in 20.7 and toreuma ’embossed or relief work’ in 20.12, both on Catiline’s lips. Then, dolus ‘trick’ in 11.2, 14.5, 26.2, and 28.2 is not clear-cut. It appears as early as the XII Tables and could be either an early borrowing from Greek (and into Oscan, perhaps via Latin) or an inherited word related to doleo ‘I feel pain’. Both an inherited word and a loan would look identical.)

As we saw in an earlier post on the dangers of diglossia, bilingualism and, here, perhaps ‘only’ a reading knowledge of Greek, were not always seen as morally-neutral attainments.

After I wrote that post, James reminded me of Quintus Sulpicius Maximus, the Roman 11-year-old wunderkind, who died shortly after winning an honourable mention, in a contest of 52 poets, for his 43 Greek hexameters on Phaethon (GVI 1924, shortly after 94 CE).

1 Comment

The dangers of diglossia and bilingualism

The Latin word for ‘bilingual’ is bilinguis e. As a formation, it “literally” means ‘having two tongues’, just as the poet Ennius said that he had three hearts  (tria cordia) because he knew how to speak Greek, Oscan, and Latin (Aulus Gellius, Attic Nights XVII 17.1). The adjective bilinguis is one of a sizeable set of compounds that begin with bi– (see OLD, pp. 232-235).

The formal equivalent of bilinguis in Greek is δί-γλωσσος (again, “literally”, ‘having two tongues’, a neat illustration of how stems used in compounds can have the force of a singular or plural or dual). Whence come διγλωσσία and English ‘diglossia’ (‘being bilingual in your own language’, as I was taught) and ‘diglot’, a technical term for a book like a Loeb.

Both bilinguis and δίγλωσσος are not just formal equivalents; they also have the same range of meanings, connotations, and applications.

OLD lists the adjective as a description of ‘things’ with two tongues, of people with two languages, and of people who are ‘double-tongued, deceitful, treacherous’. LSJ has ‘speaking two languages’ for Thucydides and Galen (in his famous discussion of the nature of Koine Greek), but also ‘interpreter, dragoman’ in Plutarch. LSJ then continues ‘double-tongued, deceitful, LXXSi.5.9, al.’ (As ever, there is a question of what ‘al.’ means: officially ‘elsewhere in the same author’. This meaning occurs elsewhere in the LXX and in Siracides at that, but also in other authors, as DGE s.v. II 2 reports. ‘etc.’ would have been appropriate this time.) From DGE, we can add a double-tongued singing cicada (Anth. IX 273.2). Since γλώσσα can be anything tongue-shaped (LSJ s.v. III), doubtless, various objects could be ‘double-tongued’.

The Persian by Plautus has one character describing another as tamquam proserpens bestia est bilinguis et scelestus (‘Like a snake he is evil and has a two-forked tongue’: line 299). Virgil, Aeneid I 661, might be better known: domum timet ambiguam Tyriosque bilinguis (‘she fears the uncertain house and the “bilingual” Tyrians’).

The historian Quintus Curtius Rufus describes the Branchidae as:
mores patrii nondum exoleverant, sed iam bilingues erant, paulatim a domestico externo sermone degeneres.
They had not ceased to follow the customs of their native land, but they were already bilingual, having gradually degenerated from their original language through the influence of a foreign tongue.

History of Alexander the Great VII 5.29

This we would describe as ‘progress’ towards ‘language death’, in the context of language contact and cultural contact. However, we would do so more charitably than Curtius, who labelled the Branchidae as bilingual degeneres. That said, the Branchidae had sided with Xerxes and, to please him, had destroyed the Didymeon sanctuary (VII 5.28). Xerxes had resettled them. In VII 5.33-35, Curtius is more sympathetic to them  as victims of genocide (or more hostile to Alexander).

In some instances, it is clear that treachery (1), not bilingualism (2), is in view, but the two go together in the case of the Branchidae and, more generally, as Rachel Mairs has discussed in  ‘Translator, Traditor: The Interpreter as Traitor in Classical Tradition’, Greece and Rome 58.1 (2011), 64-81.

For (1), consider Didache 2.4, an early Christian text only rediscovered in 1883, and its parallel in the longer-known Epistle of Barnabas 19.7a:

οὐκ ἔσῃ διγνώμων οὐδὲ δίγλωσσος· παγὶς γὰρ θανάτου ἡ διγλωσσία.
You will not be double-minded, nor double-tongued: diglossia, you see, is the snare of death.

For (2), there are, among many other instances, bilingual Carian cities in Diodorus Siculus XI 60.4 (Greek cities with Persian garrisons) and an interpreter in XVII 68.5:

ἐν δὲ τούτοις ἧκεν ἀναγόμενος ἀνὴρ δίγλωττος, εἰδὼς <τὴν μὲν Ἑλληνικὴν καὶ> τὴν Περσικὴν διάλεκτον· οὗτος δὲ ἑαυτὸν ἀπεφαίνετο Λύκιον μὲν εἶναι τὸ γένος, αἰχμάλωτον δὲ γενόμενον ποιμαίνειν κατὰ τὴν ὑποκειμένην ὀρεινὴν ἔτη πλείω· δι’ ἣν αἰτίαν ἔμπειρον γενέσθαι τῆς χώρας καὶ δύνασθαι τὴν δύναμιν ἀγαγεῖν διὰ τῆς καταδένδρου καὶκατόπιν ποιῆσαι τῶν τηρούντων τὰς παρόδους.

Among these came hopefully a man who was bilingual, knowing *<the Greek and> the Persian language. He said that he was a Lycian, had been brought there as a captive, and had pastured goats in these mountains for a number of years. He had come to know the country well and could lead a force of men over a path concealed by bushes and bring them to the rear of the Persians guarding the pass.

* an example of a saut du même au même, an omission caused by skipping from the first occurrence of a word (τὴν ‘the’) to a second occurrence.

The words ‘he said that he was a Lycian’ sound a note of suspicion of (1) here… but, in this instance, that was in Alexander’s favour.

What has all this got to do with Greek in Italy?

Well, apart from the Greek historian Diodorus the Sicilian and then Galen, who was active at Rome in the second half of the third century (at the Imperial Court no less), it is enticing to speculate that the Latin bi– compounds have been influenced, to some extent, by their Greek formal counterparts. It is possible that biurus involves Greek οὐρά ‘tail’. If so, the name that Pliny the Elder reports Cicero as reporting for animalia…, qui uites in Campania erodebant (‘animals…, who would gnaw the vines in Campania’) would be a hybrid Latin-Greek compound. Two other bi– compounds, bilycnis ‘twin-lamped’  and bisyllabus ‘disyllabic, involve words that were Greek in origin (λύχνος and συλλαβή), but had their own currency as Latin words (lychnus and syllaba).

Neither the Didache nor the Epistle of Barnabas have any known connection with Italy, unlike other Greek texts among the so-called Apostolic Fathers (Ignatius wrote to the church in Rome, the letter known as 1 Clement was sent from the church in Rome to the church in Corinth, and the Shepherd of Hermas reports events in Rome and may refer to Cumae at 1.3 and 5.1, as Dindorf conjectured [although the Greek manuscripts have εἰς κώμας ‘into the villages’, one Latin version has apud ciuitatem Ostiorum and apud regionem Cumanorum respectively]). However, Codex Claromontanus (6th c. AD), which contains the letters of St Paul in Greek and Latin and which is thought to have been copied in Sardinia, contains a stichometric list that includes both the Epistle of Barnabas and the Shepherd of Hermas. Two of the leaves of this codex are palimpsest with the Phaethon of Euripides as their undertext (plates I-IV in J. Diggle’s Euripides: Phaethon [Cambridge: Cambridge University Press, 1970]). So, there are various other Greek in Italy connections.

More than that, there is the occasion for this post: editorial work has begun on Migration, Mobility, and Language Contact, Greek in Italy’s volume arising from the 2016 Laurence Seminar of the same name. This volume will include a chapter on interpreters by Rachel Mairs.

Leave a comment

‘Greek in Italy’ in Norfolk


Although a two-week family holiday in North Norfolk was not going to be an epigraphic extravaganza on the same scale as Nick’s visit to Naples, it did provide a reminder to blog about Greek in inscriptions now in Norfolk and some online resources. That said, we happened to see the famous dialect roadsigns that instruct ‘Slow you down!’ (with thanks to the Daily Telegraph).


Some months ago, a friend and colleague asked me about Latin manuscripts at Holkham Hall, of which there are many. At that, I searched Trismegistos, the database of databases for Greek, Latin, and Demotic texts from Egypt (to say the least) for Holkham Hall, as a Trismegistos Collection. The Latin manuscripts fall outside the coverage of Trismegistos, but two Latin inscriptions were reported.

Both are from Rome, both are funerary inscriptions, and both date to early in the second century AD (one may be slightly earlier).


CIL VI 2 14155

CIL VI.3 24008

Thanks to Trismegistos’ own data and its links to Clauss-Slaby (14155 and 24008) and the Electronic Archive of Greek and Latin Epigraphy/the Epigraphic Database Roma (14155 and 24008), full details of the two marbles and the images above can readily be accessed.

Neither inscription involves any Greek sentence, phrases, or script, but there is ‘Greek in Italy’ here nonetheless in the names. 14155.3-5 mentions a Calpurnia Chrysis mater. 24008 reads in full:

D(is) M(anibus) / Petronio / Hedychro / vix(it) an(nos) XXXV m(enses) VI d(ies) VII / Petronia Trophime / conliberto idem / coniugi suo b(ene) m(erenti) fec(it).

‘To the Spirits of the Departed: for Petronius Hedychrus; he lived for 35 years, 6 months, and seven days. Petronia Trophime made <this> for her fellow freedperson and “spouse” alike, who was well-deserving’.

Chrysis is a Greek name (the <ch> and <y> are classic giveaway indications of non-Latin words), as are Hedychrus (<ch> and <y> again – Ἡδύχρους; LGPN omits this bearer) and Trophime (<ph> is one indicator). In Trophime we see also a non-Latin ending. The first-declension nominative singular ends in –a (as in Petronia), but here we have –e as a transliteration of Greek <η>, the equivalent ending for the Greek first declension. One Τροφίμᾱ with -α, the Greek first-declension dialect ending, is also known to the Lexicon of Greek Personal Names.

These three – Goldie, M(aste)r Sweet-Fleshed, and Miss Foster(ed) or Miss Plump – look like slave names, a suggestion corroborated by conliberto ‘fellow freedperson’.

The name Ἡδύχρους (or -χροος) looks like a poetic epithet, as, indeed, it was: GVI 1595.13-14 [Rome, perhaps second century AD]; cf. IG XII, 1 781.4 [Rhodes; second century AD], which has a dative -χροι from a by form in -χρως. However, it was also the name of a perfume. As such, it is also a Greek word known first from a Latin text: Cicero, Tusculan Disputations III 46.

The manuscripts there all have hedyc(r)um (one has aedicrum). Although the <h> and the  <y> were preserved, there is no trace of the <ch>. Whether Cicero spelled it with <c> or with a <ch> that has been lost in transmission, we cannot know. OLD printed the headword as hedycrum with <c>, not <ch>; some texts have the <ch> restored, as LSJ gives it.

Also, –crum indicates that the substantive (like the personal name) was borrowed into Latin from a variety of Greek that has the contraction -χρους, not the uncontracted form -χροος. (Galen has both -χροον and -χρουν).

So much, for now, for Greek in Italy via Norfolk and this experiment with ‘linked data’. I’ll leave it to James to tell the story of the Norfolk clergyperson who argued that the topography of the Iliad was based on his own – and Nelson’s – county.

1 Comment

Hail fellow, well MET: a follow up to ‘Gambling in Greek’

Another thing I saw in the Museum at Naples was the originals of the tavern frescoes from Pompeii that I wrote about James writing about some months ago (here). It’s a bit hard to find good pictures of these, so I’ve put some photos up here (all left-to-right in the original).

In the fourth one (close up directly below), which once showed two fighting drinkers being ejected by the barman, the painted figures have been mostly lost, but their words largely remain. The latest reading for the words on the left is NOXSI. / A ME / TRIA. / ECO / FVI ‘You cheat. 3 was thrown by me. I was (the winner)’. At this time the letter E was often written as two vertical strokes, like this: II. You can see this at the beginning of the fouth line, where ego is written IICO. But in the photo, especially if you zoom in, the second line looks pretty clearly like AMIII: there’s one too many strokes. This is backed up by the early drawings of the inscription, which you can see in Mary Beard’s blog here and here and which pretty clearly also show three strokes.

So what to make of these? The first two letters are pretty clearly AM, but there is no plausible Latin word AMIE or AMEI. James has an excellent suggestion, which is that the last stroke is really the vertical of a T, with the crossbar lost in the damage to the plaster. That would give AMIIT = A MET, still meaning ‘by me’, with the old-fashioned ablative form of the first person pronoun normally spelt med (uncertainty whether to spell final -d in words like sed ‘but’ with -D or -T is common in writing from at least the first century AD). Probably no-one had said med for a good two hundred years by this point, but this use of the old-fashioned form would fit in with the other old-fashioned features I noted in my previous post, like C for G. Seems pretty convincing to me!


A calque ‘under the sea’ ?

… with apologies to fans of The Little Mermaid.

The adjective ὑπό-σαλος ‘under the sea’ is suspicious as a Greek word. It functions, so it seems, as an equivalent to a prepositional phrase such as ὑπὸ τοῦ ἁλός (‘under the sea’, but more likely to mean ‘because of the sea or salt’). The chief cause of suspicion is that Greek ἇλς ἁλός, ‘salt’, did not begin with a [s], at least not in historical Greek. We should find ὕφαλος (see also Buck-Petersen, p. 357 i), as, indeed, we do in Sophocles’ Antigone and subsequently. Indeed, the adjective occurs as a name in the context of ὑπόσάλος.

Before the Greek language was even written down, it lost the sound [s] at the beginning of words and between vowels. By contrast, Latin kept [s] at the beginning of words, so we get sets like ἇλς, Latin sal, and English ‘salt’, ἑπτά, septem, and seven, etc. ([s] between vowels in Latin is a story for another day…). The loss of [s] in both enviroments means that there was no Greek stem sal– on which to build our compound adjective and, if there had been such a stem at the time of that compound’s creation, the [s] between vowels would have been lost subsequently.

Where we find Greek words that begin with a sigma, that sigma cannot be original (one problem is that σῦς, ‘pig’, appears in Homer alongside the expected ὗς; cf. Latin sus, English swine). Explanations include the simplification of a cluster of consonants or that the word in question came into Greek after the loss of initial [s] had ceased to operate. As examples of the latter, consider σιμικίνθιον (semicinctium an apron), σουδάριον (sudarium a towel), and συμψέλια (subsellia seats of a certain kind), words that were borrowed from Latin by Greek speakers alongside what we may call the lexical residue of the Roman Imperial jackboot, or sandal: e.g., κεντυρίων (centurion), λεγιονάριος (legionarius), and πραιτώριον (praetorium).

Here’s the context for that sole instance of ὑπόσαλος as ‘under the sea’ in the Periplus Maris Magni or ‘Voyage around the Great Sea’ (aka Stadiasimos or ‘Measuring by Stades’), as quoted by one third-century Hippolytus in his Chronicle:

(72) Ἀπὸ τοῦ Εὐσχοίνου ἐπὶ τοὺς Ὑφάλους στάδιοι οʹ· νησίον ἐστὶν ὑπόσαλον· ἔχει δὲ καὶ αἰγιαλὸν βαθύν.
‘From Euschoinos to Hypaloi, seventy stades; the islet is under water; and it has a deep (or thick) beach.’

Before we try to explain how the sigma in ὑπόσαλος could be the [s] lost from the Greek equivalent of English ‘salt’ and Latin sal, let us consider whether salt or the sea needs to be involved in the word at all.

The sigma in ὑπόσαλος is so suspicious that the other analysis given in the same LSJ entry is worth considering:

II. shaken underneath, undermined, γῆ Plu.2.434c (ὑπὸ σάλου codd.); ὀδόντες ὑ. loose teeth, Dsc.1.105.5′ [also in Dsc.5.102.2, but that’s the lot] .

On that analysis, the sigma is unimpeachable: it was always there in σάλος, σαλεύω, etc., whatever the origin of the word (Beekes, true to character, classifies it as ‘Pre-Greek’), and, hence, could be justified in an adjective like ἐπίσαλος, which is also used by our Stadiasmos:

(55) Ἀπὸ Ναυσίδος εἰς Πτολεμαΐδα στάδιοι σνʹ· πόλις ἐστὶ μεγίστη· ἐπίσαλός ἐστιν ὁ τόπος, καὶ νῆσον ἔχει· Ἶλος καλεῖται· ἀσφαλίζου.
‘From Nausis to Ptolemais two-hundred and sixty stades: the city is very big; the place is rough (subject to storms), and has an island; it is called Ilos: go carefully!’

Could the islet in (72) — named Ὕφαλοι ! — simply mean ‘shaken underneath’ or ‘undermined’, or even ‘slightly shaken or storm-tossed’, since ὑπο- can add the notion ‘slightly, a bit, somewhat,…’ (trust me, I am/was a lexicographer)?

There are only a few other compound adjectives in -σαλος (ἀ-, εὔ-, and κονί-; very different are βήσαλον and φύσαλος: see Buck-Petersen, p. 359 ii) and none support the idea that the -σαλος part could be the salt sea rather than ‘shaken’.

So, it is likely that since ὑπό with χρῡσός means ‘with gold underneath’ (of the ground), ‘underneath/covered by gold, gilded’ (of an iron ring), or ‘containing a mixture or proportion of gold‘ (so LSJ, for some of the children in Plato’s Republic), ὑπό with σάλος could describe the islet as ‘with turbulence below’, ‘covered by rolling swells’, or ‘somewhat storm tossed’. There might be some sense in that in the Stadiasmos, a guidebook for sailors. If so, we can delete the first section of ὑπόσαλος from LSJ and make an improvement on its Supplement and Revised Supplement.

If not, can we explain why a salty word in Greek has a sigma? Does any help come from the long-term contact between speakers of Greek and speakers of Latin, especially in nautical contexts in texts from the time of the Roman Empire?

Could the sigma reflect a Latin (partial) origin for our ‘under the sea’ adjective? If so, how?

One possibility is that we have a ‘hybrid compound’: one part of the word is from one language, another from another. If that seems strange, it is no stranger than a tele-vision or an auto-mobile  or a hetero-, homo-, or metro-sexual, all of which with a Greek start and a Latin finish. Perhaps, a seafarer who used Latin and Greek with some degree of mixture took Latin sal as the root of a new adjective and then qualified it with an element from his knowledge of Greek word-formation. Real ancient parallels would be ὑπο-καμίσ(ι)ον or ὑπο-καμάσιον ‘a shirt (Latin camisa) that is worn underneath (Greek), an undershirt’ and ὑπο-νοτάριος ‘a notary (Latin) who is underneath or subordinate (Greek), a deputy or sub-notary’.

Can we go a step further and see a ‘calque’ here? A ‘calque’ is an element-by-element translation of a term from one language into another. The Paradebeispiele or ‘Parade-Examples’ (or, in English, the ‘oft-cited examples’, if not ‘examples on parade’) are Modern Greek ουρανοξύστης, German Wolkenkratzer and French gratte-ciel, both based on English ‘skyscraper’, but not necessarily with the elements in the same order as in English.

In our case, the Latin term behind ὑπό-σαλος would be sub-salsus, ‘slightly salty’ (Oxford Latin Dictionary, p. 1848, s.v.), which is found in the medical writer Celsus and the encyclopaedist Pliny the Elder, respectively of water and of a plant. The calquer did not think to drop the [s] in his element-by-element translation: ὑπό- for sub-, Latinate σαλ(ο)-, not ἁλ- for sal, and then adjusting the formation to a Greek pattern (without sigma before the ending). It migth be a problem that σάλος is a noun, but salsus is an adjective. Should we expect ὑπ(ο-σ)άλιος as a calque of sub-salsus (cf. ἐνάλιος)? That said, a compound adjective ending in -αλ-ο-ς would be fine. Indeed, as we saw, ὕφ-αλος is attested — as is ἔναλος.

As a further complication, could the direction of calquing be the other way round: from Greek into Latin? Pliny the Elder is a major source of evidence for us of Latin borrowings of Greek words and of Greek words known only (or chiefly) from Latin authors and their texts. Celsus, as a medical writer, is exactly the kind of borrower and calquer of Greek words that we would expect to meet. (I know nothing about how much Greek influence Celsus actually shows…) An element-by-element translation of the Greek adjective would give a Latin adjective — something like sub-sal_sus. Besides, compounding is so restricted in Latin that Greek is normally involved at some level. I do not know how common it is for Latin sub- to add the notion of ‘slightly’, whereas for ὑπό- it is somewhat ubiquitous. The Oxford Latin Dictionary, p. 1835, s.v. sub- prefix., notes, ‘Before verbs and adjectives it indicates a reduced intensity in the action or quality…’. When and why sub- does so and in which authors are questions for another time. Even if the calquing happened into Latin from Greek, the sigma in ὑπόσαλος would remain unexplained.

Further, if, in ὑπόσαλος, we do have a calque from Latin and if ὑπόσαλος, ‘shaken underneath’, was already in use by the same speakers and/or in nautical registers (the writer of the Stadiasmos also used ἐπίσαλος…), we have a loan-shift or a semantic calque. That is, a Greek word gained a meaning through the use of a Latin counterpart or near-counterpart. For example, the Roman Republican historian Sallust used amare as ‘to be in the habit of’ (as well as ‘to love’), because Greek φιλεῖν could mean ‘to be in the habit of’ as well as ‘to love’. One of my suspicions about the entry in LSJ, second to the sigma between vowels in a salty word, was how the same word could have two meanings that were so different: ‘under the sea’ and ‘shaken underneath’. If one meaning is to do with salt and the  other is to do with σαλός ‘shaking’, we have homonyms and should have two entries.

One final angle: with the meaning ‘under the sea’, ὑπόσαλος would be a ‘prepositional governing compound’, one, that is, that originated in a prepositional phrase. Let us suppose that a bilingual seafarer wanted to say ‘under the sea’ and began in Greek (ὑπό), then started to continue in Latin, either with sale (ablative of sal) or with the dative sali with the genitive salis (because Greek did not have an ablative for him to use), but finished with a Greek ending. Such a hypothetical prepositional phrase would be a peculiar kind of code-switch: a speaker’s (or writer’s) switch from one language to another mid-sentence, mid-phrase, or even mid-word. My favourite code-switch is still that in the title of a paper by Poplack: ‘Sometimes I’ll start a sentence in Spanish Y TERMINO EN ESPAÑOL’.

The last word: as St Paul was on his way to Italy by sea, the Acts of the Aposles 27:14 (a turbulent text) names a wind as εὐρ-ακύλων, which has a counterpart in Latin script in euro-aquilo (CIL VIII 26652: see below) a east-north wind. The first part is Greek, the second Latin (whatever the script). The Bauer-Danker Lexicon quaintly calls it ‘a hybrid formation of Lat.-Gk. sailor’s language’. Was the Greek element ‘borrowed’ into Latin and then the whole borrowed into Greek or was the Latin element borrowed into Greek and then the whole borrowed back into Latin?