Allow me to show you how familiar a whole language can be, namely the infamously tricky Finnish language. There are more Finnish words that you, as an English speaker, can recognise than just sauna.

Fiendish Finnish?

To call Finnish “tricky” is only fair depending on your perspective. It has gained a reputation for difficulty through a general European point of view. Most of Europe today is a patchwork of spoken languages belonging to the Indo-European family, from English and Irish in the west, to Russian and Greek in the east.

The few holes in this Indo-European blanket, like Hungarian and Basque, will not unreasonably seem very different in their vocabulary, grammar and sounds. Finnish is another such gap on the map; faced with Finnish numbers from 1 to 5 (yksi, kaksi, kolme, neljä, viisi), a native-English-speaking learner may flee back to the safety of Indo-European.

From the linguistic viewpoint of an English speaker, Finnish does seem alien and arcane. People for whom it will not seem strange are of course the speakers of Finnish, and also of languages within Finnish’s own family. Clustered around the eastern shores of the Baltic Sea, this family is called Finnic.

It contains roughly eleven members (counting languages is always hard). Two have achieved the hallowed status of national languages: Finnish and, over the water, Estonian. Karelian has official standing in the Republic of Karelia, part of the Russian Federation, but it and other Finnic languages in that country are severely endangered by Russian.

Just as English does, Finnish has its own complex prehistory, stretching back millennia. Its immediate Finnic grouping is itself part of the dense family tree of Uralic. This diverse family includes other well-known languages, such as Hungarian and Sámi.

The family tree also includes many nexuses and ‘proto’-nodes of con-/divergence. The Finnic languages, for example, go back to Proto-Finnic. This was their last common ancestor, which emerged in the second millennium BC.

How exactly the Uralic puzzle should be pieced together continues to keep experts busy. The wide spread of Uralic across Eurasia, the late date of our earliest documents, and external influences from other languages make the Uralic family a much greater challenge than Indo-European. It remains tempting to speculate about a common origin of those two big language families (I’ve been tempted myself), but the signal from that distant speck in prehistory is so faint that the evidence could just be a handful of coincidences.

For our purposes then, let us treat English and Finnish as fundamentally separate in their origins. Consequently, the lack of familiar faces in Finnish vocabulary makes the language very daunting. What then should the would-be learner do? Fear Finnish? Suffer or surrender?

No! Linguistic history comes to the rescue.

Subscribe now

Finnic and Friends

English and Finnish may be part of two entirely separate languages, but there are no hard borders on lexical allegiance. Indo-European languages have for thousands of years given vocabulary to Finnish or to its ancestral states; this is completely normal for any language. Recognising these donations can make Finnish feel much friendlier.

Given the medieval and modern rulers of what is today Finland, we can expect the Finnish lexicon today to contain words of Swedish and Russian origin. These plentiful acquisitions can be useful for English speakers, especially those from Swedish, because it’s a fellow Germanic language. Within this group, in Finnish housut ‘trousers’, halli ‘hall’ and hytti ‘cabin’ you may recognise the English words hose, hall and hut.

Alongside Swedish, a language of socio-political prestige, Germanic languages of trade, like Low German, have also made contributions. Older Slavic speech, prior to what we think of as ‘Russian’, made their mark too. The Finnish city of Turku for example derives its name from an East Slavic word for ‘market’ (in Modern Russian: торг ~ torg). Find a deep dive into that word’s long links here.

In waves of varying size, different languages have been washing up against Finnish/Finnic speakers and leaving vocabulary behind for centuries now – for millennia, even.

This post henceforth concentrates on the earliest transferals, before the medieval era. Specifically, it looks at how the prehistoric ancestor of English influenced the prehistoric ancestor of Finnish. Through their early date of arrival, in the Bronze and Iron Ages, we can find these loanwords across the Finnic languages, not just in Finnish.

Long before writing in any Finnic language slowly emerged (from the 13th century AD onwards), prehistoric Proto-Finnic had absorbed all sorts of words from languages in northern Europe. These include a couple of rare gems from some very ancient Indo-European language (see *lehti ‘leaf’).

Many come from Proto-Baltic, and a great number from Proto-Germanic, the hypothetical ancestor of German, Dutch, Swedish, Norwegian and our own dear English.

Proto-Germanic is said to have had a very strong impact on the development of Proto-Finnic. Its signature can be seen in the number of words of ‘early Germanic’ origin in the vocabulary of a modern language like Finnish. Scholars have reckoned about 500 words, which is a sizable presence considering that Standard Finnish has around 6000 word stems.

The intense lexical contact with Germanic speakers was so intense that it even affected the structure of syllables that Proto-Finnic could have. This has led scholars to describe Late Proto-Finnic, spoken and broken up in the first millennium AD, as “phonologically Germanicized” (Kallio 2012: 234). I will leave the archaeologists to wonder what kind of societal relationship between the two languages’ speakers could be behind this influence.

What then are those words?

A few of them, simply through phonological chance, resemble their English cousins. It’s not hard to see how Finnish kattila ‘pot’ is related to kettle, or how naula and neula are both translated by and cognate with English nail and needle. Likewise, hamppu, haukka, kakku, leipä and valas still resemble hemp, hawk, cake, loaf and whale.

Adjectives too were borrowed; ilkeä means ‘bad’ and is related to ill, just as raaka means and is related to raw. Even functional words were adopted from Proto-Germanic, which is a sign of profound linguistic influence. For example, Finnish sama and English same match in meaning and origin.

Even that most quintessentially Finnish concept, the sauna, may have Germanic ancestry. It’s derived by some from a Proto-Germanic word that has given English stack. If true, this would be one of the many instances of Finnish/Finnic modifying new words’ sounds to fit its own strict phonological patterns.

From Proto-Finnish onwards, the family has lacked the voiced consonants /b, d, g/. It has approximated them with voiceless /p, t, k/ instead. This swapping of sounds is how we etymologists can connect Finnish palje ‘bellows’ to bellows, or patja ‘mattress’ to bed.

Finnic has also had a rule against more than one consonant at the beginning of a syllable. In the case of sauna, this rule meant that original *sta- became simply *sa-. Other Germanic borrowings ran up against this ban, and hence we get Finnish ruoho meaning and being related to English grass. Finnish kaunis ‘beautiful’ is cognate with English sheen and German schön. In this case, Proto-Germanic *skauni– would become Proto-Finnic *kaunis, losing the illegal *s.

The mishmash of examples here are just those I identified with recognisble connections to Modern English. If you bring in Old English or other Germanic languages, in particular the venerable Gothic, the Germanic legacy in Finnish comes to seem very great indeed.

Germanic and Pre-Germanic

Delving deeper still: the rich lode of Germanic is so substantial that linguists can talk about layers of loanwords, like geologically strata. By identifying the presence or absence of certain sound changes, we can pick out the different flavours of Germanic that Finnic has been in contact with. There is a recognisable footprint of Proto-Germanic, as opposed to later Germanic, to be seen.

Famously, Finnic languages have been very conservative in some respects, preserving elements of borrowed Proto-Germanic words that were lost from the Germanic languages themselves.

For example, the assembled evidence tells us that Proto-Germanic nouns could have the masculine singular ending *-az. It used this ending as part of its Indo-European inheritance; *-az was the Germanic counterpart to Latin -us and Ancient Greek -ος. The grammatical ending is no longer there at the end of English king, ring and lamb, but there it is, frozen like a fly in amber, at the end of Finnish kuningas, rengas and lammas.

Finnic has also been finicky about vowels. It has maintained some that linguists have independently reconstructed for Proto-Germanic. For instance, the Modern English word oar goes back to Old English ār. We’d reconstruct it further back as *airō-, with the initial diphthong *ai.

The Finnish for oar? It’s airo. We piece together the Proto-Germanic skeleton from fragments strewn across its descendants, only to find the full fossil perfectly preserved in Finnish!

Relatedly, English woe and one go back to their own ancestral words with *ai, which is there in Finnish vaiva ‘bother’ and ainoa ‘only’. Turning to the kitchen, the vegetable called leek in English and Lauch in German must derive its name from Proto-Germanic *lauka-. Another in the allium genus, the onion, is laukka in Finnish.

Moreover, the word for ‘mattress’ above (patja) preserves the *j sound that English bed and German Bett must have once had, since it triggered the process of i-umlaut before falling silent.

Even more exciting for historical linguists is the idea that Finnish preserves words of ‘Pre-Germanic’ origin. That is to say, there are loanwords that preserved sounds at the time as yet unaffected by sound changes that would later define Proto-Germanic.

One such change was the shift of *ā into *ō. The English words hoof, hen and seek go back to Proto-Germanic words with the vowel *ō. These were *hōfa-, *hanjō- and *sōkjan-.

However, if we assume (as we have good reason to) that these were previously *hāfa-, *hanjā- and *sākjan-, then it looks like it was from their earlier stage that Finnish gets kavio ‘hoof’, kana ‘chicken’ and hakea ‘to get’.

Similarly, the E in rengas above is an example of an *e that shifted into *i in ‘classic’ Proto-Germanic and its descendants (hence, English ring). Plus, there’s the fact that Finnic seems to have taken Germanic words whose consonants consistently look Pre-Grimm.

We Indo-Europeanists know that the initial *h in *hanjō- was previously a *k, and that old stop sound is what we find in Finnish kana. Granted, there are other explanations, but it is another potential pointer to relations with speakers of ‘Pre-Germanic’.

All in all, Finnish and other Finnic languages show evidence of sustained contact with Germanic speakers. This has lasted from the Bronze-Age prehistory of a prehistoric language, right up until the incoming English loanwords of today.

Obligatory concluding point

So, if I have a final takeaway for you, it’s that historical linguistics can break down the walls between languages! Seemingly impenetrable languages have sneaky ways in; an appreciation of history is the key. Treat Finnish as a product of a long story of human interaction, and it will start to seem much more manageable.

You know more about other languages than you think.

END.

Subscribe now

References

• Kallio, P. (2006). On the earliest Slavic loanwords in Finnic. Slavica Helsingiensia, 27, 154-166.

• Kallio, P. (2012). The prehistoric Germanic loanword strata in Finnic. A linguistic map of prehistoric Northern Europe, 225-238.

• Kroonen, G. (2013). Etymological Dictionary of Proto-Germanic (Leiden Indo-European Etymological Dictionary Series 11). Brill.

• Lehtinen, T., & Hakulinen, A. (2012). Finnish. Revue belge de Philologie et d’Histoire, 90(3), 1029-1052.

• Rajki, A. (2009). Finnish Etymological Dictionary. https://www.academia.edu/12788026/Finnish_Etymological_Dictionary

• Ringe, D. A. (2006). From Proto-Indo-European to Proto-Germanic (Vol. 1). Oxford University Press.

Modified version of You Know More Finnish Than You Think, first posted in August 2025 on dannybate.com, which this month celebrates its sixth birthday, and on which this remains its most read article.

Images my own or from Wikimedia. Cover image: Neitokainen, an artificial pond in Finland, shaped like the country. It’s real!

The talk’s topic fitted into my overall research into the word order of ancient Indo-European languages. My task was to acquire, present, evaluate and compare the available evidence, upon which I could then conclude with new findings or insights. So, my audience of Edinburgh experts might have been expecting a presentation concerned with what their junior colleague had come to understand. What they got instead was well-researched frustration about something that I didn’t understand – and still don’t.

So, forgive the reminiscences and selfish return to gone glory days; I have unfinished business with a curious piece of ancient syntax. It saw words dissected and spread around their sentences, its motives unclear. In what follows, I’ll set out what it looks like in our historical texts, before turning to how it theoretically got the better of me.

Dis-secting

Did our linguistic forebears feel flexible in how they could construct sentences? Abso-bloody-lutely.

In our historical sources, older members of the Indo-European family (to which Modern English also belongs) display great variation in the relative order of their words. The permitted permutations make it necessary to speak only of patterns and preferences, rather than hard syntactic rules.

The same flexibility could potentially be applied to an individual word, which might find itself cut and split up. Such lexical dissection is known as tmesis (‘t-me-sis’), from the Ancient Greek for ‘cutting’. It thereby shares a root with anatomy, dichotomy and atom. Occasional instances of tmesis occur in English for humorous emphasis: the aforeused abso-bloody-lutely is one victim of the verbal knife, as are ri-goddamn-diculus and boo-f*cking-hoo.

Higher up the family free, tmesis is sometimes documented in ancient languages like Latin. One famous example is attributed to Ennius, an early writer of the Roman Republic. Among the supposed fragments of his writings is the half-line:

saxō cere comminuit brum

‘With a stone he broke his skull’
(Ennius, Annales frag. 609)

While also slotting his words into a metrical template, Ennius playfully employs tmesis to suit the meaning; he splits the word for ‘skull, brain’ (cerebrum) in two.

Another oft-quoted example of tmesis comes from Sextus Pompeius Festus, a Roman linguo-geek living in the 2nd century AD. Festus documents an artifact of Latin that, to his ears at least, by then sounded old-fashioned:

ob vōs sacrō, in quibusdam precātiōnibus est, prō vōs obsecrō, ut sub vōs plācō, prō supplicō”

‘Ob vōs sacrō, in certain prayers, stands for vōs obsecrō, just as sub vōs plācō stands for supplicō’
(Festus’ epitome of De verborum significatione, O)

What Festus is getting at here is that the words obsecrō (meaning ‘I beseech’) and supplicō (also meaning ‘I beseech’) could be interrupted by the object pronoun vōs ‘you’. The resultant phrases were confined, though, to the archaic language of prayers. An English equivalent would be like inserting us into forgive and deliver in the Lord’s Prayer, and praying ‘for-us-give our sins’ and ‘de-us-liver from evil’.

Festus, a darling of historical linguists today, gives two further examples. This time, they come from the foundational laws laid down in Rome’s early years:

ut in lēgibus: trānsque datō, et endoque plōrātō

‘As in the laws: trānsque datō (‘and he shall surrender’) and endoque plōrātō (‘and he shall implore’)’
(Festus’ epitome of De verborum significatione, S)

The verbs for ‘surrender’ and ‘implore’ are cut in twain by the presence of -que, meaning ‘and’. As with religion, legal language is quite used to preserving linguistic archaisms.

There is a notable difference between Ennius’ assault on cerebrum and Festus’ remembered formulations like ob vōs sacrō. In the second, tmesis is carving at the joints; it makes a clean break within obsecrō, namely between its initial prefix (ob) and the root verb (sacrō). The break is at a morphological boundary within the word.

That break is present within hundreds (thousands?) of English words, as part of its Germanic and Latinate inheritances, although prefixes in English differ as to how lively and malleable they are. We can wield the prefix re- (redo, rewatch, reorder) far more dextrously than the prefix ob- (obvious, oblige, obfuscate).

The same break cannot be seen in cerebrum. Tmesis of such an intact word would’ve felt as strange as dividing English skull or brain and positioning the parts across your line of poetry.

Subscribe now

Re-cognising

Now, here’s the thing: the ob vōs sacrō pattern, in which a prefix and its verb bookend their clause, pops up across the Indo-European family, far beyond Latin.

Take its cousin, Ancient Greek. Its greatest representative, the poet Homer, attests an employment of this pattern that is much livelier than fossilised fragments. Separating a single prefix and putting it at the beginning seems to have been a valid grammatical option for Homer (whoever he/she/they was/were).

ἐν γὰρ Πάτροκλος φόβον ἧκεν ἅπασιν
en gàr Pátroklos phóbon hêken hápasin

‘For Patroclus had sent fear into them all’
(Iliad 16.291)

ἐν δ᾽ ἐρέτας ἐπιτηδὲς ἀγείρομεν, ἐς δ᾽ ἑκατόμβην θείομεν
en d᾽ erétas epitēdès ageíromen, es d᾽ hekatómbēn theíomen

‘Into it let us gather sufficient rowers, and onto it let us put a hecatomb’
(Iliad 1.142–3)

The poet here has cut up verbs like ἐνίημι (eníēmi) ‘to send in’ and ἐναγείρω (enageírō) ‘to gather in’, and shoved their two components to the edges of the clause. While acknowledging the many inevitable exceptions and variants, this ‘P … V’ pattern exemplified by Festus and Homer is common enough to be noteworthy.

Homer crafted his verses according to a strict metrical template, yes, but the frequency of tmesis and the general principle that his works couldn’t be unintelligible gibberish together indicate that this was a genuine part of general Greek grammar. It was at least grammatically acceptable, perhaps sometimes preferable, in everyday speech and among the wider population of speakers.

If so, then it wasn’t to last. Very few post-Homeric examples of this pattern can be found.

κατά με φόνιος Ἀΐδας ἕλοι
katá me phónios Aḯdas héloi

‘May murderous Hades take me’
(Sophocles’ Oedipus at Colonus 1688–9)

Besides, Homer already cast a long shadow that we cannot remove the possibility that Sophocles here was simply imitating the epic poet. From the classical era onwards, we don’t find this pattern, and Greek prefixes remain firmly glued to Greek verbs.

The general trend of this kind of tmesis is therefore a decline, although Homer also appears to be its chronological peak. Curiously, Myceanean Greek (the language behind Linear B, spoken centuries before Homer) to date displays no sure instances of the pattern.

𐀁𐀠𐀆𐀅𐀵
e-pi-de-da-to

‘It has been distributed’
(PY. Vn 20)

In later alphabetic Greek writing, this would be rendered ἐπιδέδασται (epidédastai). Such Mycenaean examples from the second millennium BC only show prefixes like ἐπι- immediately preceding their verbal colleague. They form what we recognise as ‘compound’ verbs.

Our sources are Mycenaean Greek are few in number, inviting the speculation that we simply haven’t yet dug up its examples of the ob vōs sacrō pattern. Nonetheless, what we do have is so consistent in not attesting it. On this basis, it has been suggested that Mycenaean Greek was innovative and ahead of the grammatical curve when it came to compound verbs, compared with the later-but-conservative Ionian dialect of Homer.

Over in India and Central Asia, Vedic Sanskrit, the language of the Vedas and a cousin of Latin and Greek, makes great use of prefix-like elements that are flexible in their position relative to the verb. A compound verb in Vedic Sanskrit may be formed with more than one such element (I believe two is the recorded record in Vedic, but four in Classical Sanskrit). The default position is for that item to stand right before the verb, but there is also a noticeable tendency to place it away, at the very beginning of their common clause.

वि जना॑ञ्छ्या॒वाः शि॑ति॒पादो॑ अख्य॒न्
ví jánāñ chyāvā́ḥ śitipā́do akhyan…

‘The white-footed black horses looked at the people’
(Rigveda 1.35.5)

प्र यो वां॑ मित्रावरुणाजि॒रो दू॒तो अद्र॑वत्
prá yó vām mitrāvaruṇā ajiró dūtó ádravat…

‘The quick messenger who ran forth to you, O Mitra and Varuna’
(Rigveda 8.101.3)

वि (ví) ‘apart, away’, प्र (prá) ‘forward’, परि॑ (pári) ‘around’, नि (ní) ‘down’, परा (párā) ‘away, along’ and others contribute to the meaning of a Vedic verb, often simply by adding the action’s direction. The प्र and अद्र॑वत् in the second example here can together be translated as ‘ran forwards’ or more literally ‘forwards … ran’, but new senses can emerge from the compounding. If two prefix-like elements are present, only one is allowed in this remote position.

पर्यू॒ षु प्र ध॑न्व॒ वाज॑सातये॒
páry ū ṣú prá dhanva vā́jasātaye

‘Around and forwards run well, for winning the prize’
(Rigveda 9.110.1)

Compared with Latin’s fragments and Homeric Greek, the Rigveda exhibits a great deal more arrangements of the ob vōs sacrō pattern.

Vedic Sanskrit’s sister language is Avestan, the oldest representative of the Iranian branch of Indo-European. Its ancient scriptures exhibit a few instances of the ob vōs sacrō pattern, such as with 𐬁 (ā), meaning ‘at, to’. Here, it stands alone and first:

ā-mо̄i rafəδrāi zauuə̄ṇg jasatā

‘Here come to my calls for support’
(Yasna 28.3)

Other venerable great-aunts of the Indo-European family display this preference for preposed prefixes, with Hittite among them. But even centuries later, in the Late Antique/Early Medieval European era, we find tmesis.

For example, in fourth-century Gothic, the oldest thoroughly documented Germanic language, a few instances can be found:

𐍆𐍂𐌰𐌷 𐌹̈𐌽𐌰 𐌲𐌰-𐌿-𐍈𐌰-𐍃𐌴𐍈𐌹
frah ina ga-u-ƕa-sēƕi

‘He asked whether he saw something’
(Mark 8.23)

𐌳𐌹𐌶-𐌿𐌷-𐌸𐌰𐌽-𐍃𐌰𐍄 𐌹̈𐌾𐍉𐍃 𐍂𐌴𐌹𐍂𐍉
diz-uh-þan-sat ijos reiro

‘And then a trembling seized them’
(Mark 16.8)

𐍃𐌰𐌴𐌹 𐌲𐌰𐌻𐌰𐌿𐌱𐌴𐌹𐌸 𐌳𐌿 𐌼𐌹𐍃 𐌸𐌰𐌿𐌷 𐌲𐌰-𐌱𐌰-𐌳𐌰𐌿𐌸𐌽𐌹𐌸 𐌻𐌹𐌱𐌰𐌹𐌳
saei galaubeiþ du mis þauh ga-ba-dauþniþ libaid

‘He who believes in me, though he were dead, will yet live’
(John 11:25)

In these Biblical lines, we see three compound verbs, 𐌲𐌰𐍃𐌰𐌹𐍈𐌰𐌽 (gasaiƕan) ‘to see’, 𐌳𐌹𐍃𐍃𐌹𐍄𐌰𐌽 (dissitan) ‘to seize’ and 𐌲𐌰𐌳𐌰𐌿𐌸𐌽𐌰𐌽 (gadauþnan) ‘to die’, broken up by intervening material. Unlike in Homeric and Vedic verse, this only involves short functional words, like the question particle 𐌿 (u) and the indefinite pronoun 𐍈𐌰 (ƕa), related to English what.

I will offer only these as relevant examples from the Germanic branch, tactfully sidestepping the question of whether the modern phenomenon of phrasal verbs in Germanic (e.g. English lift off, bring on, think up, dress down) is in some way connected to ancient tmesis. I don’t think so, but don’t feel qualified to state that definitively.

Then, finally, we turn to my beloved Old Irish. This medieval language, the ancestor of Modern Irish, Scots Gaelic and Manx, does incredible things with its verbs. Among them, we have a very minor, genre-sensitive pattern that again sees a prefix-like element and its verb stand at a distance.

fordon itge Brigte bet

‘May Brigid’s prayers be upon us’

(Thesaurus Palaeohibernicus II.348.4)

ad cruth caín cichither

‘Fair form will be seen’
(Serglige Con Culainn 694, oft quoted in the literature, including by me, but I now worry it’s a ghost quote)

Once again, we see prefix-like elements (here ad and for) standing removed from the verb. The position of that particle is not the oddity for a change; what’s unusual for Old Irish is that the verb is in the final position. The standard pattern is for the verb to come early on, either first or immediately following an element like ad.

ad·cither a-suthine-som

‘Their eternity is seen’
(Würzburg glosses 1b15)

As with Vedic Sanskrit, these prefixish things are numerous in Old Irish (a stack of several is common), and often their presence is obligatory. The compound verb in the first and third examples, meaning ‘to see’, never appears without the added ad.

Across the branches of Indo-European, we unearth historical usages of the pattern seen in Latin ob vōs sacrō: a prefix-like element, then other material, then the verb.

I have to say “prefix-like”, because really, they’re not prefixes as we know them. To be a prefix implies being fixed – unable to be moved and to function as a standalone word. Our accumulated evidence suggests that this was not the case for prefixes across the oldest Indo-European languages. Instead, they were once all free and independent.

Although today we make a terminological distinction in English between prefixes (e.g. understand, forgo) and prepositions (e.g. under the table, for the book), the members of the two groups tend to look very similar. This is indicative of their original status; in the earliest days of Indo-European, there was no such distinction. All such directional and relational words worked like adverbs. They were accented and were not bound to one position relative to a noun or a verb. Instead, they independently contributed their own particular meaning to a noun, a verb or the sentence as a whole.

This situation endured for centuries and passed into the emerging Indo-European languages; we see their old flexibility most clearly in ancient members like Vedic Sanskrit and Ancient Greek. On this basis, ‘preverb’ or ‘local particle’ are alternative terms used, avoiding the dependent quality of prefix. Out of habit, I will now switch to preverb, although it’s hardly a perfect term. The pre- bit of preverb is based on the words’ typical position before the verb that they cooperate with – typical, but not exception-less.

Over and over, though, we see the old adverbs fusing with verbs into single words (‘univerbating’), becoming prefixes as we today would recognise them. When the Indo-European languages each enter the historical record, the ob vōs sacrō pattern is already rare or not to last. We see it most clearly in the Greek, Sanskrit and Hittite record; in Latin, Gothic and Old Irish, prefixes are predominant and tmesis is a fossil.

From this historical perspective, the term tmesis is actually a misnomer. The opposite process is in fact true; examples like ob vōs sacrō aren’t instances of compound verbs cut up, but rather preserve an older grammar in which prefixes and verbs haven’t yet been stitched together.

Ex-plaining

I can now get to the thing that haunted my otherwise serene PhD days.

What we have with the examples given is evidence for syntactically independent and moveable preverbs. Within that independence, there is a significant tendency for one preverb to appear first in its clause. In the Latin, Ancient Greek, Vedic Sanskrit, Avestan, Gothic and Old Irish examples, the element in question is right at the beginning, preceding all else.

It does so with such frequency and across such a spread of languages that we have the necessary grounds to think that this pattern was part of the word order of their common ancestor. That is, just as we reconstruct its sounds and words, I believe and have argued that we might reconstruct such arrangements of words back to Proto-Indo-European itself.

Put another way, the claim would be that Proto-Indo-European speakers, six-thousand-ish years ago, had a syntactic ‘habit’ of shifting a single one of these directional words (equivalent to English towards, down, out, away) to the outset of their sentences. This we can call ‘preverb fronting’. It endured into several of their prehistoric language’s many offspring, but ultimately gave way to bolted-on prefixes. From its historically documented descendants, we gain indirect evidence for the original habit – note that I can’t formulate it as an obligatory rule.

Any attempt at the reconstruction of prehistoric syntax is fraught with theoretical difficulties. Every piece of historical support for it must be critically examined, to determine whether tmesis/preverb fronting really was an archaic element of their grammar.

An alternative explanation could be that it’s due to metrical reasons and to composers mucking around and making awkward words fit – a possible factor for Homeric Greek and Vedic Sanskrit. Another alternative is interference from and imitation of other languages – a possible factor for Gothic (its Bible being a translation from Greek) and Old Irish (its scholars being steeped in Latin). The business of historical syntax must consider all other potential factors for all relevant examples before proposing that some word-order oddity in our historical texts really ‘was syntactic’ in its cause. A familiar old headache has returned for me.

Nonetheless, preverb fronting avoids these pitfalls. It’s hard to dismiss all the occurrences all as poetic artifice or interference. En masse, a shared syntactic inheritance seems more plausible. It’s too common in Homeric and Vedic not to have been a viable option in the grammar of the community at large, beyond a handful of wordsmiths. Also, the two Gothic examples above do not mirror the Greek originals, and I look very sceptically on the claim that the few instances of Old Irish tmesis were modelled on Latin. Tmesis in Latin was fragmentary, hardly a feature worth copying.

As an aside, I still suspect the opposite was actually true: that preverb fronting was so common in an earlier stage of Old Irish that it was wholly or partly responsible for the standard grammar with the verb placed early on. This ‘Verb Subject Object’ order is now typical of the Celtic languages. In my preferred model, preverbs were so thoroughly fronted at first that eventually their verbs moved to join them.

So, having seen off the theoretical competition, the coast seems clear for us to claim that preverb fronting was indeed a feature of Proto-Indo-European word order. From this prehistoric popularity, it was passed on and ‘inherited’ by later speakers of Latin, Ancient Greek and the rest. This raises the question: what was it for? What was the linguistic benefit of putting these particles first?

I don’t know!

Having proposed it for Proto-Indo-European, I’ve struggled to state why it existed as a pattern within that language’s syntax. This was the question behind my pleading “any answers?” slide. I don’t lack possible explanations; there are several, but individually or collectively they don’t pass muster.

One is to satisfy ‘Wackernagel’s law’, the well-established observation that early Indo-European languages had groups of words that could not stand first in their clause. Specifically, these were object pronouns (meaning ‘me’, ‘you’, ‘us’, ‘something’) and conjunctions (‘and’, ‘but’, ‘for’). It’s generally agreed that they were in some way phonologically deficient, slightly or wholly unaccented, and therefore needed to tag onto the end of another word.

Something – anything – else had to precede such little words, so long as it was a ‘proper’ (i.e. accented) word, which could offer a host that for these clitics to lean on. The phenomenon was first noted for Sanskrit by Berthold Delbrück, then by Jacob Wackernagel across Indo-European more generally. Wackernagelian clauses are readily present in Vedic Sanskrit and Ancient Greek. See below how मा॒ता (mātā́) ‘mother’ and καί (kaí) ‘also’ stand as host to two clitics that follow them:

मा॒ता च॑ मे छदयथः स॒मा
mātā́ ca me chadayathaḥ samā́

‘And you and my mother appear alike’ (Literally: ‘mother and of-me appear alike’
(Rigveda 8.1.6)

καὶ δὲ σοὶ αὐτῷ μοῖρα, θεοῖς ἐπιείκελ᾽ Ἀχιλλεῦ, τείχει ὕπο Τρώων εὐηφενέων ἀπολέσθαι
kaì dè soì autôi moîra, theoîs epieíkel᾽ Akhilleû, teíkhei húpo Trṓōn euēphenéōn apolésthai

‘And for you too, god-like Achilles, is the fate to be destroyed under the wall of the rich Trojans’
(Iliad 23:80–81)

The rule is vaguely comparable with the extra do that n’t needs in English negative sentences like I don’t know. As for my preverbs problem, this explanation would say that the requirements of clitics triggered the fronting of a preverb to act as their host.

As a devotee of Wackernagel’s law, my regretful response is that this is not sufficient, on the basis of the historical evidence. The correlation between the two is not strong enough. Many of our examples of preverb fronting do indeed appear with one of these troublesome little words, but many do not, such as in the first two Rigveda examples herein. We might also predict fronted preverbs in all instances of their co-appearance with clitics, which we do not find. The prefix ἀπο- (apo-) of ἀπολέσθαι (apolésthai) ‘to be destroyed’ in the Greek example above has not detached itself and run to start of the sentence to be the host for its two clitics.

Moreover, it does not explain the perceived positional preferences of preverbs: either right next to the verb at the end, or one alone at the beginning. Nor does it explain itself: why shift a distant preverb to meet the Wackernagelian requirement? Why not simply front the next available word? A variety of hosts is indeed what we see.

Moving away from any obligatory syntactic rule, another explanation is stylistic: that this construction achieved some rhetorical or performative quality, something beneficial in narration but not obligatory. Such was the view of Louis Renou for the construction in Vedic Sanskrit:

“By highlighting the preverb, it has a privileged place in hymns that consist mainly of invocations … The separation is most prominent in stanzas of calling to prayer, where it was customary to mark, by the preverb, the invocation, the initial momentum whose content the rest of the sentence develops.”

(Renou 1933, translation my own)

This sounds good, but doesn’t (yet) convince me. What did a priest or storyteller really gain through a fronted preverb? Why not some other category of word? Far be it from me to move the goalposts and critique a renowned scholar, but this explanation doesn’t dig down to the root of the phenomenon. It does feel like it’s on the right lines, though, and I speculate about some ‘bookending’ rhetorical technique; by placing the two parts of the same verb at either end, it unified a clause for the audience.

A third explanation is simple emphasis. This view would consider the fronted preverb to be important and accordingly emphasised through its initial position, and perhaps heightened volume too. I’m not at all convinced by this. While I have no direct access to spoken Proto-Indo-European, my assessment is that the initial position of its clauses was not characterised by emphasis. Anything put there was not new and important – quite the opposite. It was topical information, already known to all parties in the conversation.

The abstract structure of the Proto-Indo-European clause that my thesis proposed begins with two slots:

[ topic ] [ clause ] … other stuff … [ verb ]

Into the first, speakers put their topical material, that which was familiar and linked the new clause back to things previously mentioned. Into the second went the words that signalled what type of clause it was. If it was a question, you put the question word there; if it was a relative clause, you put the relative word (i.e. ‘which/that/who’) there, for example the यो (yó) ‘who’ in the second Vedic example above.

This structure captures and explains a lot of the data; it works well as the prehistoric starting point for the word-order behaviour that we see in the descendants of Proto-Indo-European. But there remained for me an annoying fly in the ointment: fronted preverbs. There is nothing obviously topical about them (and always just one of them) that would see them suitably slotted into my abstract [ topic ] position.

Con-cluding

I remain open to the possibility that my proposal is flawed and ought to be redesigned or binned. Fronted preverbs may just be the card that collapses the tower. But that would at least depend on us identifying what exactly this word order was for.

After years of frustratingly circular thoughts and fun-but-fruitless chats with linguistic experts down a number of avenues, a convincing explanation so far defeated me. Even writing this enjoyably long post has not provoked new insights. Whether I will be the one to explain it seems less likely with each passing year. I’ve felt for so long that it’s something obvious, something exhibited by languages living today that I’ve missed or am yet to learn about.

So: any answers?

END.

Subscribe now

Resources

• Festus

• Gothic Bible (in the Gothic script here)

• Homer’s Iliad

• Rigveda

• Sophocles’ Oedipus at Colonus

• Thesaurus Palaeohibernicus II

• Würzburg Old Irish glosses

• Yasna (in the Avestan script here)

Un-linked references

• Danesi, S. (2013). Particle-verb constructions in Vedic: The case of Ápa. Studi e saggi linguistici, 51(2). 57–100.

• Hale, M. (1993). Tmesis and movement in Avestan. Indo-Iranian Journal, 36(1), 29–43.

• Giannakis, G. K. (2023). At the crossroads of Linguistics and Philology: The Tmesis-to-Univerbation process in Ancient Greek. In: Giannakis, G. K., Filos, P., Güemes, E. C., & de la Villa, J. (Eds.). Classical philology and linguistics: old themes and new perspectives, Vol. 1. Walter de Gruyter GmbH & Co KG. 175–211.

• Jiménez Delgado, J. M. (2023). Mycenaean Greek syntax: a diachronic perspective. In Aura Jorro, F., Del Freo, M., & Piquero, J. (Eds.)The legacy of Michael Ventris: progress and perspectives in the field of Aegean scripts and Mycenaean studies. Consiglio Nazionale delle Ricerche. 161–187.

• Morpurgo Davies, A. (1981). Mycenaean and Greek Prepositions: o-pi, e-pi, etc. In Heubeck A, & Neumann, G. (Eds.) Res Mycenaeae. Akten des VII. Internationalen Mykenologischen Colloquiums in Nürnberg vom 6.–10. April 1981. Vandenhoeck & Ruprecht. 287–310.

• Ram-Prasad, K. (2018). The Syntax of ‘Local Particles’ in Sanskrit: A Contribution to the Study of Proto-Indo-European Syntax. MA dissertation. University of Cambridge.

• Ram-Prasad, K. (2023). Clitics and the Left Periphery in the Sanskrit of the Rigveda. Journal of Historical Syntax, 7(22), 1–53.

• Renou, L. (1933), La séparation du préverbe et du verbe en védique. Bulletin de la Société de Linguistique de Paris, 34, 49–96.

• Thurneysen, R. (1975). A Grammar of Old Irish. Binchy, D. A., & Bergin, O. (Trans.). Dublin Institute for Advanced Studies.

• Watkins, C. (1963). Preliminaries to a historical and comparative analysis of the syntax of the Old Irish verb. Celtica 6, 1–49.

• Watkins, C. (1964). Preliminaries to the Reconstruction of Indo-European Sentence Structure. In Lunt, H. G. (Ed.) Proceedings of the Ninth International Congress of Linguists, Cambridge, Mass., August 27–31, 1962. De Gruyter. 1035–1045.

• Wright, J. (1910). Grammar of the Gothic Language. Clarendon Press.

• Zanchi, C. (2019). Multiple Preverbs in Ancient Indo-European Languages: A Comparative Study on Vedic, Homeric Greek, Old Church Slavic, and Old Irish. Narr Francke Attempto Verlag.

Images my own or from Wikimedia. Cover image: flint blade, housed in Egypt’s Metropolitan Museum of Art, from the Naqada culture of Predynastic Egypt, 3650–3300 BC (so maybe a cutting blade contemporaneous with Proto-Indo-European).

In Modern English, the grammatical options for nouns are limited to two. A given noun can be singular, referring to one individual thing, or plural, which is used for any amount other than one. The latter of course includes all integers greater than one up to infinity (two houses, ten billion cars), but interestingly the plural is also used with zero (zero trains, no trees) and even with one and a fraction (one and a half cakes).

This capacity of words to change according to amount is known as grammatical number. It operates both as an abstract feature that our brains syntatically compute, and as an observable (i.e. audible or visible) component of words that expresses that feature. In English today, a final S and its varied sounds are the default expression of plurality (cats, dogs), but they continue to share that role with the ending EN (oxen, children), a changed vowel (teeth, mice), or no change to form at all (sheep, deer).

In European languages, singular and plural are usually the only two options available for grammatical number. However, across human languages more generally, other grammatical numbers are possible – even in the history of English itself.

One such possibility is the dual, which is the topic of the article you’re currently reading. This is a short adventure into lost grammar, unearthing the bones of a number long since devoured by its plural sister, as well as the many connections to documented languages with a living dual, both modern and historical.

As the name implies, the dual is a feature of words that conveys two of that particular thing. If you learn Modern Standard Arabic today, for example, you’ll find that a noun can be singular, dual or plural.

• كتاب (kitāb) ~ ‘book’

• كتابين (kitābayn) ~ ‘two books’ (nominative: -ān-)

• كتب (kutub) ~ ‘books’

The dual in Arabic is not as strong as it once was, though; in the classical language, the dual was obligatory when referring to any two things. Pronouns and associated adjectives and verbs had dedicated dual forms too.

This grammatical story of dual decline in Arabic is a pertinent parallel for English, since the history of the latter language shows that English once had it too – but in English, it’s disappeared into oblivion. Unlike Arabic, we also have no documented use of a dual number that’s hale and hearty. Even when English first enters the historical record, the dual is limited and rare.

Let’s begin our journey into grammatical history with those scraps of duality.

Subscribe now

Detecting the dual in English

In our sources for Old English, nouns are straightforwardly singular or plural. Relics of their lost sister number appear though in the form of personal pronouns – that is, in how Old English speakers could say ‘we two’ and ‘you two’. These pronouns sounded similar to their plural counterparts; similar to modern we and you, the old duals were wit and ġit (pronounced like ‘yit’).

In the poem Beowulf, many instances of wit and ġit appear. See here how the hero Beowulf uses wit to refer to himself competing with his childhood friend Breca:

hæfdon swurd nacod, þā wit on sund rēon
heard on handa, wit unc wið hronfixas
werian þōhton…

‘We had bare swords as we rowed on the sea, hard in our hands, as we two thought to defend ourselves against whales’
(Beowulf 539-41)

These two pronouns come with a paradigm of different case forms. If ‘we two’ or ‘you two’ were the object of being seen, and so needed the accusative case, unc and inc would be used, instead of nominative wit and ġit.

… nelle iċ beorges weard
oferflēon fōtes trem, ac unc sċeal
weorðan æt wealle, swā unc wyrd ġetēoð

‘I will not flee a footstep from the barrow’s guard, but for us two it shall happen at the wall, as fate for us decrees’
(Beowulf 2524-6)

The dual pronouns do survive past the Norman Conquest and into the Middle English period; our sources bear witness to pronouns like wit, unk, ȝit and ink as late as the 1300s, but they fade out to nothing after that century.

Possible traces of a dual number might also be present in the unusual morphology of two inherently two-related words in Old English: twēġen ‘two’ and bēġen ‘both’. Bear in mind: those dotted Gs are to be read like a Y-sound.

These words show wildly different forms according to gender, with feminine/neuter twā (the origin of modern two) and masculine twēġen (the origin of archaic twain). Bēġen likewise alternates with feminine bā and neuter bā/bū. From the fusion of bā þā ‘both the’, perhaps with some inspiration from Norse, we get the modern word both.

Compounding and analogy are responsible for elements of the unique paradigm of twēġen and bēġen, but some aspects might stem from older endings dedicated to the dual. This means that there might be a whisper of a whisper of the dual to be heard when we say two, both or never the twain shall meet.

All things considered, these phantoms and fossils don’t dazzle us with the vitality of the English dual number. To appreciate its full history, we need to head up the family tree.

The Germanic situation

What we find in Old English is pretty much the situation across the Germanic family to which English belongs.

In Gothic, the oldest well-attested Germanic language, we find dedicated duals in the form of personal pronouns like 𐍅𐌹𐍄 (wit) ‘we two’. Here in the Gothic Bible, Jesus uses that very dual pronoun while talking to His heavenly father:

… 𐌴𐌹 𐍃𐌹𐌾𐌰𐌹𐌽𐌰 𐌰𐌹𐌽 𐍃𐍅𐌰𐍃𐍅𐌴 𐍅𐌹𐍄
ei sijaina ain swaswe wit

‘So that they may be one, as we two are’
(Codex Argenteus, John 17.11)

There are some dual noun fossils like 𐌳𐌰𐌿𐍂𐍉𐌽𐍃 (daurōns) ‘(double) gate’, and Gothic verbs can bear dual endings for the first and second persons (i.e. ‘we two do’ and ‘you two do’, but not ‘they two do’). In the following quotations, Jesus heals two blind men, and so the Gothic verbs addressed to them are grammatically dual:

𐌲𐌰-𐌿-𐌻𐌰𐌿𐌱𐌾𐌰𐍄𐍃 𐌸𐌰𐍄𐌴𐌹 𐌼𐌰𐌲𐌾𐌰𐌿 𐌸𐌰𐍄𐌰 𐍄𐌰𐌿𐌾𐌰𐌽 … 𐍃𐌰𐌹𐍈𐌰𐍄𐍃 𐌴𐌹 𐌼𐌰𐌽𐌽𐌰 𐌽𐌹 𐍅𐌹𐍄𐌹
ga-u-laubjats þatei magjau þata taujan? … saiƕats ei manna ni witi

‘Do you believe that I am able to do this? … See that no man knows it’
(Codex Argenteus, Matthew 9:30)

Later, in Old Norse, we only find the recognisble pronouns vit ‘we two’ and it ‘you two’. The latter later becomes þit, just as its second-person plural equivalent ér becomes þér. This extra consonant they apparently gained through consistently following verbs that ended with Þ.

Put together, the evidence from Gothic, Old English and Old Norse tells a story of grammatical decline. Just as with Arabic, the dual is on the retreat across the landscape of medieval Germanic grammar.

Earlier than Gothic, at the outset of this Germanic story, we can reconstruct a fairly strong dual number for the grammar of Proto-Germanic. Then, the situation in Gothic – the closest in time to the Germanic languages’ common origin – displays a vitality in its pronouns, verbs and nouns that is not to last. Later on, in Old English and Old Norse, we see the dual taking refuge in pronouns, an unsurprising bastion of linguistic weirdness.

There exceptions to this story, mind you. Having arrived on the island with sailors like Naddodd and Garðar Svavarsson, Old Norse was spoken and written in Iceland, where the dual would experience an unusual renaissance. While in English, the plural pronouns (we, you) ousted the dual, the opposite has happened in Icelandic. This is to say, the originally dual pronouns við and þið in Icelandic gradually usurped their plural counterparts vér and þér, becoming used for all plurality themselves.

Something similar happened on the Continent in Bavarian (spoken in Bavaria, Austria and northern Italy). In Bavarian, the cognate of Old English ġit is eß, with the accusative/dative form enk. In some dialects, eß has expanded beyond two-ness to become the general plural form, where Standard German would use ihr.

However, as in English in particular, all this shows is that the Germanic dual is a shadow of its prehistoric glory. To get a sense of that status, we should turn to Germanic’s ancient relatives.

To antiquity and beyond

If we broaden our scope to English’s distant cousins many-times-removed (such as Ancient Greek, Avestan, and Vedic and later Classical Sanskrit), the dual is a fully functional grammatical number in some historical Indo-European languages.

We find it not only with pronouns (e.g. Ancient Greek νώ (nṓ) 'we two'), but also, in some branches of the family tree, with nouns, adjectives and verbs that are fully dual-able in form and meaning. See here in Vedic:

च॒क्र॒वा॒केव॒ प्रति॒ वस्तो॑रुस्रा॒र्वाञ्चा॑ यातं र॒थ्ये॑व शक्रा
cakravākā́ iva práti vástor usrā / arvā́ñcā yātam rathyā̀ iva śakrā

‘Like a pair of Cakravāka-birds, come here at daybreak, like two charriot drivers, you two mighty ones’
(Rigveda 2.39.3)

In Ancient Greek, the dual number and its endings are unsurprisingly common with things that come in natural pairs. This is noticeable with certain body parts, like hands or eyes.

τὸν δὲ σκότος ὄσσε κάλυψε
tòn dè skótos ósse kálupse

‘And darkness covered his eyes’
(Homer, Iliad 4.461)

Ancient Greek and Vedic can also do some cool things with the dual. Some named individuals in the sources (such as gods or heroes) appear with a dual ending that's intended to refer to that individual and another, implied person.

For example, the Vedic deity Mitra in some lines of the Rigveda appears as grammatically dual “मित्रा” (Mitrā́). This doesn’t mean that there are two of him; it’s an abbrevation for Mitra and his common companion Varuṇa. Likewise, the Homeric hero Ajax is occasionally “Αἴαντε” (Aíante). This is argued to be another elliptical dual and not to refer to two Ajaxes (Ajantes?), but rather the belligerent half-brothers Ajax and Teucer.

However, even in these ancient languages, the dual feels like a sickly part of grammar. It did not survive past the ancient period of Sanskrit into the spoken vernaculars of Middle Indo-Aryan. Even in Homer’s Greek, it can swapped for the plural. It limped on into later Classical Greek, maintained in natural pairs and stock phrases, but no further and into the Hellenistic period.

Over the Adriatic Sea in Latin, it’s almost nowhere to be found; it survives in the unusual declension of the naturally dualistic words duo ‘two’ and ambō ‘both’. It is also surprisingly absent from the oldest documented branch of the family, Anatolian, with only traces in Hittite.

There are again only whispers of duality in early Celtic languages like Gaulish, Old Irish and Old Welsh, in particular in the form of the number ‘two’. Old Irish has certain duals like láim ‘two hands’ and súil ‘two eyes’, which must be accompanied by the number two. These are body parts and natural pairs, but Old Irish also exhibits dual forms like mnaí ‘two women’, which is something not determined by nature.

This hints at a certain vitality to the dual in old Celtic languages, and it’s worth noting that relics have endured into the modern-day languages. To say ‘my two hands’ in Scottish Gaelic is mo dhà làimh, which looks like a singular, not the plural we might expect.

All in all, the picture that this comparative evidence paints is that the dual must have been a part of the grammar of Proto-Indo-European itself. On the basis of Latin, Ancient Greek, Sanskrit, Proto-Germanic and others (but not really from Hittite, strangely), we can reconstruct it alongside the singular and the plural.

We’re even able to reconstruct some of the original endings for dual nouns. ‘Two sheep’ for instance would have been *h₂ówih₁e in PIE, and ‘two fields’ would have been *h₂éǵroh₁. (The “h₁” here is a kind of algebraic placeholder, for a consonant of uncertain quality.) The ancestral language must have used it when talking about pairs of things; it may have even been obligatory to do so.

However, what we also observe is the parallel loss of the dual across the various branches of the Indo-European family tree. Over and over, we find that the plural comes to be used instead for indicating two things, relegating the dual to pronouns and idioms.

Given this pattern of change, the roots of the rot may lie in Proto-Indo-European itself, or in an inherent redunancy of the grammatical number. If the plural can be used to count between three and infinity, it’s no great stretch to extend that use to one more number, namely two. In other words, the plural can easily absorb the dual, but not the dual the plural.

For sure, the general story of the dual number in Indo-European is one of decline – but not yet of definitive defeat!

Dual death?

‘But!’ I hear the keener linguists among you cry.

It’s true; the dual ain’t dead in Indo-European. There are languages in the family that have maintained the dual to a greater or lesser extent, either in obvious fossilised morphology or as a full-blown feature of grammar.

Lithuanian and the Slavic languages have offered a refuge for the embattled number. In some corners of Slavic, such as Czech, the dual survives in the unexpected endings of natural pairs. One eye and one hand are oko and ruka, while two or more eyes and hands are not the expected oka and ruky, but the old duals oči and ruce. Likewise, the Czech word for ‘hundred’ (sto) has a form used especially in ‘two hundred’ (dvě stě).

What’s more, Slovenian (spoken in Slovenia) and Sorbian (spoken in Germany) continue to use the dual as a productive feature of the grammar. This extends beyond natural pairs to any two things, as in the Slovenian for ‘house’:

• hiša ‘house’

• hiši ‘two houses’

• hiše ‘houses’

However, it’s reported that even here the dual is retreating to natural and typical pairs, and is:

“… also becoming unstable in dialectal and colloquial use. It is likely to remain operational in the standard languages by virtue of being enshrined in the authoritative prescriptive grammars of these languages … the dual has been saved in Slovenian through the ‘intervention of grammarians’.”

(Sussex & Cubberley 2006: 225-6)

Even in these dens of duality, there seems to be an element of the artificial to the number’s continued usage.

So, this has been a quick tour through a ghost of English grammar, a forgotten sister of the dominant singular and plural. If anything, I hope it serves as an interesting example of family-wide language change, and of how we can’t take the features of our modern grammar as given. Other grammatical possibilities exist, and may even haunt our very own languages!

END.

Subscribe now

Bonus fun facts

• The dual is a historical feature of the whole Semitic language family, not just Arabic. On that note, מצרים (mitzrayim) is the Hebrew name for 'Egypt', and its -ayim ending curiously appears to be that of the dual grammatical number, for two things. I remain delighted by the explanation that it is originally dual, referring to the two ancient regions of Upper and Lower Egypt.

• Slovenian hiša is related to English house, a product of an early Germanic borrowing into early Slavic. I just think that’s neat.

Resources

• Gothic Bible

• Homer’s Iliad

• Rigveda

References

• Blanc, H. (1970). Dual and Pseudo-Dual in the Arabic Dialects. Language, 46(1), 42–57.

• https://bosworthtoller.com/

• Clackson, J. (2007). Indo-European Linguistics: An Introduction. Cambridge University Press.

• Hoffner, H. A., & Melchert, H. C. (2008). A Grammar of the Hittite Language. Part 1, Reference grammar. Eisenbrauns.

• Hogg, R. M., & Fulk, R. D. (2011). Grammar of Old English: Volume II: Morphology. Wiley-Blackwell.

• Majewski, K. (2022). The Ruthwell Cross and Its Texts: A New Reconstruction and an Edition of The Ruthwell Crucifixion Poem (Vol. 132). Walter de Gruyter GmbH & Co KG.

• oed.com

• Pronk-Tiethoff, S. (2013). The Germanic loanwords in Proto-Slavic. Brill.

• Ringe, D. A. (2006). From Proto-Indo-European to Proto-Germanic. A Linguistic History of English (Vol. 1). Oxford University Press.

• Sihler, A. L. (1995). New Comparative Grammar of Greek and Latin (1st ed.). Oxford University Press.

• Sussex, R., & Cubberley, P. (2006). The Slavic Languages. Cambridge University Press.

• Thurneysen, R. (1975). A Grammar of Old Irish. Translated by D. A. Binchy & Osborn Bergin. Dublin Institute for Advanced Studies.

• Wackernagel, J. (1877). Zum homerischen dual. Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der Indogermanischen Sprachen 23. 302–310.

• Wiesinger, P. (2013). The central and southern Bavarian dialects in Bavaria and Austria. The Dialects of Modern German. 438–519. Routledge.

Images from Wikimedia, except for the memes.

Article originally posted on dannybate.com in March 2024. Updated and reshared now on Substack, because I can.

Across most of English today, these are two separate vowels dependably represented by two separate letters: E and I. Yet, at the beginning of English, the former muscles in on the phonetic territory of the latter.

This is quite the oddity. To my linguistically obsessed brain, no comparable words with this E-I mismatch* readily come to mind. Without such allies, it can’t be formulated and tolerated as a predictable rule. Upward and Davidson, in their mighty History of English Spelling, refer to the word as “a sound-symbol anomaly” (2011: 40).

Nonetheless, we can try to explain and understand the lonely quirk. In this effort, we encounter grand forces at work behind language, namely the complex dances between speech and spelling, logic and emotion.

Subscribe now

As a rule of thumb, if a written English word looks to be pronounced a certain way, it most probably is or was thus spoken. This rule certainly covers English.

For one thing, it’s perfectly possible to pronounce the first syllable of English as it’s written. Weird as it will sound, there’s nothing impossible or phonetically awkward about such an ‘eng’ sequence of sounds; German speakers manage it in their equivalent word Englisch, just as English speakers manage it in penguin and ginseng tea.

Besides, the E of English is consistently present throughout the written record of the language, right back to the Old English period (c. 450–1100 AD). In texts from those days of yore, Englisc is the usual spelling for the language of the Engle living in Engla land. True, there is variation (including “Ænglisc” and even the rare “Onglisc”), but the E-forms are the norm.

“ic Ælfrīc wolde þās lytlan bōc āwendan tō engliscum gereorde of ðām stæfcræfte þe is gehāten grammatica”

‘I Ælfric wanted to translate into the English language this little book about the writing skill that is called grammatica’

(Ælfric’s Grammar, c. 990 AD)

With such principles of articulation and consistency of attestation, we have no reason to think that the initial vowel of Engle and Englisc was pronounced as anything other than a normal short E. That is, it was a sound made with the tongue forwards and at a medium height in the available mouth space for vowels:

This sound was the product of a shift in the prehistoric development of English. Known as i-umlaut, this change moved vowels in words forwards in the mouth to become ‘front’ vowels, on the condition that the following syllable in the word also contained a particular front vowel or semivowel. Those subsequent syllables have since largely dropped off, but the umlaut that they triggered remains very much present in English.

On the vowel trapezoid above, we can visualise i-umlaut as a leftward shift, as back vowels on the right became their matching front vowel on the left (while some already front-ish vowels became higher and front-er too). It shunted vowels not only in Pre-Old English, but across the emerging Germanic languages, with the curious exception of Gothic.

The change can explain so much of English’s grammar and vocabulary today, being responsible for the difference in vowels between man and men, tooth and teeth, foot and feet, full and fill, gold and gild, long and length, food and feed, bank and bench, Wales and Welsh, Frank and French, and many words more. Note the frequent E in the second word of these pairs, which is the umlauted one. The change has also been at work in the word English.

The English-to-be first appear in history in Roman writings. The historian and politician Tacitus, writing in the first century AD, provides us with a rare mention in his description of the barbarous foreigners to the empire’s north:

“Reudignī deinde et Aviōnes et Angliī et Varīnī et Eudosēs et Suardōnēs et Nuithōnēs flūminibus aut silvīs mūniuntur.”

‘Then the Reudigni, the Aviones, the Anglii, the Varini, the Eudoses, the Suardones, and Nuithones, protected by rivers or forests.’

(Tacitus, Germania 40)

The “Angliī” listed here are the English, at least etymologically. Their Latinised name is spelled with an A because such a low back vowel was what the Romans heard at the time. It was most likely a tribal name with a geographical basis, assumed by the inhabitants of a peninsula in what is today the north tip of Germany.

From this self-designation, the Romans coined their noun Anglus. This has since given English vocabulary like Angle, Anglo- and anglicise.

Back in the local Germanic speech of the Anglii, i-umlaut would in time take effect. It shifted the original A-type vowel forwards and upwards, resulting in Old English words like Engle and Englisc. A parallel story belongs to the great river known to the Romans as the Albis, but later to the Germans as the Elbe.

The evidence (present, historical and pre-historical) indicates that the E of English was once a ‘normal’ E in earlier stages of English, i.e. spelling the umlaut-fronted, mid-high vowel that we’d expect it to. This harmony between speech and spelling, however, became discordant in the post-1066 period of Middle English.

We can detect a distinct sound change, small in scope, that affected a handful of spoken words in the Middle English period. It was all a matter of the right sounds in the right order.

When a vowel of middle tongue-height occurred before the consonants /ŋg/ or /ŋk/ (hear the NG in finger and the NK in pink), this environment had the power to raise up the vowel up. In other words, that specific combination of sounds triggered a vocalic shift among medieval speakers, and turned mid-high vowels into high ones. Speculating, I suppose that this was a kind of articulatory assimilation. In the vowel space, the simple shift looks so:

Thankfully for us linguists, this shift is reflected in ye olde spelling. In Middle English texts from before 1500, we find words like weng, streng and lenger turning into their modern forms wing, string and linger. Possibly also, fleng became fling.

Likewise, from the thirteenth century onwards, we observe the substance enk shifting to ink. This change in sound and letter obscures the loanword’s etymology, being a sister of French encre and derived from Ancient Greek ἔγκαυστος (énkaustos), literally ‘in-burnt’.

Even the common verb think could belong to this exclusive set of words. It does derive from Old English þencan, although modern think might instead be the offspring of a confusion between the similar verbs þencan ‘to think’ and þyncan ‘to seem’. The latter, by the way, is behind the odd and affected formulation methinks. It was originally no stranger than saying ‘it seems to me’ today.

It wasn’t only the front E-vowel that was a candidate for elevation; the counterpart back O-vowel could find itself raised up into the territory of U. This could be behind the pronunciation of among, which rhymes with sung, not song.

With an appreciation of this change, we can partly explain English. Since its first syllable included the same environment of sounds, it too underwent the shift heard with wing, string, linger and ink. Unlike them, of course, there was a complication; the spoken word left its spelling behind.

The truth of the matter is therefore that the English sounds in English have developed regularly; it’s spelling that is instead the source of bother, being in need of the E-to-I update that other words unproblematically received.

Why then isn’t English spelled Inglish, to match wing and ink? Well, it has been, and occasionally still is.

Across the eras of Middle English and Early Modern English, our sources for the language attest a great number of shifted spellings: “Inglysshe”, “Inglis”, “Inglish”, “Inglishe”, “Inglisce”, and also same-sounding Y-initial variants like “Ynglis” and “Ynglysche”. These are all in tune with the change in speech.

This alternative is still current in some places. Over in Ireland, at a distance from the writing conventions of Britain, the Ulster English dialect can be referred to in writing as Ulstèr Inglis. Elsewhere, varieties of English that are heavily influenced by Chinese languages are not Chenglish, but rather Chinglish. In this case, the letter I ensures the right pronunciation. Of course, Inglish may also occur as a spelling ‘mistake’ by writers with less exposure to the standard.

Such variants, past and present, are not to be dismissed. All deserve consideration by scholars, because spelling is a matter of consensus and community. Although one way of spelling a word may rise to such a status that alternatives seem weird or inferior by comparison, that standard fundamentally remains one possible variant among others, its God-given standing just an illusion of inevitability. So often, it’s not a matter of how a word is written that wins it status, but rather who writes it that way – and whether they’re aligned with the powers of an elite.

Contemplating the ban on Inglish pokes a hole in the perceived certainty and soundness of the standard. Since the early modern period, the writers-that-be have overwhelmingly rallied behind the conservative spelling English, despite its phonetic flaw. And why? My answer has to be: if not passive imitation, then active sentimentality.

As strange and non-linguistic as that answer may feel, methinks it’s the most plausible explanation. True, there is a great deal of unfeeling inertia to be seen in the history of spelling. Yet conservatism alone cannot say why English couldn’t shift in its written form, when weng, streng, lenger and enk clearly could.

Instead, its status as the title of the language – that truest tongue of England, penned by Shakespeare and Chaucer, wielded since the days of Offa and Alfred the Great – might have inspired an idiosyncratic reluctance to recognise in spelling any changes in speech. By comparison, innovative Inglish would feel odd and alien, as coldly logical as the Newspeak promoted under Ingsoc (English Socialism) in George Orwell’s 1984.

Singular affection for English may therefore be to blame for this singular “sound-symbol anomaly”. If so, it’s a sobering reminder for the historical linguist who so often deals in sweeping, language-wide changes and the unconscious workings of linguistic logic – you can’t take the human out of the picture, in particular our capacity to get emotionally attached to individual words.

English speakers and their identities are reflected in the E of the word English itself. While not rational or practical, this nonetheless, as said, seems all too appropriate.

END.

Subscribe now

Footnote

*No other examples came to mind at time of writing, but Yoïn van Spijk has since noted the same mismatch in pretty. All due credit to him for the good catch. Without the same nasal environment, this won’t be due to the sound change discussed here, nor do my spelling speculations for English apply. Methinks pretty needs its own article.

Primary sources

• Ælfric’s Grammar

• Tacitus’ Germania

Secondary sources and references

• bosworthtoller.com

• Dobson, E. J. (1968). English pronunciation 1500–1700, Vol. II, 2nd edition. Clarendon Press.

• Minkova, D. (2013). Historical phonology of English. Edinburgh University Press.

• oed.com

• Upward, C., & Davidson, G. (2011). The History of English Spelling. John Wiley & Sons.

• Wright, J., & Wright, E. M. (1923). An Elementary Middle English Grammar. Oxford University Press.

Images either my own masterpieces or taken from Wikimedia/Wikipedia and other public-domain sources.

This piece is all about one particular consonant sound, to be heard often in spoken English today, but not in older stages of the language. Because of this limited chronology, this consonant may well be the newest addition to the band – although I just can’t bring myself to formulate it any more strongly than “may well be”.

As I’ll first try to demonstrate, what it means to be a particular sound of a particular language is not straightforward. What it definitely is, though, is a linguistically rich and interesting topic. So, stick with me.

Subscribe now

The frustrating elusiveness of sounds

If I sneeze in the middle of uttering an English sentence, is a sneeze one of the sounds of English speech?

This is an intentional absurdity, likely to elicit a prompt response of no, but it has its uses. It tests our conception or intuition for how some sounds, but not others, belong to the phonetic ingredients of certain spoken languages. Some of the sounds that we humans make while speaking, like sneezes and coughs, are easily excluded from that set. For one reason, they contain no obvious linguistic content, interrupting rather than contributing to what we want to get across.

There are also those sounds that are well used in the world’s tongues, just not in English. The Ll-sound of Welsh, the pharyngeal consonant behind Arabic ﻉ, and Czech’s infamous Ř are common in their respective languages, but I’d bet that neither the laylinguist nor the expert could reasonably include them when drawing a border around the sounds of English.

Then there are those sounds whose participation and status in speech is much less clear cut. English speakers, for example, produce a bewildering plethora of consonants with every sentence. Some of them seem alien in isolation, but in fact do occur in English, and with such consistency that they help to create ‘authentic’-sounding English speech.

This great video by the YouTube platform Pronunciation Studio discusses five of them, and I’ll respectfully pinch his example of the labiodental sound [ɱ]. As its IPA symbol suggests, this is a nasal sound, much like a lippy M, but with the bottom lip lightly touched by the top teeth to halt the departing airflow temporarily.

Uttering a repeated sequence of [ɱ] on its own sounds bizarre and unfamiliar, but it’s what English speakers tend to say in casual speech between some words (e.g. on form) and within others (e.g. symphony, infant, invisible). The option always remains, though, to pronounce the Ms and Ns in such words as [m] and [n], especially when we say the words more slowly and loudly for the purpose of emphasis or education – that is, in ‘hyperspeech’.

For another example, in my English, I might say what or cat seven out of ten times with a glottal stop [ʔ] for the final T-sound. However, if I’m asking what? with a particular degree of venom, or if I’m enthusiastically indicating a cat, I’ll most likely pronounce the T as [t].

These cases demonstrate that among all the many sounds that speech can include, there is a subset that we think of (consciously or not) as the proper ingredients of English speech. If I didn’t, I’d have no problem with pronouncing what as [wɒʔ] in all contexts, stressed or unstressed. For other English speakers who are fortunately not me, that may be fine for them.

That most exclusive subset of sounds contains the phonemes of a spoken language. These are its key sounds. My go-to metaphor is that phonemes are the ‘building blocks’ of speech. I don’t want to get into how some linguists have greatly objected to the concept; it suffices to mention that some linguists have greatly objected to the concept. With that debate acknowledged, the fact remains that phonemes are a very useful and successful theory.

We can identify sounds like [t] and [m] that hold the status of independent phonemes of English. They can appear anywhere in any word and in any conversational context. When we write about a language’s phonemes, the sounds in question don slashes: /t/ and /m/. Their status accords with our intuition, although the exact reality of phonemes is again the subject of debate: are they something in the mouth or in the brain?

Meanwhile, [ʔ] and [ɱ] are possible variants or ‘allophones’ of /t/ and /m/ – common sounds for sure, but dependent on the context of where and when they’re used.¹ Allophones often get replaced in instances of hyperspeech and emphasis. Phonemes, not all their various allophones, are also what we’d expect would be given their own dedicated letters in an alphabet, although this is a flimsy test for phonemes, with some good historical counterexamples.

The classic test for phonemes whether or not they make a meaningful difference to a word. If I were to say [mæp] and [næp], the two sounds would lead other English speakers to identify the two words map and nap. These form a minimal pair, two words that are the same in all respects except for the sound under scrutiny. Map and nap are different English words, so /m/ and /n/ gain entry to the phoneme set.

If I instead said [mæp] and [ɱæp], the second would be received as a weird mispronounced version of the first word – at least by English speakers. The two sounds might be phonemes in another language, in which [mæp] and [ɱæp] begin with two separate consonants and function as two separate words.

With all this in mind, a typical inventory of the consonants of Modern English will include around twenty-four phonemes. These can be said to ‘belong’ to the sounds of English, with all due caveats and provisos applied.

Just as phonemes can differ across the geography of the world’s languages, so too can they differ across time. A sound can achieve phonemic status over time, especially if it occurs often enough and helps speakers to distinguish one from another. Where later speakers would recognise it and say ‘Sure, you’re making an X sound’, their linguistic forebears would say ‘That sounds like a kind of weird Y sound’.

That the phonemes that define a spoken language can change is an essential fact for my purposes. I want to discuss the consonant that I think most recently joined the club of English VIPs (Very Important Phonemes).

Greet the new kid

Of course, in addition to the difficulty in defining ‘a sound of English’, there’s the further difficulty in defining English. It’s such a broad family of accents and dialects nowadays that one sound could be on its way to gaining (or losing) phoneme-hood for some speakers and not for others.

Vowel sounds in particular fluctuate like the wind, so I won’t bother trying to identify English’s newest among those. But I should have more luck among the air-restricting group of speech-sounds that we call consonants. Indeed, I have one in mind that seems to be both stable and common to most accents under the English umbrella (one exception being Philippine English).

But enough with the academic preamble and caveated caution; I’ve teased you long enough. So, what is the newest consonant of English?

Well, here it is: /ʒ/.

This is the smooth-talking fricative sound that zhooshes up English words like pleasure, visual, confusion, seizure, luxury, television, Asia, sabotage, regime and beige. The variety of methods for spelling this consonant (here: SU, ZU, XU, SI and GE) is indicative of its marginal status in the language and in speakers’ minds, and of its youth. It developed in English long after any window of opportunity for being granted its own letter.

Consequently, it’s one of the many changes in speech that standard English spelling remains largely ignorant of, a theme explored in detail in Why Q Needs U. We English readers instead know to infer /ʒ/ from written contexts like the SU in usual. But occasionally it needs to be taken out of that context, so English writers have had to improvise. ZH has stepped up to spell the sound, such as when the usual is abbreviated to ‘the uzhe’.

The consonant has no such identity issues in other languages, mind you. French has the sound and spells it with J or soft G (e.g. je, joli, manger, plonger). Ukrainian speech employs it too and renders it in writing as Ж, while Czech reserves Ž for the same consonant.

English /ʒ/ is also a variable sound, sometimes shifting from speaking to speaker, word to word. For example, it’s inconstantly present in how I pronounce nausea. I reckon I lean towards ‘nor-ʒuh’, but trisyllabic ‘nor-zi-yuh’ is an option for me too. The latter is perhaps because of how I pronounce the derived adjective nauseous (‘nor-zi-us’). I can also hear variation with amnesia: ‘am-nee-ʒuh’ or ‘am-nee-zi-yuh’.

You might also hear speakers pronounce seizure as ‘seez-yuh’ and visual as ‘viz-yul’, with the internal sequence of two sounds [zj] instead of just [ʒ]. Such speakers have very good reason to do so; theirs is a conservative pronunciation. The [zj] sequence in older English that they maintain has been one of the sources of the sound [ʒ].

Once, all words like measure and leisure, adopted into English from French or Norman, happened to be pronounced with the sequence [zj]. It was spelled with an S, since using Z for the consonant [z] wasn’t an established thing yet. This constancy in spelling makes it hard to pinpoint when the shift began, but we can at least say that it’s been happening within post-medieval English.

Speakers gradually merged the two sounds into one, for reasons of phonetic efficiency. The old [j] sound in the sequence (spelled with I or implicit within the ‘French U’ of measure) pulled the preceding [z] sound back in the mouth, towards the hard palate, and merged with it. The position of the stressed syllable in a particular word likely played a role in whether and when the [z] shifted.

These instances of [z] acquired a new palatoalveolar position in the mouth by means of the process of ‘assimilatory palatalisation’. A similar thing has gone with the [sj] sequence once pronounced in station and pressure.

The other source of [ʒ], of course, is loanwords adopted from other languages with no such questions over their phonemes. Modern French is the donor par excellence, responsible for the ‘super-soft G’ in lingerie, massage and rouge. Acquisitions from other languages are much rarer, but modern famous names, like that of the USSR’s Marshal Zhukov, may have made a small contribution.

The question does deserve to be asked and answered, is /ʒ/ really an English phoneme? Some linguists would bar it from the VIP club, on account of its limitations. They might argue instead that [ʒ] is just an allophone of /ʃ/ (the shushing sound in shoe and fish) or of /d͡ʒ/ (the jarring consonant in judge).

As mentioned, to be admitted, a phoneme needs to be able to distinguish two words of a language apart. The number of minimal pairs for /ʒ/, /ʃ/ and /d͡ʒ/ is, well, minimal. But we do have measure and mesher (‘a person who meshes’), and also pleasure and pledger (‘a person who pledges’) for the second contrast. A near-minimal pair is offered by vision and fission. By the skin of its teeth, /ʒ/ passes this test.²

Can /ʒ/ appear anywhere in a word, thereby being independent from a particular sound-context? Genre is its best hope here, providing rare evidence of /ʒ/ at the start of a word. It’s always pronounced with an initial /ʒ/ for me, but it’s true that many people say genre with /d͡ʒ/ instead, like John.

This is a case of ‘nativising’ the sound, swapping it for one that’s much more common and comfortable. See also: the second G of garage, which is a super-soft /ʒ/ for some, but a less soft /d͡ʒ/ for others, including me. I rhyme garage with marriage.

This has a long tradition. In the late medieval and early modern eras, English consistently adopted French words with /ʒ/ but said them like /d͡ʒ/ instead. In time, though, the incoming words and the new pronunciation of [zj] increased the phonological pressure.

At last, the dam burst, and /ʒ/ has established itself a new phoneme. As for when this happened, I’d say it occurred recently in historical-linguistics terms; I’d narrow it down to ‘over the course of the modern era’.

In a strange way, though, English speech was expecting this development. It had room ready for the newbie. The thing is, the sounds of speech work through contrasting features; English has historically relied on the quality of voicing (vibrating in your larynx) to distinguish sounds from one another.

For example, the /s/ in words like sue has no voicing, unlike the voiced /z/ in zoo. Voicing likewise separates fats from vats (/f/ vs. /v/). The difference in plosive sounds between pat and bat (/b/ vs. /p/) and tin and din (/t/ vs. /d/) is also traditionally framed as a voicing contrast, but in the English speech of today, it’s become something quite complex.

All these consonants of English had partnered up (one voiceless, the other voiced) during the Middle Ages – except for /ʃ/. What was the voiced companion for the /ʃ/ in shoe and fish? It was a lonely sound, and the English system of consonants had a vacancy.

To become a phoneme, /ʒ/ sure had to work for it, but English was ready to accept it for many centuries. It now functions as the voiced counterpart to /ʃ/, which is why I find its ZH spelling very appealing; ZH contrasts with SH, together mirroring Z and S. It’s why [ʒ] may ‘feel’ like a natural and easy sound to produce, despite its position on the periphery of speech and spelling.

I’ve seen it suggested that /ʒ/ should be excluded from any reckoning of the sounds of English, because it’s a ‘foreign’ sound. This is to say, because it appears exclusively in loanwords (mainly from French or Latin), it’s not a natural phenomenon within English. I disagree with this; the borders between languages are not fixed, and the bulk of the instances of /ʒ/ are due to English speakers changing the [zj] in visual and measure into a single sound. That development can’t be blamed on the French.

That said, an exotic air still lingers around /ʒ/. Touching on the domain of phonaesthetics, the sound seems to be coloured with hues and nuances of foreignness, if only in the minds and mouths of English speakers. There are words and place-names which many of us pronounce with /ʒ/, despite there being no etymological reason for us to do so.

Azerbaijan, Beijing and Taj Mahal are three prominent examples. I’ll hold my hands up to pronouncing them with /ʒ/ (that is, as if ‘Bey-zhing’). Yet that’s not backed up by their local languages. In all three (Azerbaijani, Mandarin Chinese, Hindi), the English J corresponds to an affricate sound, similar to or the same as the /d͡ʒ/ in jewel. In the absence of a better explanation, this looks like hyperforeignism and an exoticising of places abroad.

It seems that /ʒ/ can’t break free from its marginalised status within English speech – a phoneme now, yes, but still considered new and a bit weird. With time, its lot may improve. More uses for it could appear, ideally with chances to establish its ZH spelling better. It could take its place alongside not only voiceless SH, but also the CH, PH, TH and WH digraphs.

If you’re coining new words anytime soon, why not consider including it? The new phoneme on the block will thank you.

END.

Subscribe now

Footnotes

1. I remain uncertain of whether the definition and criteria for phonemes accidentally exclude what is arguably the most common sound in English speech, the humble schwa vowel [ə].

2. Sometimes the vocabulary of a language has just worked out in such a way that many minimal pairs can’t be found. The two TH-sounds of English in thin and then (fricative /θ/ and voiced /ð/) also struggle to prove their phoneme-hood this way. There’s thigh and thy at least, and mouth (the body part) and mouth (the action). I remember linguists at Edinburgh being grateful to the Scottish touristy shop Thistle Do Nicely for another example of the contrast, this time between thistle and this’ll.

Non-linked references

• Martínez, J., & de Vaan, M. (2014). Introduction to Avestan (Sandell, R. Trans.) Brill.

• Minkova, D. (2013). Historical phonology of English. Edinburgh University Press.

Images my own or from Wikimedia/Wikipedia.

This isn’t my contrarian streak speaking. I take genuine joy in the fact that tracing the etymology of this everyday word has posed questions and challenges as yet unresolved, and has inspired a corpus of academic papers devoted to the noun’s prehistory.

It all comes down to sounds. Put briefly, apple has a problematic consonant, a sound that makes the word difficult to derive from the earliest conceivable point in the story of English.

In the business of etymology, we follow spoken words and their constituent sounds through history and into prehistory. When we quit the safety of our historical sources and cross the threshold of that undiscovered country, etymology relies on comparing words and their sounds across related languages. From this comparative method, we can ‘reconstruct’ ancestors from which those historically documented offspring descend.

For one example, historical sources for the words of English take us only as far back as the early medieval period, when an apple was an æppel. To rewind further, we bring in English’s siblings and cousins. These are the Germanic family, with sister Scots, cousins German, Dutch, Frisian, Norwegian, Danish, Swedish, Norwegian and Icelandic, great-uncle Gothic, and many relatives more.

These together get our etymology back to the Proto-Germanic point in linguistic prehistory. At that point, *apla- was the noun for an apple or apple-like fruit – or rather, it must’ve been, but we lack the direct witnesses to it, hence the asterisk of uncertainty.

Then, if we bring in Latin, Greek, Hittite, Sanskrit, Avestan, Old Irish, Old Persian, Old Prussian and many more Olds, we can unearth the even older strata of Proto-Indo-European, spoken six-thousand-ish years ago.

Since the resuscitation of this grandparent language began in the eighteenth century, we’ve developed a sophisticated account of what sounds, words and grammar we need Proto-Indo-European to have had. One aspect is a conspicuous absence.

It’s generally agreed by scholars, based on the accumulated Indo-European evidence, that the prehistoric ancestor eschewed words with *b, a lippy sound very much like the ordinary consonant in English by, bow or bob. Introductions and summaries for the reconstructed sounds of Proto-Indo-European typically contain comments like:

“The voiced bilabial stop was unexpectedly rare”

(Ringe 2006: 8)

“The reconstructed sound *b … may have been absent from the language”

(Clackson 2007: 33)

When setting out the sounds of this ghostly language, it’s typical to bracket *b in order to indicate its uncertain involvement.

This absentee consonant is a problem for the word apple.

Subscribe now

Confronted with the claim that the common ancestor of Indo-European languages had no *b sound, it’s okay if your reaction is one of incredulity. After all, aren’t Indo-European languages (English being one) brimming with B-words? Shouldn’t that mass of examples point to its presence in their shared grandparent, as per the axioms of linguistic comparison and reconstruction?

It’s on the experts of the field, and on me as their proxy, to justify this theory. The thing is, the evidence does lead us to this conclusion. As we sift through the vocabulary of the different Indo-European branches, the pattern is that the B-consonant is present, but is so often the outcome of changes to an alternative original sound. Genuinely ancient B is much harder to find.

To begin, take English. Having placed to one side the lexical loot that English has purloined from other languages, we are left with a ‘core’ of vocabulary that English has inherited from Proto-Germanic, and before that from Proto-Indo-European. Many of that set’s members do indeed contain B, like be, brother, brew, best, bend and beaver.

Case closed? Well, no. We must match these words up with their cousins across the Indo-European family. To build a bridge back to Proto-Indo-European, we require meaning-matching relatives that include B in the same place.

Some indeed do; the word for ‘brother’ in Old Irish is bráthair, and in ancient Avestan it’s brātā. Yet many don’t match the B; the Latin kin of brother is frāter, the Sanskrit is bhrā́tā, and the Ancient Greek is phrā́tēr. These words, geographically distant from each other, are clearly related via a shared origin. But, to explain their disagreement in initial sounds, it’s best to propose that it began not with *b, but with *bʰ.

The difference between the two is slight, but not trivial. The second (*bʰ) was an aspirated consonant, a breathy kind of B. Sanskrit preserved it (bhrā́tā), but it changed in the development of Latin (with fricative frāter) and Ancient Greek (with phi-spelled phrā́tēr). In proposing *bʰ as one of the consonants of Proto-Indo-European, we gain a common denominator that connects its varied descendants.

For the Germanic branch to which English belongs, *bʰ became an unaspirated *b via the sequential shifts collectively known as Grimm’s law.

This is how it’s ended up in the aforementioned be, brother, brew, best, bend and beaver. These once began with *bʰ, prior to Germanic. The same correspondence of consonants is well exemplified with the English verb bear, related to Sanskrit bhárati, Latin ferō and Ancient Greek phérō, which all mean more or less the same thing.

Meanwhile, Grimm’s law would change any instances of Proto-Indo-European *b into Proto-Germanic *p, and more famously instances of original *p into *f. Hence, we get Latin-English cognate pairs like pater~father, piscis~fish, pedēs~feet, because Latin wasn’t subject to the same shifts.

Latin and also Ancient Greek sounds did not pass through the mangle of Grimm’s law, but regardless, their vocabularies don’t include a great deal of B-beginning words. If you turn to your trusty Latin dictionary, you’ll observe that the chapter for B is relatively short. In my Pocket Oxford Latin Dictionary, A takes up pages 1 to 17, while B finishes halfway down page 19.

The list shrinks if we subtract those words derived from another or sharing a prefix. For instance, my dictionary includes all the words that begin with the prefix bi- ‘two’ (from biceps ‘two-headed’ to bivium ‘where two roads meet’), and these take up most of page 18.

The Latin B-team for sure includes vital vocabulary like bonus ‘good’, bellum ‘war’ and bis ‘twice’. Yet we have evidence that demonstrates that their B was a later development. The Duenos Inscription from the 6th century BC attests duenos, the older form of bonus.

Likewise, bellum was once duellum (which has survived as the word duel), and bis ‘twice’ is related to duo ‘two’. Such examples of B at the start of Latin words are products of a ‘DU > B’ sound change. Meanwhile, most instances of B in the middle or at the end of words can be derived from an original aspirated *dʰ. Through this PIE consonant, Latin verbum (hence the loanwords verb and verbal) is connected to inherited English word.

Chapter Β in your Ancient Greek dictionary might be a bit longer. In it you’ll find words like basileús ‘chief, king’ (hence English basil, basilica and basilisk) and boûs ‘cow’ (hence English butter and boustrophedon). Yet this initial Β also looks to be a secondary sound.

Mycenaean writers, our oldest evidence for Greek, spelled basileús with an initial Q-type character (𐀣). Comparative evidence, such as from English cow and Sanskrit gó, also leads us to propose that the Β of boûs was once pronounced back in the mouth. For these reasons, we can confidently think that both basileús and boûs formerly began with a *gʷ consonant, as in the name Gwen, before a later shift.

The point here is that Latin and Ancient Greek display a scarcity of secure B-words that we can trace back to Proto-Indo-European, especially once we remove the products of later sound changes. The same applies to Sanskrit over in India. As a spoken language, it certainly made use of words containing the sound /b/, usually committed to writing under the Devanagari character ब.

For one, there’s the Buddha! This term, literally meaning ‘awake’, shares a root with the verb bódhati ‘to wake’.

Yet such Sanskrit vocabulary has also been shaped by sound change. The B of Buddha was also once an aspirated *bʰ too. It would’ve remained so, were it not for Grassmann’s law. This was a rule against a spoken word having two aspirates in quick succession; the first got unaspirated, turning ‘bhódhati’ into bódhati. This restriction was operational in the prehistory of both Sanskrit and Greek – the jury’s out on whether this parallel occurrence was just a coincidence.

All of this language-specific theory has one purpose: to show that the amount of evidence for a Proto-Indo-European *b sound is considerably smaller than first appearances indicate. So many of those appearances in Indo-European languages past and present are best explained as the outcomes of shifts and mergers of consonants. The set of Indo-European words which point to *b in PIE is small and complicated. Our apple is a member.

Long ago – seven paragraphs, to be precise – I mentioned Grimm’s law, which took the consonants of Proto-Indo-European and shifted them. Although there aren’t many good examples of this specific change happening, the programming of Grimm’s law would turn a *b into Germanic *p. To find evidence for PIE *b in an English word, the first condition is that the word should contain a P. The second condition is that the word should not have been taken from another language, but rather be part of English’s begrimmed Germanic inheritance.

Apple meets both of these conditions, as words like path, play and plot may do too. However, the second two don’t help us look for *b in Proto-Indo-European, because they don’t satisfy a third condition: to have cognates from other branches of the Indo-European family tree. To reach back to Proto-Indo-European, we have to build our theories as we would a chair. Attaching just one leg is far too wobbly; two legs isn’t much better; three is okay; four is ideal; and five is fine but overkill.

Play and plot lack that external corroboration. Where they came from before Germanic is a mystery. Path, meanwhile, is a word rich in history that I’ve written about previously, and might not even satisfy the second condition. Apple clears the bar again.

Of course, we have the Germanic evidence: German Apfel, Icelandic epli, etc. It even appears as “apel” in Crimean Gothic, a lost Germanic cousin fortunately recorded by a Flemish diplomat in the sixteenth century. But, in addition to this, we also find words for the fruit elsewhere in Indo-European that look and sound like apple.

Where I live, the local word is Czech jablko, just like it is jabłko in Polish and jabǎlka in Bulgarian. Strip away their Slavic accretions, and you have ‘abl’. Up by the Baltic Sea, the Lithuanian and Latvian words are obuolys and ābols. The Elbing Vocabulary, a rare surviving source for Old Prussian, names the fruit as “woble”. From the Celtic branch, we can harvest the Old Irish word ubull and Modern Welsh afal. The latter might be beyond the mythic island of Avalon, apparently a place rich in apple trees.

Moreover, down in southern Italy, there’s a speculative relative in the city of Avella, known as Abella in antiquity.

This windfall of words across Europe provides matches in sounds and meaning with English apple. There’s a key difference between these words and apple, which is that they include a B, not P. Since these words and their languages have not undergone Grimm’s law, this is what we expect to find. We have all the substance we need to propose that they all descend from a common ancestor in some early stage of Indo-European development. This ancestor would’ve included the consonant *b.

To strengthen this case, we can add that some descendants show traces of ablaut, the vowel-swapping system of grammar that is definitive of ancient Indo-European. Also, I believe that the tree of which apples are the fruit comes from Central Asia, not a million miles away from the steppe homeland of Proto-Indo-European. This makes the presence of a word for ‘apple’ in the proto-language plausible.

So, now: case closed? Don’t words like apple support the existence of *b in Proto-Indo-European? The pool of evidence may be shallow, but such words are clearly of great antiquity.

It’s still a no. The ancestor of apple cannot confidently provide evidence of *b in Proto-Indo-European, because we cannot confidently reconstruct it that far back. Returning to my wobbly-chair metaphor, the strongest structure is one that has its four legs spaced well apart in geography and time. This is to say, it should draw from evidence from across the Eurasian spread of Indo-European languages, and from the family’s oldest-documented members, like Hittite. This way, descent down from Proto-Indo-European becomes the best explanation for the wide distribution.

Our apple-family of words is more or less limited to the Baltic, Celtic, Germanic and Slavic branches of Europe. The Abella-connection from ancient Italy is tenuous, while the candidates for relatives over in Indo-Iranian are unconvincing. We have a possible word for apples among the Hittites, šam(a)lu-, but this looks more like unrelated words in Latin (mālum) and Ancient Greek (mêlon).

The Balto-Celto-Germano-Slavic limitation of apple-words cannot avoid the alternative explanation that a loanword has been at work – that apple doesn’t go back to Proto-Indo-European, but rather to an external word that was adopted into its later offshoots in Europe.

This theoretical alternative would account for both the presence and absences of related apple-words across the family. Having been borrowed in Europe, it wouldn’t appear in the Asian branches. It would also offer a solution for why it contained a rare consonant (*b) otherwise lacking from the earliest Indo-European speech: namely, because it wasn’t an Indo-European word.

Like so much English vocabulary, apple might be an acquisition from elsewhere. Since the 1960s, some scholars have suggested that apple falls into the category of a ‘Wanderwort’, a piece of vocabulary that made its way through the various languages of a broad geographical area, likely accompanied by the thing to which it referred.

If apple can’t offer a sure witness to *b in Proto-Indo-European, then the consonant’s list of allies grows thin. Of the other candidates proposed by scholars, none are convincing. Among inherited English vocabulary, lip, deep and thorp are among the few, each displaying the post-Grimm P. The problems throughout the small set have led Thomas Olander, in a well-titled article, to summarise the situation thus:

“Our analysis has shown that there are no candidates for initial *b in Proto-Indo-European … in earlier stages there is no evidence whatsoever for initial *b and only disputable reconstructions containing non-initial *b.”

(Olander 2022: 14-15)

Of the few candidates considered, Olander hesitantly reconstructs only two words (or rather, roots) containing *b back to the oldest level of Indo-European development. The apple-root isn’t one of them.

So, the evidence for claiming that Proto-Indo-European had a B-sound doesn’t pass muster. This is not the same as to claim that it didn’t have one – perhaps it did, but the prehistoric sound’s signal has faded so much as to be unhearable now from the evidence. The whole topic is also under threat of circular thinking; if we are led to believe that there was no *b in PIE, we might then miss or dismiss evidence for it.

Is it odd for a spoken language to lack a particular sound? Certainly not! The world’s countless varieties of speech use only a subset of the consonants and vowels that humans can produce. A sound that’s normal in European languages can easily be alien elsewhere; Arabic, for example, has historically lacked the plosive /p/. Speakers may substitute it in foreign words with /b/ instead.

Nonetheless, Proto-Indo-European seems to have utilised a lot of consonants (twenty-four by my reckoning), among which were both voiceless *p and aspirated *bʰ. The consonant *b would’ve also fitted in along side the unaspirated stops *d and *g, for which we have plentiful evidence. The hole has certainly been the subject of much speculation:

“The marginal status of *b is difficult to understand from a typological viewpoint and is totally unexplainable within the traditional framework. This problem was investigated in 1951 by the Danish scholar Holger Pedersen. Pedersen noted that, in natural languages having a voicing contrast in stops, if there is a missing member in the labial series, it is /p/ that is missing and not /b/”

(Bomhard 1988: 9)

The gap in the inventory between the two is therefore pretty odd, for sure, but it’s not beyond the borders of possibility.

This altogether provides a good example of how following the trail of etymology can lead us deep into the woods, where lurk languages of prehistory that we would scarcely recognise as related to our own. A word as down-to-earth as apple can be the starting point for that etymological encounter.

So, that’s the complicated etymology behind the Ps in apple. In another post, I might tell you what the problem is with its initial A.

END.

Subscribe now

References

• Bomhard, A. R. (1988). Recent Trends in the Reconstruction of the Proto-Indo-European Consonant System. Historische Sprachforschung 101(1). 2–25.

• Clackson, J. (2007). Indo-European Linguistics: An Introduction. Cambridge University Press.

• Hamp, E. P. (1979). The North European word for ‘apple’. Zeitschrift für celtische Philologie 37. 158–166.

• Joki, A. (1964). Der wandernde Apfel. Studia Orientalia 28. 1–17.

• Kroonen, G. (2013). Etymological Dictionary of Proto-Germanic (Leiden Indo-European Etymological Dictionary Series 11). Brill.

• Kroonen, G. (2016). On the origin of Greek μῆλον, Latin mālum, Albanian mollë and Hittite šam(a)lu- ‘apple’. Journal of Indo-European Studies 44. 85–91.

• Markey, T.L. (1988). Eurasian ‘apple’ as arboreal unit and item of culture. The Journal of Indo-European Studies 16. 49–68.

• Olander, T. (2022). To *b or not to *b: Proto-Indo-European *b in a phylogenetic perspective. Historische Sprachforschung 133(1). 182–208.

• Piwowarczyk, D. R. (2014). The Proto-Indo-European root for ‘apple’ and the problem of comparative reconstruction. Studia Etymologica Cracoviensia 19. 161–167.

• Ringe, D. A. (2006). From Proto-Indo-European to Proto-Germanic (Vol. 1). Oxford University Press.

• Stifter, D. (2019). An apple a day…. Indogermanische Forschungen 124(1). 171–218.

Images my own or from Wikimedia/Wikipedia.

For reasons of pure linguistico-historical nerdery and joy, I’d like to introduce you to a special document from my beloved era of Late Antiquity. It comes from Egypt, which today we broadly think of as an Islamic, politically independent country where people speak Arabic. In the days of this document, though, Egypt was Christian, ruled by Roman emperors in Constantinople, and a place where people spoke Greek – among other languages, one of them apparently being Armenian.

Written in Armenian letters for an unknown individual navigating the Greek-speaking society of Roman Egypt, this document is an absolute goldmine of historical and linguistic information. It’s both a testament to a multicultural Mediterranean world, and a valuable early witness to the Armenian language and its speakers. This is in spite of the fact that it doesn’t contain a single word of Armenian.

Subscribe now

First things first: what we have here is an ancient piece of papyrus, written over with neat ink lettering. The script is Armenian, an alphabet exclusively devised for Armenian speech in the early 5th century AD. Its invention is attributed to the saintly Mesrop Mashtots, and, although there are some apparent external influences (such as in the alphabetical order), it’s very distinct from other writing systems of the world.

The creation of the alphabet was a natural consequence of Armenia’s embrace of Christianity in the previous century – the first state to do so officially, something that Armenian Christians are proud of. Armed with their own script, Armenians were ready to translate the Word and its holy texts. We therefore see an eruption of Armenian language and literature around this time. From then on, we can get the internal perspective, where previously we are reliant on references by outsiders (e.g. the Persians, the Greeks).

This document, cautiously dated to around the 5th–7th century AD, is a very early example of the Armenian alphabet, and the only one written with papyrus for its material. Yet it doesn’t come from anywhere near lands ever known as ‘Armenia’, nor does it write down Armenian speech. Its provenance is unclear. The French scholar Auguste Carrière bought the parpyrus from a dealer at the end of the 19th century. Scholars worked off a photograph of just one side until the original was rediscovered in 1993 by historian Dickran Kouymjian at the French Bibliothèque Nationale (designation: BnF Arm 332). Before Carrière, the trail goes cold, but the arid, papyrus-preserving climate of Egypt is the likeliest resting place. As for its language, the document is nothing but words of Greek.

Line after line, the document faithfully renders nouns, adjectives, verbs, phrases and whole sentences of Greek in Armenian letters. Following James Clackson’s (2000) reading and transcription, this is a taste of how it looks:

… ԱՆԹՐՈՊՈՍ:ԻԼԻԿԻԱ:ԿԵՓԱՂ:ՈՓԹԱՂՄՈՍ:ՈՒՏԻՆ…

(BnF Arm 332a: line 26)

Matching each Armenian letter up with its closest Greek equivalent gives us this version:

… ανθροποσ:ιλικια:κεφαλ:οφθαλμοσ:ουτιν…

Finally, these words, helpfully kept separate by a colon, we can render in standard Ancient Greek and translate:

… ἄνθρωπος (‘man’): ἡλικία (‘age’): κεφαλή (‘head’): ὀφθαλμός (‘eye’): ὠτίον (‘(little) ear’)…

Behind the Armenian and the Greek, we find a list of body parts waiting for us! This is the essence of the Graeco-Armenian papyrus; it’s not great literature, but rather a delightfully practical document.

In other parts, we have humble vocabulary for the kitchen and the table (“pan: jug: cup: plate: cup”) and for the family (“sister: daughter: aunt: uncle: companion: lap: wifeless”). There are practice verbs, familiar to any student of Greek: “we have: I have: you have: he has”. There are useful Greek phrases for everyday life: “you did not pay the price… I have a book… I want to learn… today I am not at leisure, tomorrow I am going… where do we go: show me the road… I ate well: eat don’t wait”.

There are more complex sentences too, such as memorable sayings about the philosopher Diogenes. Biblical quotations, monotheistic well-wishing (“God help you”) and the twelve months of the Egyptian calendar together point to the document’s origin in Greek-speaking Christian Egypt.

Now, I see two seams of information to be excavated from the papyrus: one about historical language (quelle surprise), but another about historical society. Let’s dig into the first.

Evidence for language past

The thing is, the papyrus is an excellent acoustic witness to how Greek sounded back then. The language, caught between its Koine and Medieval forms, is plentifully attested in historical sources from Late Antiquity, but such sources are necessarily silent. We know that Greek speech has undergone many changes down the millennia, but pinpointing when (and where) these changes occurred tends to be imprecise – we say things like ‘Oh, that consonant shifted during the Koine period’, which narrows things down to about nine hundred years.

This touches on a tension that I explored in Why Q Needs U: when speech changes, writing can often stand perfectly still. For alphabets and other systems based on the sounds of a language, if instances of a particular sound change across the board, without exception, then its associated written symbol need not shift too. It straightforwardly spells one sound, both before and after the change.

For instance, the Greek letter B originally stood for the consonant /b/. That stop sound has since shifted into the fricative /v/. This has turned the letter’s name from beta into ‘veta’, but the letter is otherwise unbothered; now, it spells /v/. The letter’s inertia obscures the fact that any change in speech has taken place.

Alternative scripts are therefore of great importance for the historical linguist. Rendering speech in new letters is not bound to any archaic spelling and established standard, but instead is accurate to sound. This Armenian spelling of Late Antique Greek lifts the veil on the spoken language, giving us a precious glimpse of what changes had (or hadn’t) occurred.

There is good evidence that the author was not reading and transliterating written Greek, but rather listening attentively to the spoken language. The Armenian alphabet has a broad array of letters at its disposal, and the author of the papyrus chose carefully among them, on the basis of either what they heard in Greek speech, or what they thought they heard.

The Armenian alphabet has two letters, Լ and Ղ, which at the time spelled two L-like sounds. These were a ‘light L’ and a ‘dark L’, as can be heard in many accents of English. In my accent, Armenian Լ would spell the L at the start of like and language, while Ղ was for the L at the end of ball and ill. Both are used consistently in the papyrus, in spelling words for which Greek has only one letter: Λ (lambda). Rather than chaotic employment of Լ/Ղ for Λ, the author appears to have identified a ‘light/dark’ difference in spoken Greek and applied the letters accordingly. If correct, this is news to me; I’ve not read of such a distinction in Greek elsewhere.

The active choice of Armenian letters offers us whispers of the accent behind the words. For example, the alphabet has Բ, which stood and still stands for the voiced stop sound /b/. Greek words spelled with B are here mostly Armenian-ised with Բ, rather than another letter that would indicate a thorough shift towards the voiced fricative /v/, such as the W-letter Ւ. There are a couple of spellings that hint at the shift’s onset, though. For instance, “ՍԱՒԱՆ” (‘sawan’) spells σάβανον ‘linen cloth’.

The consonant behind the Greek letter Χ has also changed over the centuries, from an aspirated /kʰ/ to a fricative /x/, as in Scottish loch. The Armenian alphabet could provide suitable letters for both the older and the younger sounds (namely, Ք and Խ). We observe that Χ-words in usual Greek writing are spelled in the papyrus with Ք, indicating the older sound.

Relatedly, Armenian has a letter for the breathy /h/ sound in hat: Հ. The author of the papyrus often uses it at the start of Greek words that have since dropped their Hs. It’s there in “ՀԷՄԱ”. This is the Ancient Greek word for ‘blood’, αἷμα, the origin of English haemo-. It’s pronounced like a breathless ‘ema’ in Modern Greek, but was ‘hema’ still for our author. That said, the Հ is absent from other possible places. The variability gives the impression that H-dropping in Egyptian Greek was common, but not yet ‘good’ Greek.

These features together give the Egyptian accent of Greek behind the papyrus a fairly conservative, quasi-classical feel. Many of the sound changes that are standard and normal in Greek today don’t seem to have been fully present in Late Antique Egypt.

Some are, mind you. An infamous change in post-classical Greek is ‘iotacism’. This has been a gradual merger of several vowels. Once, only the letter Ι (iota) represented the /i/ vowel heard in English sheep and Greek. Now, it has been joined in that pronunciation by Η (eta), Υ (upsilon) and three pairs of letters. There is evidence for iotacism in the papyrus, based on spelling misalignments or ‘mistakes’, but the process is just getting started.

The vowels behind Ι and Η have apparently merged, because they’re both spelled with the Armenian equivalent to iota: Ի. This explains the “ԻԼԻԿԻԱ ~ ιλικια ~ ἡλικία’ example above. The author likewise spells the eta-words χήρα ‘widow’ and ἥβη ‘youth’ with Ի.

Yet Greek Υ gets its own rendering in Armenian: ԻՒ (‘iw’). This is clearly the author’s way to represent the front vowel /y/ that the letter historically stood for. ԻՒ is for the most part kept unmerged and unmistaken for Ի. We find “ԴԻՒՆԱՏՈՍ” (‘diwnatos’) for δυνατός ‘strong, able’. But there are several exceptions, like ՓԻՍԻՍ (‘pʼisis’) for φύσις ‘nature’. On this basis, we can say that the /y/ vowel behind Greek Υ was still mostly pronounced separately from the /i/ vowel behind Ι and Η, but it was beginning to shift and merge.

As well as innovations in sound, there are some in the realm of words and their construction. Modern Greek is today littered with words of the neuter gender that tend in -ι. These were once diminutives, smaller or more casual versions of basic nouns, formed through the ending -ιον. The Greek behind the Armenian appears to have fully embraced these derived diminutives; instead of σπάθη ‘blade, sword’ and κοντός ‘pole, spear’, its source preferred diminutive σπαθίον and κοντάριον. These look ahead to the Modern Greek words σπαθί and κοντάρι.

Altogether, the evidence for the Greek language behind the Armenian letters is just what we expect to see: in some places conservation, in others innovation. Like any stage of any living language, it preserves elements from the past, while also developing features that will define its future. The Greek of the Graeco-Armenian papyrus sits on the turn of the language’s ancient and medieval chapters. What’s significant is that it gives us close-to-direct evidence for these changes in progress.

But why does this witness to living Greek even exist? What was the purpose of the papyrus? And who was it for?

Evidence for people past

While the document is anonymous, both its content and its context reveal a great deal about its writer and reader. The penmanship is consistent, indicative of a single author, and naturally arose within an Armenian social environment. This in itself isn’t unbelievable; Egypt today is some distance from the modern county of Armenia, but the Armenian state and people could historically be found over a much larger area, even touching the eastern Mediterranean sometimes.

Besides, Egypt was a powerhouse of antiquity and an obvious destination for people travelling, trading and migrating. As I’ve sought to demonstrate in many of my articles, people in the ancient world really did travel around, and left samples of their languages behind. For example, a beautiful mosaic of birds entwined in vines, found in Jerusalem and dated to the 6th century AD, includes a strip of early Armenian writing at the top:

Returning to our papyrus, the linguistic evidence, specifically the choice of Armenian letters, indicates a free hand in writing down spoken Greek. Correspondingly, we can infer a lack of familiarity with written Greek and its conservative spelling. Otherwise, it might have shaped the author’s choices. The same unfamiliarity extends to the society and lifestyle of Hellenic Egypt, from household objects to the local calendar and smart-sounding references.

This altogether draws an outline of a newcomer to Egypt, having travelled from the northeast and now finding their (I’ve not spotted any direct indicators of their gender) way around the bustling, multicultural and multilingual streets of its cities. Some, such as Alexandria, had become world-changing centres of Greek language and literature. Being already fluently literate in Armenian, our newcomer was most likely an adult, or at least an elder youth.

It’s reasonable to describe this papyrus as a kind of phrasebook or aide-mémoire, an informal guide for remembering essential Greek vocabulary. It might be created for personal use, or even to distribute among new arrivals from Armenia. Yet, there is a glaring omission that speaks against this: there are no translations into Armenian.

It’d be a poor phrasebook that doesn’t make its vocabulary and phrases understandable in the reader’s own language. Instead, as James Clackson (2001a) suggests, a likelier context is one in which the papyrus didn’t stand on its own, but rather was a product of an educational setting, with a teacher who supplied the Greek words’ meanings. That is, its author was an adult Armenian learning from a compatriot already fluent in Greek. What they learned, they wrote down, in their own letters, in a kind of exercise book.

“It is possible that the writer took down information by ear from an informant who himself was only dimly able to remember his own educational experience.”

(Clackson 2001a: 216)

If our author was receiving Greek via a fellow Armenian, there might have been some interference in the signal; the instructor’s accent could have been non-native-like. This provides a possible origin of the ‘dark/light’ difference between Լ and Ղ: not in native Greek speech itself, but in someone’s Armenian accent when speaking Greek.

The casual context would also explain the apparent lack of overarching structure behind the entries. They read like someone rattling off whatever comes to mind: “Ah yes, there’s the foot, nerve, vein, blood, body, skull, brain. Also: sheath, sword, dagger, bow, but hopefully you won’t need those. Now, as for verbs, it can be tricky…” This also accounts for the inclusion of quotes and sayings, giving the learner a crash course in both the linguistic and the cultural local lexicon – “Oh, there’s this saying about Diogenes that goes down well with the Greeks…”

If this interpretation is correct, then the papyrus attests to at least two Armenians in Late Antique Egypt, with an established resident teaching a recent arrival. While still finding his feet, the newcomer is clearly committed to learning Greek and making a go of his new situation. Many of us today will be all too familiar with the daunting experience of being a stranger in a strange land. We can sympathise and imagine the daily challenges and stress. Thankfully, our author had a mentor on hand.

I have loved this papyrus (from afar; we haven’t met in person) for four or five years now. No other such documents come to mind as true rivals for my affection; I think I’ve made a good case for why. This unique survival from Late Antiquity is an invaluable witness to everyday real language, and therefore to the real people who lived through that language.

END.

Subscribe now

References

• Clackson, J. (2000). A Greek Papyrus in Armenian Script. Zeitschrift für Papyrologie und Epigraphik, 129. 223–258.

• Clackson, J. (2001a). A Greek Educational Papyrus in Armenian Script. In Andorlini, I. et al. (Eds.) Atti Del XXII Congresso Internazionale Di Papirologia, Firenze, 23–29 Agosto 1998. 207–218.

• Clackson, J. (2001b). Vox armeniaca: new evidence for the pronunciation of Classical Armenian. Slovo, 26(1). 43–52.

• Kouymjian, D. (1996). Unique Armenian Papyrus. In Sakayan, D. (Ed.) Proceedings of the Fifth International Conference on Armenian Linguistics Delmar NY. 381–386.

• Matasović, R. (2019). A grammatical sketch of Classical Armenian. University of Zagreb.

Original Substack article. Images my own or from Wikimedia. Cover image: photo (from Wikimedia) of a fragment of BnF Arm 332.

This can be of a single named individual or of an uncountable number of nameless victims. Violence is more prevalent in some books of the Bible than others¹; the Book of Judges, for one example, explores the consequences of (un)faithfulness to God through the narrative of its stars, the Israelites, doing awful things and having awful things done to them. Judges 3 moves quickly from the bloody oppression of the Israelites by the neighbouring Moabites, to the gutsy assassination of the Moabite king and the killing of 10,000 of his subjects, all in the short space of seventeen verses.

Yet, despite the horror of the deeds and the magnitude of the slain, it’s possible for these passages to wash over their readers, to seem unremarkable, vapid. This may be due partly to our unnerving cognitive ability to process high numbers of fatalities unfeelingly (as Stalin probably didn’t once comment on), and partly to scepticism in the numbers. I have no doubt that life in the Bronze Age could be nasty, brutish and short, but the thousands of dead have the ring of authorial exaggeration, written centuries after the fact. If actually true, the total death toll of the Hebrew Bible would surely have left the region a depopulated wasteland.

Perhaps the indifference of readers is also in response to how the passages are constructed. Their simple, often impassive description of events, combined with an unfamiliar social context and repetitive conjunction (‘and … and … and…’), means that the passages – and, importantly, subsequent translations of them – are unlikely to provoke emotional responses from their modern audience.

Add in a deep cultural immersion in these texts and their presentation in unusual, archaic English, and you have a situation where Monty Python can make them a target of parody:

“...And spotteth twice they the camels before the third hour. And so the Midianites went forth to Ram Gilead in Kadesh Bilgemath, by Shor Ethra Regalion, to the house of Gash-Bil-Bethuel-Bazda, he who brought the butter dish to Balshazar and the tent peg to the house of Rashomon. And there slew they the goats, yea, and placed they the bits in little pots.”

(Monty Python’s The Meaning of Life)

I am fully guilty of this dispassionate attitude towards ancient texts, Biblical or not. Consequently, I deeply appreciate any means of jolting and resetting my conception of them, thawing out stories frozen by genre, and getting back to the real people behind the texts. I think that one way to do so, you will not be surprised to learn, is via language.

Subscribe now

Shibboleth is a great English word. True, its greatness is a subjective quality that the word itself is unaware of, but I’d bet that other English speakers will agree with me. Any pleasure derived from shibboleth is due to its lexical rarity, fricative sounds, and composition as a word that unconsciously marks it out as a borrowing into English. Shibboleth does indeed come from elsewhere, specifically from Biblical Hebrew.

It can be used in English with a handful of related meanings; the Cambridge Dictionary provides two:

“a belief or custom that is not now considered as important and correct as it was in the past”

“a word, phrase, custom, etc., only known to a particular group of people, that you can use to prove to them that you are a real member of that group”

These contrast greatly with its meanings for speakers of its language of origin: ‘ear of grain’ (i.e. of wheat) and/or ‘flood, stream’.

The English usages arise from a single context in which the word was written, a story recounted in the twelfth chapter of the Book of Judges. It’s not unusual for the book, being yet another account of one Biblical people (the Gileadites) slaughtering another (the Ephraimites). The Ephraimite dead are reckoned at 42,000 at the cessation of hostilities. Yet the episode is notable for a small detail that gives it an air of remembered reality.

In this story, the major geographical landmark is the river Jordan, flowing, as it still does, from north to south. Looking out from their home on the west side (Cisjordan) was the Tribe of Ephraim. On its east side (Transjordan) lay the region and people of Gilead, chief among them being the judge Jephthah. Don’t imagine a legal bigwig for the term judge in this context; the judges of the Hebrew Bible are more active, more martial individuals.

The necessary context is that the Ephraimites and Gileadites were feuding. The former hadn’t helped the latter during an earlier time of peril, and then the Ephraimites had the nerve to be bitter that they were not invited to join in with the Gileadites’ battle and eventual victory against the Ammonites. This, by the way, was not a fight against fossilised shelled mollusks, but rather a separate people, living to the south of Gilead in what is today Jordan. The Ammonites gave their name to the capital city of the modern-day country of Jordan, Amman.

The Ephraimites fought the Gileadites, lost against them, then fled back towards the Jordan and home. But the Gileadites were waiting for them there, having secured the crossings already. You could pass their blockade and reach the safety of Cisjordan by passing a test: just say the word shibboleth. It sounds simple enough, Pythonesque again in its anticlimax, but the Ephraimites fatally failed by the thousands.

The Ephraimites could not pronounce the word correctly, thereby signallying their social allegiance. The Hebrew text of Judges conveys the right and wrong pronunciation with two spellings: harmless שִׁבֹּלֶת and deadly סִבֹּלֶת. They differ only in their first letter: ש‎ (rendered Roman as shibbolet) and ס (in Latin letters: sibbolet). After an Ephraimite answered with sibbolet, death was dealt out.

It’s from this test that the English word gets its meaning of a distinguishing feature for a particular group. As said, its meaning at that time was maybe ‘ear of grain’, maybe ‘flood, stream’. Its varied translations could be due to a case of homophony through a merger of two unrelated words. Given the riverside setting, the ‘flood, stream’ sense would make more sense. Regardless, the meaning was unimportant. What mattered was its phonetics; mispronunciation was a fatal misstep.

This challenged later translators of the Hebrew Bible. Do they translate the word, to make it intelligible to their new readership? Or stay true to its phonetic purpose by leaving it untranslated? Responses have differed. The scholars behind the Septuagint went with the first option, rendering it as “στάχυς”, the Ancient Greek word for ‘ear of grain’. St. Jerome in his Latin Vulgate left it as “scibboleth” instead. SC represents the sibilant that the Roman alphabet has no single letter for, being absent from native Latin words. I have to wonder how the Romans shushed one another.

But what is actually going on with this story? Is this a tale of a tongue-twister, unknown among the Ephraimites? It’s actually something bigger than this one word; it was a matter of dialect.

Based on the animosity and conflict chronicled in books like Judges, you could be forgiven for thinking that the peoples and groups mentioned were totally alien to one another. Instead, the historical and archaeological evidence indicates that all of this disharmony was actually happening within a shared culture of belief, lifestyle and language.

Ephraimites, Gileadites, Ammonites and many other characters of the Biblical story (even the Phoenicians to the north) were speaking flavours of a broader ‘Canaanite’ language. This in turn belonged to the wider Semitic family. It would at one point have been unified, before its disintegration into new accents and dialects across a broad geographical area. Of the offspring of Canaanite, only Hebrew has survived until today, thanks to its enduring sacral function and return to speech in the modern era.

The line between dialect and language is vague (the Ephraimites and Gileadites had armies but probably not navies). Nonetheless, the story of slaughter beside the Jordan implies that the two groups could communicate with each other perfectly – so well that a fleeing Ephraimite soldier wouldn’t give himself away after saying “Let me cross over”, but only after the single trap word.

Accents and dialects emerge through accumulated small differences, such as in the sounds of speech. A particular group of speakers may shift a sound that another group retains; the innovative and the conservative now differ, but correspond systematically. One such difference lies behind the shibboleth story. A leading explanation is that of E. A. Speiser (1942), namely that the Gileadites pronounced the initial sound of shibboleth conservatively, and the Ephraimites innovatively.

The theory goes that, east of the Jordan, Gileadites and Ammonites preserved a /θ/ consonant (as in English thistle and thorn) in relevant words – the safe pronunciation to their ears was ‘thibbolet’. This /θ/ was conservative, inherited from Proto-Semitic, the ancient ancestor language.

Meanwhile, over in Cisjordan, instances of /θ/ had shifted and merged with the existing sound /ʃ/ (as in English words like shift and shibboleth). This accentual innovation cost many Ephraimites dearly. They couldn’t pronounce /θ/, and tried in vain to substitute it for a /s/ sound. This is the reason for the ‘wrong’ answer of סִבֹּלֶת (sibbolet). Our English pronunciation of shibboleth is indicative of the scriptures’ origin west of the river.

Whether the phonetic details of this theory are correct or not (it has its gainsayers), what is surer is that the Jordan was a faultline not only within the land, but also within a colourful spectrum of Canaanite speech. The different hues could signify, if not also contribute to, social frictions that might spark violence.

As an aside, I had to get these Semitic sound shifts fixed in my head while writing Why Q Needs U, because it’s in this corner of space and time that our alphabet is developing. The key players in that development are the aforementioned Phoenicians. Their version of the already-old letters, well adapted to their version of Canaanite speech, was highly influential in the region of the eastern Mediterranean. It held so much prestige that, in addition to being embraced by the Greeks, other Semitic languages adopted it without adaptations that would better suit their particular phonological needs.

The Phoenician accent seems to have been distinct in what consonants it had merged in speech; the three Proto-Semitic sounds */ʃ/, */θ/ and */ɬ/ had become simply /ʃ/ for the Phoenicians. The primary evidence for this is that words that once included them were spelled there with just one letter: 𐤔. This is the origin of Greek Σ and Roman S.

Elsewhere, the southwestern accent that would become Hebrew merged the first and second, but kept the third consonant separate. Yet the Phoenician alphabet was the foundation for the Hebrew, and could offer no extra letter. The Hebrew letter ש‎ (from Phoenician 𐤔) consequently has to double up and spell two unmerged sounds. These are today /ʃ/ and /s/. Later scribes solved the ambiguity by adding the separate letter ס after and above ש, when it stood for /s/. This has since simplified into a little dot: שׂ‎.

What do I want to achieve with all of this Biblical and philological information? Do I want to share a sinister delight in the story of the massacre of the Ephraimites? No, I do not. Do I want to provide an accessible introduction to some hardcore historical linguistics? Yes, obviously, but I have a bigger goal in mind: to restore a bit of life to ancient people.

Behind the possibly historical and definitely horrible shibboleth incident, there lies a lost world of linguistic and social variation – not of impermeable blocks of two-dimensional actors, but of real people, getting along and getting past minor differences. In between instances of conflict, we can imagine many more moments of harmless inter-linguistic awareness in Biblical times, as people said things like ‘What, “shibbolet”? Talk properly!’ and ‘That lot over the Jordan do talk funny’ (this could work in both directions).

If the Speiser account is correct, then we have a case of people swapping ‘TH’ for ‘S’, something that happens among plenty of adult learners of Modern English. It reminds me of the famous Berlitz advert, in which a German junior coastguard asks the crew of a ship in peril “What are you sinking about?” It’s a reasonable phonetic approximation, one that would’ve also occurred during happier interactions across the Jordan.

An appreciation of the dialects that ancient people must have had – hearing and speaking in them, assigning and associating them with certain groups, just as we do today – allows me to appreciate the reality of ancient people. Even in the dark places of the texts that document these people, amidst all the drama and death, there are glimpses of real life to be enjoyed.

END.

Subscribe now

Footnote:

1. That said, I think I’m correct in rejecting the simple characterisation of the God of the Hebrew Bible/Old Testament as wrathful, in contrast with a God of love and peace in the Christian New Testament. You can find tenderness and quietude with the former, as well as violence or threats thereof with the latter.

References:

• Blau, J. (2010). Phonology and Morphology of Biblical Hebrew: An Introduction. Linguistic Studies in Ancient West Semitic. Eisenbrauns.

• oed.com

• Rendsburg, G. A. (1988). The Ammonite phoneme /ṯ/. Bulletin of the American Schools of Oriental Research, 269(1). 73–79.

• Rendsburg. G. A. (2013). Shibboleth. In Khan, G. et al. (Eds.) Encyclopedia of Hebrew language and linguistics (Vol. 3). 556–557. Brill.

• Speiser, E. A. (1942). The Shibboleth Incident (Judges 12:6). Bulletin of the American Schools of Oriental Research, 85. 10–13.

Original Substack article. Images my own, from Wikimedia, or from credited sources. Cover image: 14th-century illustration of the character of Shamgar, mentioned in the Book of Judges.

With my thanks, as always, to Benjamin Suchard for letting me bother him with DMs about ancient Semitic phonology.

Such a contrasting blend of light and dark colours its history too; this region is now a magnet for swish skiers and rucksacked hikers, but barely hides the scars of its wartime experience.

The towns and villages of Krkonoše were overwhelmingly German in identity from the Middle Ages up to 1945-1946. These Bohemian Germans’ ancestors had gone east to mine, cultivate and settle the uplands; the territories where they settled are known collectively as the Sudetenland. The defining population was expelled at the end of the Second World War, but fragments of their successful lives and non-Slavic language endure.

For me, arrival into Krkonoše is marked by the sight of giant factories and mansions for management, many of them now skeletal in dilapidation. I sense I’m crossing a fault-line not just within the country, but in Europe. Here the faded burned-yellow stone and plaster of Bohemian Baroque, the most northern bastion of Italy and Counter-reformation, gave way. It surrendered to the confident southward march of granite-grey industry, Prussian and Protestant. That overlapping wave in turn broke; now a torrent of tourism (but not much else) washes over it.

With such turbulence in its human geography, we might look for calm in the nature of Krkonoše – such as in its rivers. But no, these too bear witness to the maelstrom of history, albeit with slower currents.

Subscribe now

The chief of the rivers that flow through Krkonoše is the Elbe; I’ve previously written about that prince of waterways in great detail. My most recent visit instead inspired linguistic reflections about a considerably less famous river: the Mumlava.

Etymologically, it’s the mumbler (in older German: mummeln; in Czech: mumlat). It by no means ranks among the great rivers of Europe; the Mumlava rises just to the south of the source of the Elbe, mumbles its way for twelve kilometres, then spills out into the Jizera. It wouldn’t be known at all beyond the wardens and fans of Krkonoše, were it not for the Mumlava Waterfall, the largest in the country. This theatre of crashing water is a popular destination, a breezy stroll from the town of Harrachov, although the winter-iced path down to it this time offered me a crash-course in luge.

Given the region’s historical inhabitants, it seems to have been on the basis of the German verb mummeln that the Mumlava was first christened the Mummel. Czech speakers then modelled their own name for it on that German original. This they achieved by adding the ending -ava.

This Czechification of the name brought the Mumlava into line and rhyme with other rivers in the country; there’s the Sázava, the Jihlava, the Úhlava, the Otava, the Oskava, the Opava, the Morava and the Vltava. Last weekend saw me hiking along the south bank of the first until it flows into the last, outside the small Bohemian town of Davle.

Their common -ava ending was bestowed on the Mumlava, a sort of hydronymic suffix to make it sound like a proper Czech river. The thing is, this ending is not part of Czech’s core of Slavic vocabulary – it’s not something the language has inherited from its prehistoric Slavic origins. Instead, naming rivers with -ava is a later practice that the Czechs-to-be extracted from names already in use when they first arrived in Bohemia and Moravia.

When that happened, probably in the 6th century AD, those lands were hardly blank canvases, devoid of inhabitants. Bohemia and Moravia had for centuries been home to Germanic-speaking groups, their language(s) being related to English, German and Dutch.

These people are more than abstract archaeology; they are a colourful presence in Latin and Greek sources from antiquity. One group, the Marcomanni, were the staunch enemies of the emperor Marcus Aurelius (121–180 AD). From their base in Bohemia and Moravia, their attacks on the Danube frontier diverted Roman resources and attention. If you’ve seen the 2000 film Gladiator, the Marcomanni are the fearless yet hopeless enemies in its initial battle.

The Germanic linguistic ingredient can still be found in the Czech lands. Despite the strength of the later Slavic wave, geographical terms attest to their predecessors. Bohemia, a term that English gets from Latin, was most likely coined to mean the ‘Boii-home’ originally. This etymology makes the -hem- of Bohemia a cognate of English home and German Heim.

Landmarks of the landscape maintain prehistoric language too. The Elbe is one river whose name predates the coming of the Czechs; another is the Vltava. This would’ve been the *Wilþ-ahwō to the Marcomanni – the wild river. Here’s another link to English: the Vlt- of Vltava to the adjective wild.

Meanwhile, the second bit, *ahwō (meaning ‘river’), is behind the -ava part of the river’s name. It would’ve sounded like ‘akhwoh’, just a couple of sound changes away from -ava. Several others among the -ava rivers have been derived from Germanic names that include *ahwō. There are enough of them for us to suppose that the incoming Slavs adopted some of the old names, and from them extracted the -ava element (which was presumably meaningless to them) to be applied to new coinings.

This endurance of river names amidst population changes is nothing unusual. All across the world, waterways speak for people and language now otherwise silenced. For example, a single root (via a couple of lost languages) seems to lie behind Europe’s Danube, Don, Dniester, Dnieper and Donets. The distribution of river names of Baltic origin flows far beyond where we today find Baltic languages.

Some rivers of Britain, including the Thames, are widely derived from languages predating even the Celts’ arrival on the island. The Celtic language of Iron-Age Britain has retreated westwards in the face of English, leaving stranded in England the river Avon – straightforwardly a sister of the Welsh word for ‘river’ (afon).

All of this makes the field of hydronymy intoxicatingly interesting, as it redraws the map and resurrects ghosts from geography. Rivers can transport us as far back into linguistic time as we can go. For some, they rise from a barely visible Europe, before the coming of the Indo-European languages that would later dominate it.

My speculative explanation for this longevity is that rivers have to be shared; it is harder to limit a river to one community or culture, compared with a fort, village or town. A language that settles on the bank of a river will not quickly find moorings all along the river’s length, or even on the other side. Its speakers will at first share the river with established communities, and adopt the current name from them.

That Germanic word *ahwō, which gave Czech rivers their -ava, may have rung a small bell in my reader’s mind. It is ultimately a prehistoric word, but one supported by plenty of evidence. It can be found in our sources for historical Germanic languages, such as Gothic. It appears once in our fourth-century Gothic translation of the Bible.

𐌳𐌰𐌿𐍀𐌹𐌳𐌰𐌹 𐍅𐌴𐍃𐌿𐌽 𐌰𐌻𐌻𐌰𐌹 𐌹𐌽 𐌹𐌰𐌿𐍂𐌳𐌰𐌽𐌴 𐌰𐍈𐌰𐌹 𐍆𐍂𐌰𐌼 𐌹𐌼𐌼𐌰
daupidai wesun allai in Iaurdane aƕai fram imma
‘Baptised were all in the Jordan river by him’

(Gothic Bible, Mark 1.5)

From centuries later, *ahwō also appears, greatly reduced, as words for rivers and streams in Nordic languages: å in Swedish, Danish and Norwegian, á in Icelandic and Faroese. It doesn’t stand alone in their cousin English, but has met with success as the first part of the word island. This old compound word, literally ‘river-land’, is first written in English as “iland”, “igland” and “ealond”. Our spelling with S today is a product of the early modern period; the first element was confused with the unrelated word isle, and English hasn’t yet dared to eject the unetymological letter.

But none of that is likely why that lone bell of recognition is tolling. Instead, the appearance and watery meaning of *ahwō bear a resemblance to Latin aqua. This was the Romans’ word for ‘water’. It continued to be so in the emerging Romance languages, such as Italian (acqua), Spanish (agua), Romanian (apă) and consonant-crushing French (eau).

Aqua is cognate with the Germanic word *ahwō. True, one means ‘water’, the other ‘river’, but we can envision the latter sense developing from the former. They also differ in their first consonant, but the *h of the Germanic word we can explain through the shifts of Grimm’s law that define the Germanic languages. All things considered, we can consider the two to have sprung from the same source.

What that source was may well have been Proto-Indo-European, ancestor of the whole family of which Latin and Germanic are members. It could have had a word for ‘water’ that we may label *ak^wā. From this, aqua and *ahwō could emerge.

Now, the strongest proposals (‘reconstructions’) of vocabulary for Proto-Indo-European spin together many threads; they draw evidence from several branches of the family tree strewn across Eurasia: the Germanic, the Indo-Iranian, the Slavic, the Celtic, the Hellenic, the Anatolian and so on. They thereby mitigate for mutual influence and chance, so that ancestry from Proto-Indo-European becomes the most plausible theory.

In this case, with aqua and *ahwō, these are all the threads we have to weave with. No certain trace of cognate words can be tracked elsewhere in Indo-European. A two-legged structure for reaching Proto-Indo-European has nothing else to lean on. It stands, but wobbles.

The limited distribution of cognates invites healthy doubt and speculation. It remains a possibility that the single source of aqua and *ahwō was indeed stored in the lexicon of Proto-Indo-European, only to be lost in all of its descendant lines save two. But the door is also open to an instance of borrowing.

This alternative explanation would envision a word, in a language local to central and southern Europe, that was adopted into the Indo-European family from outside. Such an external origin was considered by the scholar Robert S. P. Beekes, for one. In Beekes’ view, *ak^wā belonged to the prehistoric language behind so many European river names. It acted as a ‘substratum’ that donated words to the ascendant Indo-European languages. Those words in time became all that remained of it.

If this alternative is correct – and it’s a heavy if – then we have a world of historical language behind the ending of the humble river Mumlava. It can tell a long tale of shifting populations and their languages.

The German speakers who settled the Bohemian border along the Mumlava were cast out in 1945 and 1946; Czech speakers have since filled the void. Centuries earlier, the Czechs-to-be had gained both river names and a hydronymic suffix (-ava) from the Germanic language present on their arrival into Bohemia and Moravia. This -ava was a word in that Germanic language, one that meant ‘river’. The word developed out of a pre-Germanic ancestor, shared with Latin aqua. That in turn may have previously belonged to an unrelated and unclear language, spoken in the dark undercurrents of European prehistory.

That language is barely conceivable today, known from only whispers and fragments. Yet it once might have been hale and hearty, even displacing other tongues in the minds and mouths of ancient speakers. On and on it has gone, the flow of life and language, in which I too am of course submerged. Occasionally, I feel the truth of that fact: that my life and my language are a drop in that flow, now new, but doomed to be old and, in time, washed out to sea and oblivion.

END.

Subscribe now

References

• Beekes, R. S. (1998). The origin of Lat. aqua, and of *teutā ‘people’. Journal of Indo-European Studies, 26(3/4). 459–466.

• Blažek, V. (2010). Etymological analysis of toponyms from Ptolemy‘s description of Central Europe. Brozović Rončević, D., Fomin, M., & Matasović, R. (Eds.) (2010). Celts and Slavs in Central and Southeastern Europe. Proceedings of the Third International Colloquium of Societas Celto-Slavica, held between 18-20 September 2008 in Dubrovnik (Croatia): Studia Celto-Slavica III. Institut za Hrvatski Jezik i Jezikoslovlje. 21–45.

• Dávná a někdy trochu tajemná - jména našich řek: Český rozhlas feature (2011).

• Kroonen, G. (2013). Etymological Dictionary of Proto-Germanic (Leiden Indo-European Etymological Dictionary Series). Leiden, Boston: Brill.

• oed.com

Resources:

• Gothic Bible text and translation

Original Substack article. Images my own, from Wikimedia, or from credited sources. Cover image: the Mumlava Waterfall.

For one reason, it brings a great number of modern female first names to a close. A sample of famous names, especially if skewed towards the West, will typically include a fair few examples. Among the female winners of a Nobel Prize, we find Claudia, Bertha, Rigoberta, Gabriela, Donna, Svetlana and several others, united by both the accolade and their names’ final letter. The first ladies of the United States over the past 100 years have included another Claudia, a Thelma, a Barbara, a Laura and a Melania twice. In a somewhat less august list, eight of the nine ladies cited in Lou Bega’s 1999 hit Mambo No. 5 also bear names that end in A.

But the association between A and the feminine goes beyond proper names, of course. Several national and international languages rely on the letter and its sound to help assign nouns to a feminine grammatical gender. Some of these words have travelled into English (e.g. from Spanish: vanilla, armada, salsa).

The gender and the A are then shared with nouns’ associated words, like adjectives and determiners. To render a tall woman into Italian would be una donna alta, just as the dear grandmother in Spanish would be la querida abuela. The same association holds in other linguistic corners of the world. In my local Slavic language, Czech, short A and long Á string together phrases about female folk like:

ta nová, slavná a úspěšná žena přišla a ona je spokojená

(‘that new, famous and successful woman has arrived and she is pleased’)

Perhaps you’re now wondering why? What’s the reason for this apparent correlation between vowel and gender? Is there something intrinsically feminine about it? (No.)

As a historical linguist, my very short response would be: it’s because of Proto-Indo-European. But that, except to a handful of already au-fait nerds sagely nodding in agreement, would be a deeply unsatisfying explanation. So, what follows is my less short response. It will take you, dear reader, into a world of ancient speech and grammatical computation, and ultimately up against the bars of the cage of linguistic thinking. Brace yourself, and keep an eye on the nerdery level of each section, lest you be exposed to levels that you cannot handle.

Subscribe now

The feminine A across Indo-European

(Nerdery level: 2/5)

The single reason for so much of the aforementioned phenomena is a particular family of languages: Indo-European. This wide grouping embraces French, Spanish, Romanian, Greek, Kurdish, Persian, Hindi-Urdu, Armenian, Albanian, Pashto, Polish, our own dear English, and countless others.

It is of course perfectly possible for speakers of languages outside of Indo-European to coin and wield names that end in A, just as many female names forged within Indo-European may terminate otherwise.

With some names, the story is complicated. The Semitic name Maryám/Miryám was introduced by Christianity into the Indo-European fold, and became Ancient Greek María. This jump in language family saw it gain a final A. It then lost this through regular sound change on its way to becoming French Marie and English Mary.

Nonetheless, there is a millennia-old connection at work in Indo-European languages between A and things feminine. This is because of the languages’ shared ancestry; once upon a time, they were one. During those early days, a class of words, characterised by A, was set up by speakers of Proto-Indo-European.

This development was prehistoric, prior to our historical sources, but we can believe in it on the basis of descendant languages. The use of feminine A-endings is something that most branches of the family tree have exhibited at some point, although they differ as to how long it has remained a part of them.

For example, Latin had a whole class of nouns defined by the vowel A, nearly all of them feminine in grammatical gender. This ‘first declension’ unsurprisingly included nouns like fēmina ‘woman’ and puella ‘girl’. Yet it contains a majority of words with no connection whatsoever to humanity or its different flavours: mensa ‘table’, taberna ‘hut’, āra ‘altar’, fortūna ‘luck’, and many more.

The feminine A is very visible in ancient Indo-European languages, such as Latin and its distant cousin, Sanskrit. The latter has a whole set of feminine ‘ā-stems’. Again, some of these are obviously female in their meaning (e.g. gajā ‘female elephant’) but others not at all (e.g. sénā ‘army’).

This behaviour of Latin and Sanskrit words instantly drives a wedge into any idea that human gender and grammatical gender map onto each other. Their relationship is complex; they are separate phenomena, but not entirely unconnected. The former can shape and drive the latter. If speakers recognise that one of their grammatical genders happens to contain many words relating to a particular gender or sex within a species, they may strengthen the association by assigning new similar-meaning words to that grammatical gender, or even by switching the gender of existing words. Nonetheless, this is just an association, and some of the world’s languages have genders (also called ‘word classes’) that work differently. Others don’t reflect human categories in their grammar at all.

Many modern descendants of ancient Indo-European, like Spanish and Italian, have maintained the A-class of words up to the present day. But another daughter of Latin, French, has seen that A modified into E via sound change. This E is still a signifier of the feminine; it can be read, and less often heard, at the end of French nouns, adjectives, verbs and personal names. Some of the last group have made it into English, like Caroline and Diane.

Sound change has altered the original feminine Α elsewhere in Indo-European. Greek is a branch separate from Latin, and it has for centuries had alpha-final feminine nouns (e.g. thálassa ‘sea’, sophía ‘wisdom’) and names (e.g. Kleopátra), but it used to have many more. A large chunk of them in antiquity underwent a change from alpha to eta. This context-dependent shift took feminine nouns and names like níka (‘victory’) and Heléna, and turned them into níkē and Helénē. It was originally limited to the Attic and Ionic dialects of the ancient language, leaving us plenty of evidence for the feminine alpha elsewhere. But because it was from Attic, with Ionic input, that a unifying Greek language later emerged, these eta-final feminines have become a fixture of the modern standard.

English itself belongs to a branch of the Indo-European tree that changed and lost the feminine A centuries ago, long before our written record begins. Its branch is Germanic, and our earliest prose texts for this sub-family, composed in the 4th-century Gothic language, show that Germanic had shifted the feminine A into a long О̄ vowel. A few centuries later, when we get our first texts in Old English, we see that long О̄ has been reduced. It’s either absent or a short vowel, like the U of giefu (‘gift’) and granu (‘moustache’). Bearing this ending went hand in hand with belonging to the feminine gender, one of three categories that English grammar used to have. Both the categories and the endings would in time whither away, part of the loss of structural complexity that characterises the history of English words.

At this stage, I can offer a preliminary answer to the question of why a final A is feminine: because many of the words and names that exhibit it come from Indo-European languages, and these historically have employed an A-ending for words that are grammatically and humanly feminine.

This goes back to a single feminine ending in their common ancestor, Proto-Indo-European. Some members of the family have retained the A (Spanish, Italian, Slavic languages); others have modified it (Greek, French) or lost it (English, and modern Celtic languages too). Within the first group are languages, like Latin and Spanish, that have happened to gain cultural prestige, and consequently have donated A-final words and names to other languages.

But we word nerds want to dive to great depths. Why did A become this marker of femininity? How did it start? And why does this ancient A-class include a bulk of words that have nothing to do with women, femininity or humanity in general? Take a deep breath, and follow me deeper still.

Subscribe now

Shifting genders in prehistory

(Nerdery level: 4/5)

Strange to say, the feminine gender and its characteristic A are comparatively young. We have evidence for an early stage of Indo-European without them.

The assembled evidence has allowed scholars to identify two stages within the prehistoric development of Indo-European speech: a later stage with three grammatical genders (masculine, feminine, neuter), and an earlier stage with only two. These word groupings are generally labelled the animate and the inanimate.

Rather than any sub-types within humanity, speakers sorted between things that speakers recognised as self-moving, living, active individuals, and non-self-moving, non-alive passive objects and substances. A human, a horse or a bird would likely be assigned to the animate word class. Stone, water, earth, wind and fire would be grammatically inanimate.¹

The difference in animacy was expressed in the word’s ending, and was actually very minimal. Animate words would end in *-s if they were the do-er of the action (the nominative function). They bore the ending *-m if they were having the action done to them (the accusative function). We have a wealth of historical evidence for these prehistoric endings; Latin for example displays them in deus (a nominative god) and deum (an accusative god). Inanimate words, meanwhile, did not have this distinction. Their ending stayed the same regardless of whether they were nominative or accusative. This is a rule that has continued across the Indo-European family.

As a nerdy aside, this ancient grammar is reminiscent of ergative-absolutive languages (like Georgian and Basque). Also, the difference between the two classes was so slim that it’s possible to envisage a pre-prehistoric time when there were no classes or genders in Proto-Indo-European grammar at all. Such a situation is conceivable; it’s how Uralic languages (like Finnish) and Turkic languages (like Turkish) have functioned for centuries.

How we get from Stage 1 (animate, inanimate) to Stage 2 (masculine, feminine, neuter) is that the animate gives birth to a third grouping. Like Eve from Adam’s rib, the feminine gender began life as a subset of animate words. The feminine having established itself, the remainder by default became its counterpart masculine gender. The inanimate category, unbothered, henceforth gets rechristened by scholars as the neuter.

How can we believe any of this? Newcomers might be incredulous, faced with such discussion about stages within a prehistoric language that we have no direct evidence of. But there are many traces of the earlier stage still to be spotted in Indo-European languages with the three-gender system. Older daughters, like Latin, exhibit a fair few in their declensions. Even English words still whisper of it.

For example, English preserves the ancient division in its question words: who (animate) vs. what (inanimate). Who has the archaic form whom, the rule being that who is the asker and whom the asked. It therefore changes from nominative to accusative, as animate words always would. Meanwhile, what never changes in its ending, like a good inanimate. Nominative he and she inflect into him and her, but English’s inanimate pronoun, it, does not bend.

An overwhelming flood of evidence, however, comes from the Anatolian languages, most famous among them being Hittite. Following its decipherment in the 1910s, Hittite has been found not to show a single sign of a third gender, the feminine. Instead, its grammar functioned through a two-way divide of common (animate) and neuter (inanimate) words.

This difference is explained through complexifying the family tree. While Latin, Ancient Greek, Sanskrit, English and many others go back to that common Stage 2 point, Hittite and its Anatolian crew branched off from the rest after Stage 1. Put another way, after Anatolian-to-be had gone on its way, the remaining speakers of ‘Core Proto-Indo-European’² innovated the feminine gender that has endured until today. But why? Why make your own grammar more complicated through this innovation?

Engendering a gender

(Nerdery level: 5/5)

Those speakers’ reasons for the innovation are not at all clear. I cannot stress how much of a complex debate this aspect of language history has provoked among Indo-Europeanists. The prospect of doing justice to this headache-inducing issue daunts me, but I owe it to you, my reader, to try.

There is at least agreement over the point that the A typical of the feminine grammatical gender was once a suffix – that is, an extra thing added to the end of existing words to create new ones. It found itself added to a selection of animate nouns. The accusative *-m ending could follow it in turn, being necessary to indicate the role of the animate noun in the sentence and event.

The meaning of the suffix and what it contributed to the noun are in no way obvious. It can’t have been something exclusively female-referring; there are so many grammatically feminine Indo-European nouns that have no obvious connection to women, human gender or biological sex (remember: tables, huts, altars, etc.). All speculation is severely complicated by the fact that the same suffix seems to have created plural neuter nouns too.

Across Indo-European languages, A is not only associated with the feminine, but also used as the ending for nominative/accusative neuter plural nouns. In Latin, we have templa ‘temples’, tempora ‘times’, corpora ‘bodies’ and cornua ‘horns’. The neuter has since died out among Latin’s daughters. But over in Polish, a representative member of the Slavic branch, we find the plurals miasta ‘cities’, okna ‘windows’ and piwa ‘beers’. Likewise, in Modern Greek, the neuter plural Α remains strong; for example, singular neuter to déntro (’the tree’) is rendered plural as ta déntra (’the trees’).

We can consider the feminine A and the neuter plural A to be unrelated Indo-European phenomena, just a coincidence. But then we have two very similar-sounding suffixes to explain etymologically. Plus, there are whispers of a prehistoric time when the neuter plural didn’t indicate several separate things, but rather a collection or mass of things. The line between the two ways of sorting entities is very fine, I grant you. In this function, the category was treated as grammatically singular. The primary evidence comes from archaic quirks of Greek grammar, as well as subtle differences between words like kúkloi ‘wheels’ vs. kúkla ‘a set of wheels’ in Ancient Greek.

But on the opposite path, we also encounter problems. If we assume (at least for reasons of economy) that the same ancient Indo-European suffix is responsible for both feminine words like fēmina and neuter plurals like templa, it’s hard to pinpoint what its original function was.

I’ve submerged myself in literature on this topic, to little satisfaction. The usage of the suffix and the question of which came first (the neuter or the feminine) baffle the experts. Their proposals include an original ‘collective’ function (creating nouns that referred to collections of things, some of which happened to be women), an ‘appurtenance’ function (indicating that the suffixed nouns belongs or relates to something else), and an origin in adjectives. Painting a good image with the available evidence is so complex that H. Craig Melchert, a truly great Indo-European scholar, ends one paper on the subject with the pessimist note:

“I have no illusions that I have come close to “solving” the problem of the rise of the Indo-European feminine grammatical gender.”

(Melchert 2014: 269)

If you want to chase this wild goose yourself, you’ll need to know that the prehistoric suffix in question is referred to in the scholarship not as *-a, but as *-h₂. This is a sort of algebraic placeholder for an unclear consonant sound. It’s used for reasons that are very interesting, and really demand their own dedicated article. Until that happens, here’s Simon Roper on it.

What is semi-clear at least is there was once a suffix, added to a group of animate nouns in the ancient Proto-Indo-European language. This subset was the kernel of the A-feminine words. In time, the number of members referring to women and female entities that it contained (by design or by accident) led people to think of it as the ‘feminine’ gender. Its animate counterpart became the ‘masculine’. When this recognition occurred is unknown, but the ancient Greeks at least sorted their words into masculine, feminine and neuter. The third was the metaxú (‘between’) gender according to Aristotle. The Romans then followed suit, coining their terms masculīnus, fēminīnus and neuter (‘not-either’).

But it would take more than that prehistoric affix to turn the feminine into a fully fledged gender. What it needed was for the noun to share that affix with associated adjectives and functional words. It may seem counterintuitive at first, but what grammatical gender is first and foremost is a feature shared between words.

“Genders are classes of nouns reflected in the behavior of associated words”

(Hockett 1958: 231)

Gender is crucially something that a noun donates to other words, as a way of indicating that they belong together.

In Latin, stella ‘star’, turris ‘tower’, manus ‘hand’ and rēs ‘thing’ were all designated as feminine in gender, even though they don’t all exhibit the characteristic A. Nauta 'sailor', which does exhibit it, was grammatically masculine, an oddity probably due to its meaning. Grammatical gender is potentially so complex that it often detaches itself from the sounds and shape of its nouns. Some languages weave the shape, gender and meaning of words very tightly together; others are happy to pick the three threads apart.

Among them, the sharing of a quality between cooperating words is the essence of gender. Latin stella, turris and the rest were feminine not because of their appearance, but because they required feminine adjectives and determiners. Bonus ‘good’ and ille ‘that’ had to change into feminine bona and illa if they were to cooperate with stella or turris and provide information about the same thing.

In later Italian and Spanish, you can more or less know the gender of a word from its appearance; this was a development within non-standard Latin, arguably one for the better.

“The marking of declension class was becoming equatable to gender marking. This makes for an easier grammar than one in which gender must be learned separately from form.”

(Alkire & Rosen 2010: 195)

Revisiting my short Italian example above (“una donna alta”), if Italian nouns like donna didn’t share their gender with articles (una) or adjectives (alta), then all we’d have is a large group of Italian nouns that happen to end in A.

English has plenty of nouns that happen to end in T (cat, hat, rat), but we don’t think of them as forming a gender. If they shared that T around, then we might – that is, ‘thist bigt, beautifult and grumpyt cat’.

Luckily for the feminine gender, the suffix started to be shared between words in Proto-Indo-European phrases. Why? Because it was somewhat useful to do so. For language learners who think of grammatical gender as a nuisance and a challenge, it might seem strange to contemplate that it only survives by being useful or easy for native speakers, or both. Things that are unhelpful and hard tend not to survive in language. Grammatical gender can get started by providing a service for speakers.

In the case of Indo-European, it helped to thread words together across a complex sentence, a detail that confirmed to the listener which nouns belonged with which adjectives, verbs and determiners.

“It is a mistake to think of gender systems as systems for classifying things: to the extent that they do so it is secondary to their function to make it easier to keep track of links between constituents.”

(Dahl 2000: 113)

This was especially useful for a language that was flexible in its word order. I am well and truly qualified to say that Proto-Indo-European was such a language. Gender becomes less useful when word order is more rigid, as in the case of English. A weakening of gender might cause, or be caused by, a stiffening of syntax.

Clearer genders for English might allow its speakers more freedom to mess around with word order. We could say

I saw thist bigt cat outside the house yesterday morning beautifult and grumpyt

and we’d know that beautifult and grumpyt work with thist and bigt in referring to the cat, despite their separation across the sentence. Does that sound silly? The Romans and Greeks wouldn’t have thought so.

The limits of my language

(Nerdery level: 3/5)

Returning to the surface after this deep dive into prehistoric grammar and language change, I’m left with one final piece of reflection: that the nature and history of our languages shape our thinking and impressions. I refuse to get all ‘Sapir-Whorf’ on you here, mind you. What I mean instead is that our language gives us a sense of being natural and normal, when it is in fact arbitrary. For people thinking within the Indo-European box, there is something essentially feminine about A. New names for baby girls or products marketed at women may well be coined with A in final position. Activia yoghurts and perfumes from Sephora come to mind.

But it’s all an accident of language history. There is nothing inherently feminine about the letter A, nor the vowel that it spells. Its status today derives from a prehistoric suffix in a language whose offspring happen to have been socially successful. There are other traditions of linguistic femininity that put no such store in A. In Semitic languages, to give one example, the feminine marker has historically been T. Native English speakers may have encountered this ending in a Jewish woman’s bat mitzvah or the name Judith (deriving from Hebrew, etymologically ‘Judean woman’).

It is us who put meaning, associations, notions and nuances into our language; we don’t discover them in it. Contemplating the feminine A and its fundamental arbitrariness are for me a path towards bumping into the bars of our caged thinking. They’re also a key for unlocking the door and stepping outside it for a while.

END.

Subscribe now

Endnotes

1. That said, we actually reconstruct two words for ‘fire’ in Proto-Indo-European. One was animate (the origin of Latin ignis), the other inanimate (the origin of English fire). It seems that speakers perceived both fire as an active force and fire as a tamed substance.

2. Note for my fellow nerds: I don’t know where to locate Tocharian in this story of gender development. It’s complicated.

Bonus fun fact

The word gender, liberally used in this article, comes from Latin genus, which just meant ‘type’. Genus is also the origin of general and generic. What marks gender out from the other offspring is its D. This isn’t etymological; it’s epenthetic. The consonant was inserted into the word by post-classical speakers after genus, specifically the oblique form genere, had become gen’re. This contained a phonetically uncomfortable sequence of tongue-tipped sounds that the extra consonant could assist with. If you’d like to read more about epenthesis, I have an article all about it to hand…

References

• Alkire, T., & Rosen, C. (2010). Romance languages: A Historical Introduction. Cambridge: Cambridge University Press.

• Dahl, Ö. (2000). Animacy and the notion of semantic gender. In Unterbeck, B. et al. (Eds.) Gender in Grammar and Cognition I: Approaches to Gender. II: Manifestations of Gender. De Gruyter. 99–115.

• Hockett, C. F. (1958). A Course in Modern Linguistics. New York: Macmillan.

• Hoffner Jr, H. A., & Melchert, H. C. (2008). A Grammar of the Hittite Language: Part 1: Reference Grammar. Winona Lake, Indiana: Eisenbrauns.

• Kim, R. I. (2009). The feminine gender in Tocharian and Indo-European. In Yoshida, K., & Vine, B. (Eds.) East and West: Papers in Indo-European Studies. Bremen: Hempen. 69–87.

• Kim, R. I. (2014). A Tale of Two Suffixes: *‑h₂‑, *‑ih₂‑, and the Evolution of Feminine Gender in Indo-European. In Neri, S., & Schuhmann, R. (Eds.) Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden/Boston: Brill. 115–136.

• Ledo-Lemos, F. J. (2003). Femininum Genus: A Study of the Origins of the Indo-European Feminine Grammatical Gender. München/Newcastle: LINCOM Europa.

• Luraghi, S. (2009). The origin of the feminine gender in PIE. Grammatical Change in Indo-European Languages. Amsterdam, Philadelphia.

• Luraghi, S. (2011). The origin of the Proto-Indo-European gender system: Typological considerations. Folia linguistica 45(2). 435–464.

• Melchert, H. C. (2014). PIE*-eh₂ as an ‘Individualizing’ Suffix. In Neri, S., & Schuhmann, R. (Eds.) Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden/Boston: Brill. 257–272.

• Ringe, D. (2017). From Proto-Indo-European to Proto-Germanic. Vol. 1. Oxford: Oxford University Press.

Original Substack article. Images my own, from Wikimedia, or from credited sources. Cover image: woodcut print from 16th-century Germany, in the care of the British Museum.

“... Love is not love

Which alters when it alteration finds,

Or bends with the remover to remove.

O no! it is an ever-fixed mark

That looks on tempests and is never shaken.”

(William Shakespeare, Sonnet 116)

February 14th has come round again, and love is in the air!

‘It was different in my day,’ older souls may grumble, thinking fondly back to when relationships were courtships and dating was wooing. Before that day, we imagine dark days of marriage by command and affection as an optional extra. After that day, we now find a romantically frustrated youth, often ghosted and getting by on hook-ups.

There’s an argument to be made that love is fickle, a passing fad that has had its moment in the cultural sun. ‘Love is dead, long live lust!’ Now dawns a day of transactional pragmatism and selfish desire. Love’s blossom blown away, the cynic will assign the word to the simulacra of red hearts, roses and soppy sayings required to sell more Valentine’s cards.

I disagree; I have more faith in love. Would you believe that that faith comes in part from etymology?

If you know me, and how I call the romantic realm of historical linguistics home, perhaps you would. I’ll let poets and historians battle it out over the exact nuances of old references, but I can at least say that there is constancy and a long history behind this word love of ours. When we investigate our words and their journey across time and space, we see how we are connected to linguistic ancestors (not necessarily biological ones). They may have lived thousands of years ago, but they apparently loved as we still do. If love can survive for so long, what can our times do to dislodge it from our hearts?

Subscribe now

Seeking the word love in our records for the English language turns up a cornucopia of occurrences. It appears in surviving texts from the period of Old English, the language prior to the arrival of a linguistically significant army of Norman warriors in 1066 AD. The concept is recognisably our own, high in its emotional content and broad in its application.

Love is there, for example, in the ancient poem Beowulf, included among its monsters, manly heroes and mead halls. The poem speaks of good Queen Hygd, who “lufode ða leode” (‘loved the people’) she was dispensing drinks to in her hall. They surely loved her back. The title character Beowulf tells King Hygelac man to man that he would always “þinre modlufan maran tilian” (‘strive for more of your mind-love’). These were days of yore before the adoption of certain signs of modernity, namely the letter V. English writers back then made do with F.

Though its form has changed in sound and spelling over the centuries, the word love has clearly remained a core item of English vocabulary. It’s been accompanied by daughter words too, derived from it, like lovely and beloved, all of them full of warmth and affection. They may be undergoing a kind of weakening, as love muscles in on the territory of like. A kind of semantic inflation has meant that I love pizza has become a reasonable expression. This use of the word would’ve confused Shakespeare, along with the pizza. Nonetheless, love retains its rank as the supreme affection.

Love has sister words too, born from the same root, but some of these have fallen into obsolescence. Lief is an adjective meaning ‘dear’, which was commonly employed in Old English (Beowulf himself is twice affectionately called “Beowulf leofa”), but which has since retreated into archaic language. That said, lief limps on as the first part of the unusual word livelong – as in the livelong day. Originally, it was literally ‘the dear long day’, with lief brought in to add emphasis to the length. Throughout its employment, those who were lief to you were valued, treasured.

If we adjust our scope to consider geography as well as chronology, similar-looking instances of love appear in other languages. Over the North Sea from England lie the Netherlands and Germany, where people speak Dutch and German. An Amsterdamer would know his or her strong feelings for a compatriot as liefde in Dutch. Over the border in Germany, the same high attitude is known as Liebe, and a dear thing is lieb.

But how to explain these similarities? The linguistic archaeologist has three options. One is to put them down to simple coincidence. Another is to ascribe this to influence, as one language donates words to another. These are both unlikely scenarios. The words bear too many similarities and too many speakers in one corner of the world for this to be just luck. As for influence, love is such a fundamental feeling that it would be brave to argue that the English gave the word and its concept to the Germans and the Dutch, or one of those two to the English. Instead, the third option stands out as most plausible: that love, liefde and Liebe share a common ancestry, and are products of a time when English, German and Dutch were one language.

The three emerged from their common gradually, forming a linguistic family of relatives. This ‘Germanic’ family isn’t limited to the aforementioned three. When an Icelandic speaker calls a dear thing ljúfur and a Swede describes something lovely as ljuvlig, we have evidence to group their languages within Germanic too. These five and others trace descent from a common Germanic forefather, spoken in northern Europe from around the year 500 BC. But this language wasn’t written down in texts that we now can read. What we’ve done here is step off the edge of history, guided by language into prehistoric darkness.

Within that original language (Proto-Germanic), the ancestor of our love was a potent concept, just as it remains for us today. Within its purview, it contained not only romantic and sexual attraction between two people, but also tenderness, affection, desire, respect and goodwill. When Germanic speakers later encountered Christianity (which puts a high premium on love), it was with love and like words that its central, transcendental concepts were translated. Speakers and writers of Gothic, a deceased aunt of English, coined the compound term brōþralubōn (‘brother-love’) in the 4th century AD. This was a literal translation of the Greek word philadelphía, found in St Paul’s letter to the Romans, but now better known today as home to the Eagles and Phillies. No doubt the coming of Christ brought and fostered new perspectives on love, but it wasn’t an alien concept, unmatched by existing Germanic vocabulary.

We can still travel further back. Around the same time that Germanic speakers were living in northern Europe, the Romans were putting themselves about the Mediterranean. In their Latin language, we can spot a similar-looking word likewise used to express a positive disposition. This was lubet, meaning ‘it pleases’. It was how a republican-era Roman would translate our modern English word like. Plautus, a playwright penning his words around the year 200 BC, has the character Philematium flirtatiously quip “quod tibi lubet, idem mi lubet” (‘what you like, I like too’).

Lubet was an everyday verb for the Romans, and it shared its lub- core with lubido, a noun meaning ‘desire’ or ‘pleasure’. In the later classical language of Caesar and Cicero, lubido had shifted in one sound to become libido. This noun might have remained an antique, had it not been resurrected in the late 19th century, when leaders in the newfound field of psychoanalysis, chiefly Sigmund Freud, summoned libido to serve in their psycho-sexual analyses.

Bear that lub- root in mind, because we find it elsewhere in Europe. To declare fondness for something or someone in Polish, for instance, you’ll need the verb lubić. Bring out a phrase like lubię Polskę (‘I like Poland’) in Warsaw, and you’re sure to make friends. Polish is one of many Slavic languages found across Europe, and their familial connection is substantiated by very similar words. Just as Polish has lubić, Russian has ljubítʹ, Serbo-Croatian ljubiti, Slovak ľúbiť and Czech líbit – all to do with loving, liking, kissing and adjacent activities. Their co-existence points to a source word, which specifically meant ‘love’, in a prehistoric Slavic parent.

But again, how to explain these similarities? Latin and the Slavic languages are too distant to be included within the Germanic club, both linguistically and geographically. Latin emerged in Italy around the river Tiber, while the likely homeland of Slavic speakers lay in eastern Europe. Nonetheless, Latin and the two families share a significant amount of similar vocabulary, among L-words related to loving, pleasing and desiring among them.

Our scope has to expand even further, because words resembling love, lubet and lubić in appearance and meaning pop up in India too! Great compositions in Sanskrit, an ancient and prestigious language of India, include words built around the cores lubh- and lobh-. We find verbs like lúbhyati, meaning ‘to desire’. The negative consequences of these positive emotions are reflected too in such Sanskrit words; the Mahābhārata, the world’s longest epic poem, often warns against lobha (translatable as ‘greed’, ‘covetousness’ or ‘lust’). This was a secondary development, though; love and desire were primary.

One of these languages – from among Latin, Sanskrit, the Germanic group and the Slavic set – cannot subsume the others. Instead, they represent visible branches of a dense family tree, the common trunk of which is lost to time. Just as the immediate ancestor of the Germanic languages is a prehistoric hypothetical, so too is its own grandparent. But while Germanic alone took us only back to 500 BC, we now have the means to travel back perhaps as far as 4000 BC. That grandparent, located even further back in prehistory, would beget not only Latin, Sanskrit and our dear English, but a family that, before the modern era, was already strewn across Eurasia, from eastern India to western Europe. Hence, we know it as Indo-European.

Estimates vary as to when the single proto-ancestor of the Indo-European group was spoken, but six thousand years before the present day is a common suggestion. I appreciate that, to newcomers, this idea may feel like a flimsy construction, built on nothing but language. Yet the idea is a couple of centuries old and has stood the test of time. Countless words and bits of grammar have together built a steady platform for us to reach so far back. One of those words is love.

Six thousand years ago, give or take a century, somewhere out on the Eurasian Steppe (a large expanse of grassland), there were people who lived, died, ate, drank, hunted, farmed, fought, built and, more importantly, loved. Their words to do with loving would have built around a linguistic core. This, on the basis of what it became, was something like *leubʰ-. (That asterisk indicates that we cannot be sure, but this hypothetical form is what the evidence points us to.)

These people haven’t left us today much in terms of artefacts, but they did pass on their vocabulary. Handed down over millennia, their *leubʰ- words would eventually produce our word love, as well as our libido via the Romans. The collective linguistic evidence points to *leubʰ- giving expression to very high feelings; its descendant words are united in desire and delight.

The people out on the Steppe loved and spoke about their love, and now so do we. Their words and their concepts have endured countless societal shifts since, and it’s this etymology that gives me a degree of hope for the survival of love in the immediate future. We today, when we use the word love, are connected to ancient human affections, desires and passions via the very words we use. I personally think that that, for no want of a better word, is lovely.

END.

Subscribe now

References

• Digitales Wörterbuch der deutschen Sprache

• Kroonen, G. (2013). Etymological Dictionary of Proto-Germanic (Leiden Indo-European Etymological Dictionary Series). Leiden, Boston: Brill.

• oed.com

Sources

• Beowulf

• Plautus’ Mostellaria

Original Substack article. Cover image, from Wikimedia, of a 13th-century manuscript illustration from the Roman de la poire, in which a suitor offers a somewhat anatomically heart to his lady.

That post and this one are united by an appreciation of how interconnected the ancient world was. True, travel could be dangerous and laborious (a fact perhaps reflected in the etymology of the word travel itself, a sister of French travail ‘work’). It certainly did not progress at the consistent speed of a modern car, train or plane.

Nonetheless, people did move around, and they took their languages along for the ride. One result of their peregrinations is that languages can pop up in the historical record in surprising locations.

To make it into that record is by no means guaranteed. What few texts and witnesses to ancient language have come down to us must surely reflect only a fraction of the speech of their day. Political prestige is a factor in their endurance; if a society’s elite patronised a language, it stood a good chance of being committed to the medium of writing.

While Greek, the Asian and African Language was partly concerned with Greek’s success to the east, the star of this Part 2 is a language that met with social success far to the west of its associated region (India). It makes a brief appearance in ancient sources that continues to intrigue historians and linguists alike. Well, at least one of them.

First, though, here’s a lot of necessary (but no less interesting) linguistico-historical background.

Subscribe now

Among the Aryans – no, not those ones

A sizable chunk of my adult life has been spent in the company of one particular group of languages, a family of sisters and cousins that share a prehistoric grandmother, and now encircle the world. This is the Indo-European family of languages. It gets its name from the distribution of its members, scattered across a broad sweep of Eurasia, prior to further enlargement by European powers in the early modern period.

It was during that same period that the recognition of the languages’ relationship began to occur, thereby connecting Gujarati to Greek, Sinhala to Swedish, Persian to Portuguese. In particular from the late 18th century onwards, the Indo-European family established itself as an object and field of study, and it has since blossomed into a rich garden of language and history.

Among the branches of the Indo-European family tree that scholars have perceived, there are subordinate Indo-somethings. While Indo-European refers to the whole set, Indo-Iranian is a subset of those languages. This is one of the primary branches of the tree as we understand it; it’s comparable in that status with the Germanic branch to which English belongs.

Within Indo-Iranian, Iranian and Indo-Aryan are then two sub-subsets. Each subsequent set has its own ancestral nexus point, and branches out from it like roots. Each emerged through the natural process of linguistic splitting, through which accents, dialects and full-blown languages are born.

But make no mistake; Indo-Iranian may be one step subordinate to Indo-European, but the term covers a cosmos of languages. From Kurdish in the west to Assamese in the east, Indo-Iranian is a major grouping in its own right.

As said, the central divide within Indo-Iranian is into Iranian and Indo-Aryan languages. Yes, dear language nerds, there are also the Nuristani languages, but I will set them and their place in the story to one side.

The first of the two major sub-subsets contains Persian, Pashto, Balochi and beyond. The second contains the linguistic cornucopia spoken and written across Pakistan, north and central India, Bangladesh, and up into Nepal; it includes major languages like Hindi-Urdu, Bengali, Marathi and Punjabi, each with speakers numbering in the millions.

Now, some readers may raise an eyebrow or furrow a brow at the use of the term Aryan with such abandon. For sure, the term was weaponised to evil ends in 20th-century Europe. Yet its use in that century and continent make up only one chapter in its long employment. It has its origins far away from Europe, in the northern reaches of prehistoric Central Asia.

Its journey from there to the posters and pages of Nazi propaganda was made possible through a combination of academic scholarship and twisted logic. European ‘experts’ of the 19th century seized on the term as a signifier of an ancestral superior people, and concurrently a racial category that could exclude as much as include. They hijacked and perverted an ethnic term that still has currency in Asia today.

The word Aryan is in fact cognate with Iranian. The two go back to a single identity self-applied in the Central Eurasian Steppe. Its etymology is much debated, but the reconstructed etymon *Áryas seems to have referred to a member of a people who lived, died and rode chariots on either side of the year 2000 BC.

“The original homeland of the Aryans, the speakers of Common Indo-Iranian, cannot be precisely identified, but is thought to have been in western Central Asia, to the east and north-east of the Caspian Sea.”

(Sims-William 1998: 127)

From the linguistic and archaeological artefacts they left behind, we can sketch the vague outline of the actual Aryans. They shared an Indo-European language, a semi-nomadic pastoralist lifestyle, and a sophisticated understanding of the supernatural. Their gods constituted a complex pantheon, among them still famous names like Mitra and Indra. Centuries later, Mitra would find favour across the Roman Empire, venerated through the mysteries of Mithraism.

The common language of the Aryans (‘Proto-Indo-Iranian’) would go on to split, split and split again. Over the course of the second millennium BC, migration in two general southwest and southeast directions would gradually motivate the ancient divide into Iranian and Indo-Aryan. Iranian languages would end up in Iran, unsurprisingly, but also all along the vast east-west Steppe; the Scythians and Sarmatians of the classical era seem to have spoken Iranian tongues. Meanwhile, Indo-Aryan would end up in India. On the journey and at their destinations, each would continue to burgeon into today’s dense canopy of languages.

That linguistic disintegration would’ve gone hand in hand with the dilution of the label of *Áryas and its importance. That said, it has retained relevance; in as late as 1935, Reza Shah Pahlavi asked the rest of the world not to call his country Persia, but rather the local name Iran (etymologically ‘of the Aryans’). The rest of the world listened.

The hyphenated term Indo-Aryan is intended to acknowledge the linguistic heritage that the languages under its wing share with Iranian languages, but not with all the languages of India. That country and its neighbours are home to other big language families, like Dravidian. With the exceptions of Sinhala and Dhivehi, Indo-Aryan languages generally find themselves to the north of Dravidian ones within South Asia. This modern-day spread points to their prehistory away to the north, in the Eurasian Steppe.

Supporters of the term Indo-Aryan would argue that its ‘Indian-but-not-all-India’ quality is an essential aspect that the alternative label Indic struggles to convey and distinguish.

When you look at the early samples of language from either side of the Iranian/Indo-Aryan division, their inherited linguistic and cultural affinity is clear. Representing the Iranian side, we have Avestan, the language of the corpus of Zoroastrian literature known as the Avesta. For Indo-Aryan, we have Vedic (also called Sanskrit and Vedic Sanskrit, but I lean towards plain Vedic). This was the language behind great religio-literary compositions like the Rigveda, rich in hymnal praise of the gods.

By the second half of the second millennium BC, Iranian and Indo-Aryan speech had clearly developed into two distinct daughters of Proto-Indo-Iranian. However, from what our surviving sources witness, the two traditions of Zoroastrian and Vedic verse nonetheless still shared so much in terms of sounds, words, grammar and gods.

On the language side, we can find words that are more or less still the same. Numbers, for instance, remained fairly united; five was paṇca for Avestan speaker-writers, and páñca- in Vedic. Yet deeper disagreements had already emerged too; the number one is a lexical faultline: aēuua- in Avestan, but *aika- for the Indo-Aryan speakers composing the Vedas.

In between concord and discord, we also find many correspondences – that is, consistent differences, each with a common origin. For example, Avestan shows a /h/ sound where Vedic has a /s/. An Avestan speaker would count seven things with the number hapta, a Vedic speaker with saptá-. The former would include themselves with ahmī ‘I am’, the latter with ásmi. In this aspect of phonology, the Iranian /h/ is the innovative sound, the Indo-Aryan /s/ the conservative.

For another example of this, the Rigveda often mentions and praises soma, a powerful ritual drink of unclear botanical origin. A whole mandala (‘book’, for want of a better word) of the Rigveda is devoted to it.

ápa soma mŕ̥dho jahi / áthā no vásyasas kr̥dhi

‘Drive away enemies, soma, then make us better’

(Rigveda 9.4.3)

At or around the time, the composers of the Avesta were likewise fairly fond of the sacred plant haoma.

āat̰. aoxta. zaraθuštrō. / nəmō. haomāi. / vaŋhuš. haomō. / huδātō. haomō.

‘Then Zarathustra said: reverence to haoma, haoma is good, haoma is well formed’

(Hōm-Yašt 16)

These ancient characters and substances remained so potent that they inspired European writers of the 19th and 20th centuries; Nietzsche adopted the central figure of Zoroastrianism for his 1880s book Thus Spoke Zarathustra, while soma gave its name to the infamous drug in Aldous Huxley’s 1932 novel Brave New World.

Among the gods, a curious kind of divine inversion occurred that upset the original Aryan religious cosmology. Different types of deity, who presumably once all lived in relative harmony, could either ascend further or be cast down.

In the Zoroastrian hymns and their Avestan language, a daēuua- is a demon and an adversary. Meanwhile, ahura- is an honorific meaning ‘lord’, as bestowed on the Zoroastrian supreme being, Ahura Mazda. Contrarily, in Vedic religion, the cognate devas come to be the good guys, and asuras the ones to avoid. For example, Indra, king of the devas, is a deity supremely exalted in the Rigveda.

índraṃ vayám mahādhané / índram árbhe havāmahe

‘Indra we call on in the great battle, Indra we call on in the small battle’

(Rigveda 1.7.5)

Meanwhile, the Indra of the Avesta is a second-rank demon, and barely mentioned.

paiti. pərəne. iṇdrəm.

‘I charge against Indra’

(Vidēvdād 10.9)

Society and divinity were splitting along the same faultline as language, and the shifting status of these divine species is actually detectable in our sources. Composed over centuries before its final consolidation, the Rigveda is built up of discernible layers of belief. Textual archaeology can excavate the rise of the devas and the fall of the asuras from within its thousands of verses. This disagreement has been explained as the product of reform, with one if not two movements seeking to change the original Aryan religion from within.

Out of all this language, prehistory and religion, what I’d have you bear in mind is that Indo-Iranian is a big group of historical and current languages, divisible into Iranian and Indo-Aryan sets. All of them were once one language; their common ancestor, which we call ‘Proto-Indo-Iranian’, was itself an early member of the Indo-European family tree. It broke up through the migration of its speakers.

Speakers of Iranian-to-be walked or rode south and west, both along the Steppe and towards what is today Iran. Speakers of future Indo-Aryan headed south and east, descending from the mountains into South Asia. This gives us the general distribution of the two today, with the former to the west of the latter. Indo-Aryan languages are understandably closely associated with India, hence the name.

This linguistico-historical background makes it all the stranger that our earliest evidence for the Indo-Iranian family (specifically from its Indo-Aryan branch) actually comes from the Middle East.

Subscribe now

Horse riders can go far

^D.MESMitrassil ^D.MESArunassil ^DIndara ^D.MESNasattiyanna

‘... both Mitra and Varuṇa, Indra and the Nasatya …’

(KBo I 3 Vo 24)

A treaty in c. 1400 BC invoked these gods as its guarantors. They were top gods in the ancient Aryan pantheon, and have been venerated in India for millennia now. Mitra and Indra we have already met, and they all appear together, in the same order, in the Rigveda:

ahám mitrā́váruṇobhā́ bibharmi / ahám indrāgnī́ ahám aśvínobhā́

‘I bear Mitra and Varuṇa both, Indra and Agni, and the Aṣvins (a.k.a. Nasatya) both’

(Rigveda 10.125.1)

But this treaty wasn’t unearthed in India; it was found in what is now Turkey.

The treaty committed to clay a new-found harmony between the Hittites and Mitanni. While the Hittites were found in the 1910s to have spoken an Indo-European language (like a niece of Proto-Indo-Iranian), most people in Mitanni spoke Hurrian. This was a non-Indo-European language of northern Mesopotamia, written in the widespread cuneiform system.

Mitanni was established as a kingdom around the year 1500 BC, covering a large territory today torn between the countries of Syria, Turkey and Iraq. While no doubt a linguistically dynamic people – as humans historically have tended to be – the majority in Mitanni seem to have been native Hurrian speakers. For that reason, we might expect the kingdom’s elite to have likewise been Hurrian in speech, writing and outlook, give or take some external influences from local superpowers like Egypt.

Instead, what we find is a heavy linguistic influence from Indo-Aryan. This is in fact our earliest evidence for the Indo-Aryan family, and the bigger Indo-Iranian branch to which it belongs. Far away from India and the Central Eurasian Steppe, Mitanni texts bear witness to a cultural layer imposed on the general Hurrian population.

Kings and gods form part of this ‘superstratum’; in addition to invoking gods like Indra, the kings of Mitanni style themselves with regnal names that are discernably Indo-Aryan in their etymology. These are names refracted through the cuneiform tradition of writing, which can make recognition a little uncertain, but the name recorded for king Tushratta, for one example, has been derived from the Indo-Aryan word tveṣárathaḥ. This means ‘quick-chariot-ed’, and is again a word that appears in the Rigveda (5.61.13). Tushratta was an ally of the Egyptian New Kingdom and father-in-law to the radical pharaoh Akhenaten. His messages were found among the Amarna corpus of official letters from Upper Egypt, which was even further for an Indo-Aryan word to travel.

Heck, it’s been proposed that the name Mitanni itself has an Indo-Aryan origin. But beyond kings, states and gods, everyday words from Indo-Aryan also appear. The kingdom relied on a class of chariot-driving young warriors, known as maryannu. This term, documented in sources from both Mesopotamia and the east coast of the Mediterranean, has been derived from Indo-Aryan márya-. This means ‘young man’ in the Vedas, but a shift in meaning to ‘warrior’ is not implausible. We can imagine young men making a name for themselves through their skillful operation of the ancient world’s equivalent of a tank.

Chariots and horses are a common theme in the fragments of Indo-Aryan language of Mitanni. Kikkuli, a Hurrian individual whose name we actually know, left us a delightful document that shares his equine expertise. Beginning “thus speaks Kikkuli, master horse-trainer of the land of Mitanni”, Kikkuli’s guide discusses not only how to get horses to follow your orders, but also how to care for their needs.

It was written in c. 1345 BC in the Hittite language, yet contains several words of Indo-Aryan origin. Apparently these came with the whole tradition of horsemanship, and Kikkuli was unable to translate them properly with single Hittite words.

In his text, we find the numbers aika- ‘one’, tera- ‘three’, panza- ‘five’, satta- ‘seven’, na[ua]- ‘nine’, each in combination with a word for ‘turn’ or ‘circuit’, -u̯arttana. Compare these numbers with their equivalents in Classical Sanskrit: éka-, trí-, páñca-, saptá- and náva-. The job title that Kikkuli introduces himself with (“aššuššanni”) may also contain the Indo-Aryan word for ‘horse’ (in Sanskrit: áśva-). We can assume that Kikkuli was rewarded for his service to the Hittites, although I do wonder if teaching them Mitanni horsemanship constituted a kind of treason.

Since Indo-Aryan and Hittite were both ultimately Indo-European languages (albeit greatly separated by time), Kikkuli’s manual displays a rare meeting of two ancient members of the family, like two branches of the same tree crossing over in the forest.

What we have here is a prestigious elite language of the Mitanni kingdom, whose speakers, at some point, must have travelled a considerable distance to bring it close enough to Hurrian lands. It also seems to have come with a culture attached, or at least a tradition of prowess in horsemanship and exceptional hippological knowledge. We’ll get to what social circumstances might have made Mitanni Indo-Aryan possible in a moment, but first, a word on that distance.

If we accept that these interloping words in our Mitanni texts are indeed Indo-European, and specifically Indo-Iranian in origin, why not derive them from an early Iranian language? Why Indo-Aryan, when Iranian speakers ultimately travelled and settled westwards? In proposing contact with the Indo-Aryan branch, we have a greater distance to account for. Yet that’s what the linguistic evidence points to.

As mentioned above, differences emerged between Iranian and Indo-Aryan, and what words we have from Mitanni identify themselves as from the latter. The number one, for example, is aika-. Remember that this is a hard line of disagreement. Iranian languages have a primary numeral with an internal /w/ sound – for example, Avestan aēuua-. It’s Indo-Aryan that is team K. Moreover, satta- for ‘seven’ preserves a /s/ sibilant, rather than an Iranian /h/.

Beyond the sounds of the words from Mitanni, their reference is noteworthy too. The appearance of Varuṇa in the invocation above aligns it with the Vedic religion and their deities; Varuṇa is prominent in the Rigveda, but absent from the Avesta.

Yes, the divergence of language and people is complicated, but the accumulated evidence indicates an early kind of Indo-Aryan as the donor of this vocabulary, even though Indo-Aryan would end up in the east. Strange to say, it overtook Iranian, appearing in the historical record to both its sister’s east and its west.

In terms of linguistic history, we can at least say from the evidence that Mitanni’s contact with that language happened after the break-up of Indo-Iranian. But it is very hard to say where its speakers had travelled from to come close to Mitanni. The Indo-Aryan language might be closely associated with India, but that does not entail that India (even maximally defined) was the geographical starting point for this linguistic contact. Some scholars have stated that “the Indo-Aryans of the Mittanni most probably were never in India” (Lubotsky and Kloekhorst 2021: 331). Instead, they might have set off from somewhere to the northwest, somewhere closer to the Hindu Kush, like Bactria.

Nonetheless, wherever they started, it was still rather a long way from there to the banks of the river Tigris. Who then were these people, whose language and culture met with such success in Mitanni?

Of course, there is more than one plausible explanation for this meeting of languages in antiquity. One, a ‘maximal’ view, is that Mitanni was “dominated by an Aryan aristocracy” (Lazzeroni 1998: 98). In other words, Aryans arrived in person from the east, and through conquest, diplomacy, alliance or matrimony ended up ruling the Hurrian-majority kingdom. I note that the Britannica article goes as far as to say that Aryans founded the whole kingdom of Mitanni.

Archaeology and genetics demonstrate that they would’ve come too late to introduce the domestication of the horse to the Middle East, but they may have had outstanding equine skills that gave them a military and political edge. The chariot may have been their specific field of expertise.

This scenario, I think, accords with the linguistic evidence pretty well. A language needs to be backed up by some serious social currency to become the medium for naming kings and invoking the gods. Historical evidence too leans towards an active inequality in language and people in Mitanni, in how those gods were addressed:

“The order in the invocations of deities is telling: Akkadian first, Sumerian second, Mitanni Aryan next, Hurrian last. This implicit textual hierarchy actually supports the idea that Mitanni Aryan was in a position of socio-cultural superior prestige when compared to Hurrian in the eyes of Hurrian people themselves, who made up the majority of the inhabitants of the Mitanni Kingdom.”

(Fournet 2010: 5)

Yet there are ‘minimal’ explanations, which afford the Aryans and their Indo-Aryan language only a small or indirect role in the affairs of Mitanni. Under this view, the exchange of words and ideas was lighter, more partial and probably a lot less turbulent.

It may have happened in Mitanni, perhaps with a small contingent of advisory Aryans serving as experts in new technology and new deities. Alternatively, the point of contact could have occurred elsewhere, or via a chain of transmitted knowledge that began somewhere between Mitanni and India.

Instead of a living culture, Indo-Aryan technical terms such as we find in Kikkuli’s manual may have been “piously handed down as fossils” (Kammenhuber 1988: 788), acquired by the Hurrians and then the Hittites as opaque jargon. It’s true that these words are few in number and are fully plugged into local Hurrian grammar. They therefore may not “reflect the presence of speakers of Indo-Aryan, rather they are relics of interaction with Indo-Aryan speakers” (von Dassow 2008: 85).

For myself, if I can humbly offer up an opinion, I bend towards the maximal view. For a people to provide the name of an ancient kingdom, and to furnish it with gods, royal names and an elite class of warriors, I cannot see how the Aryans could not have been a powerful personal presence among the Hurrians. They may have been linguistically assimilated by the time of our evidence for their Indo-Aryan language in Mitanni, but the nature of that evidence speaks to me of at least some direct contact between Aryans and Hurrians.

If I’m right in that, then we have here a prime instance of the interwoven fabric of antiquity. Set aside a stale image of stagnant populations stubbornly sticking to their lands and settlements; the ancient world can surprise us with its dynamism. Societies were interacting and overlapping in ways that required considerable means and determination. Now, here in the present day, I’m keen to fly the flag for language as a way to unearth that impressive past.

END.

Note:

Words prefixed with an *asterisk are hypothetical words, not directly witnessed by historical sources, but still very likely to have existed. Words suffixed with a hyphen- designate incomplete stems, without the grammatical endings necessary for their use in a sentence.

Non-linked references:

• Fournet, A. (2010). About the Mitanni Aryan Gods. Journal of Indo-European Studies 38(12). 26–40.

• von Dassow, E.,(2008). State and society in the late Bronze Age: Alalaḫ under the Mittani Empire. In Owen, D. I., & Wilhelm, G. (eds.), Studies on the Civilization and Culture of Nuzi and the Hurrians, vol. 17. Bethesda, MD: CDL Press.

• Hrozný, B. (1931). L’entraînement des chevaux chez les anciens Indo-Européens d’après un texte Mitannien-Hittite provenant du 14e siècle av. J.-C. Archiv Orientální 3(3). 431–463.

• Kammenhuber, A. (1988) On Hittites, Mitanni-Hurrians, Indo-Aryans and Horse-Tablets in the 2nd Millennium B.C. In HRH Mikasa, T. (ed.), Essays on Anatolian Studies in the Second Millennium B.C. 35–51. Wiesbaden: Otto Harrassowitz Verlag.

• Lazzeroni, R. (1998). Sanskrit. In Ramat, A. G., & Ramat, P. (eds.), The Indo-European Languages. 98–124. Routledge.

• Lubotsky, A. (2023). Indo-European and Indo-Iranian Wagon Terminology and the Date of the Indo-Iranian Split. In K. Kristiansen, G. Kroonen, & E. Willerslev (eds.), The Indo-European Puzzle Revisited: Integrating Archaeology, Genetics, and Linguistics. 257–262. Cambridge: Cambridge University Press.

• Lubotsky, A., & Kloekhorst, A. (2021). Indo-Aryan -(a)u̯artanna in the Kikkuli treatise. In H. Fellner, M. Malzahn, & M. Peyrot (eds.), Lyuke wmer ra: Indo-European Studies in Honor of Georges-Jean Pinault. 331–336. Ann Arbor NY: Beech Stave Press.

• Martínez García, F. J., Vaan, M. D., & Sandell, R. (2014). Introduction to Avestan. Leiden: Brill.

• Parpola, A. (2015). Early Indo-Iranians on the Eurasian Steppes. The Roots of Hinduism: The Early Aryans and The Indus Civilization. Oxford University Press.

• Raulwing, P. (2009). The Kikkuli Text. Hittite Training Instructions for Chariot Horses in the Second Half of the 2nd Millennium B.C. and Their Interdisciplinary Context. From here.

• Sadovski, V. (2009). Ritual Formulae and Ritual Pragmatics in Veda and Avesta. Die Sprache 48(1). 156–166.

• Sims-Williams, N. (1998). The Iranian Languages. In Ramat, A. G., & Ramat, P. (eds.), The Indo-European Languages. 125–153. Routledge.

Sources:

• Hōm-Yašt text

• Rigveda text

• Vidēvdād text

Original Substack article. Images my own, from Wikimedia, or from credited sources. Cover image: a decorated bowl from Iran, dated to 2000-1800 BC, on display at the Louvre.

A highlight of the journey must be the visit to Reichenau, a precious stone set in Lake Constance, today a southern part of Germany. Reichenau is known the world over for its ancient abbey and medieval history, and known more locally as the Gemüseinsel – the vegetable island, abundant in edible produce.

For the historical linguist too, it’s a place of significance. Here’s my introduction to a text to which Reichenau has given its name, treasured for its evidence of the slow birth of a new language.

Subscribe now

What are the Reichenau Glosses and why should I care?

It’s fair to say that the Reichenau Glossary isn’t the best known historical document. Yet it deserves its moment in the spotlight, being of great interest and importance in tracing the history of the Romance languages.

This modern-day language family (including French, Italian, Spanish, Catalan, Portuguese and Romanian) emerged out of Latin, which the Romans had spread across Europe through their conquests.

There’s really no hard line though to say when Latin ended and Romance began. This is partly because our sources for early Romance are scarce, but also because language change is slow and gradual, barely perceptible when you are living through it. There are no clean breaks between generations of language users to which we could easily add our labels.

The Reichenau Glossary is one text, though, that tells us a great deal about when certain features of the Romance languages had arisen, and how far the Late Latin language had gradually diverged from its older, classical form by the time the glossary was composed.

As the name suggests, the glossary is several long lists of words, compiled in the 8th century AD, and found in 1863 among the library of the abbey on Reichenau. The glosses are principally written in the manuscript Karlsruhe 115, housed today in the Badische Landesbibliothek. The language of the text is Latin, and the purpose of the glosses is to explain and provide synonyms for words used in St Jerome’s 4th/5th-century Latin translation of the Bible, known as the Vulgate. One part goes through tricky words in the Bible book by book, another part glosses words alphabetically.

Based on the vocabulary used, we can say that it was not actually created on Reichenau, despite the abbey’s prowess in making manuscripts. Instead, it seems to have originally been written for scholars in what is now northern France, specifically at the abbey of Corbie. The language of that region would have been something very Latin-like, and people would have thought of it still as the lingua Romana. From this linguistic identification, the term Romance would eventually emerge.

Yet the simple fact that the glossary was written implies that some readers of the Vulgate were having difficulty in understanding the language of Jerome from three centuries earlier. Many shifts in vocabulary had occurred since then, altering the meaning of some words, while others had been lost from everyday speech and replaced by new vocabulary.

In the glosses given, we see first or very early appearances of words and meanings that are still common in French today. It’s not accurate to say that the language of the people reading the glosses was recognisably or distinctly French yet, but it was heading in that direction. In the Reichenau Glosses and the lexical shifts that necessitated them, we can witness the slow birth of French.

“From the multitude of words that were deemed difficult for eighth-century readers, one can see how much the lexicon had changed in the intervening centuries. The words given as glosses were generally those destined to survive in Romance.”

(Alkire & Rosen 2010: 319)

Let’s dive into some of those words to see how.

Late Latin and Future French: examples

Here is a serving of examples, with my explanations, of what the Reichenau Glosses look like. For each, we first have the word used in the more classical language of Jerome’s Bible, followed by a synonym that users of the glossary would have better understood. For a longer list, with brief notes, have a look here.

ager ~ campus

‘field’

While Jerome uses the Latin word ager for ‘field’ (hence English agriculture), it’s here glossed as campus. The latter is a well-documented word in our classical sources, but its usual meaning is of a large, flat area, not necessarily one cultivated for farming. By the time of the glosses though, campus had clearly taken over the agricultural sense of ager, which it still has today in French champ.

liberos ~ infantes

‘children’

Liberi is a common classical term for ‘children’, specifically used in the plural. Infantes meanwhile were very young children, of the non-speaking (in–fans) age. Yet infantes then broadened in its meaning, coming to refer to children in general, as in French enfants.

arena ~ sabulonem

‘sand’

It’s true; an arena was originally just a lot of sand. The glosses reveal that in Late Latin, it was losing ground to sabulo, once a rougher kind of sand, or gravel. The latter is behind the usual words for ‘sand’ in French and Italian today, sable and sabbia.

caseum ~ formaticum

‘cheese’

English is like Spanish, in that the words for ‘cheese’ (cheese and queso) go back to Latin caseus. Yet French’s word, fromage, is the result of a new formulation, once referring to the mould (forma) in which cheeses were made. This gloss is one of those that help to pinpoint the document’s Gallic origins.

forum ~ mercatum

‘market(place)’

The forum, that Romanest of city sites, was giving way to a mercatum, where merchants continued to do business. This is recognisably the source of English market and French marché.

ore ~ bucce

‘mouth’

Classical Latin mainly used os for ‘mouth’ (hence English oral), while bucca meant ‘cheek’. Perhaps through casual, colloquial use, bucca was destined to take over from os. This gave rise to words in Romance today like Spanish boca, Italian bocca and French bouche.

vespertiliones ~ calvas sorices

‘bats’

Bats got renamed too, going from the etymologically sensible Latin word vespertilio (from vesper ‘evening’) to calvas sorices ‘bald mice’ (or ‘shrews’ originally), a designation maintained until today in French chauves-souris. Not all lexical change may be change for the better.

rerum ~ causarum

‘(of) things’

The useful Latin word res is the source of English words like real and republic (the ‘public matter’). Yet the root noun itself lost out to causa as the Romance word for ‘thing’, as in Italian cosa or French chose.

pulcra ~ bella

‘beautiful’

While classical authors might have called a beautiful thing pulcher, in the glosses we witness the revival of the adjective bellus, responsible for French beau and Italian bello.

optimos ~ meliores

‘best’

The glosses also give us glimpses of grammatical restructuring. Like how English has good/better/best, Classical Latin has bonus/melior/optimus with three distinct words to express the three degrees. What we see here though is that melior, originally only ‘better’, is taking over from the superlative optimus. This anticipates the situation in French today, in which meilleur is both ‘better’ and ‘best’. What differs now is whether the definite article is used too (‘best’ is le meilleur).

saniore ~ plus sano

‘healthier’

Likewise, this gloss hints that the comparative form of adjectives is switching from being expressed through an ending (-ior) to being expressed through a separate word (plus). This is still how French tends to form comparatives, as in plus petit ‘smaller’ or plus sage ‘wiser’.

edunt ~ manducant

‘they eat’

Manger is well known as the verb for ‘to eat’ in French today, and here we see how it took over from edere. The latter is part of a large word family, including English eat, but it was usurped by manducare, originally ‘to chew’. This verbal root is why the body part that is tasked with chewing is your mandible.

abio ~ vado

‘I go’

This is another case of a core verb being replaced. Abire ‘to depart, to go’ builds on the verb ire. This is still a part of French today (e.g. j’irai ‘I will go’), but it’s been joined in the overall conjugation of the verb aller ‘to go’ by other roots. This is known as suppletion. So, while Latin ire has given French the future-tense form j’irai, Latin vado is the source of present-tense je vais.

cecinit ~ cantavit

‘he/she/it sang’

As well as verbs getting messier, many relics of grammar within Classical Latin were tidied up and ‘regularised’ in the subsequent centuries. The verb canere ‘to sing’ contained within its paradigm vestiges of reduplication: doubling the first consonant, as in the perfect-tense verb cecinit. This was surely an oddity for later speakers, and canere lost out to the more comfortably regular derivative cantare.

adversum ~ contra

‘against, opposite’

Prepositions were changing too! It seems that adversus (literally ‘turned-to’) had become unfamiliar, with contra assuming its functions, as contre does in French today.

semel ~ una vice

‘once’

It’s not just single words that the Reichenau Glossary gives us information about; some phrases have their earliest appearances in it. While Cicero and Jerome might have said semel for ‘once’, the readers of the glosses were instead used to the phrase una vice, just as the French today say une fois.

singulariter ~ solamente

‘only’

This is an early attestation of a very Romance construction: forming adverbs with -mente/-ment. This is the typical means of doing so in French today (e.g. heureusement ‘happily, facilement ‘easily’, étrangement ‘strangely’, seulement ‘only’, etc.).

It’s an ending today, but -mente/-ment was actually once a noun and a separate word. It comes from Latin mens ‘mind’, as in English mental. This noun fused with the adjectives that used to modify it, creating a new class of adverbs. For instance, to do something clara mente ‘with a clear mind’ came to mean simply ‘clearly’.

Italia ~ Langobardia

‘Italy’

The migrations of the post-Roman period in western Europe had led to some geographical renaming. The Lombards (etymologically: the ‘Long-beards’) were a Germanic-speaking people who rushed into Italy after the peninsula’s existing powers had exhausted themselves through the Gothic War. The new arrivals were so successful that, from a northern Gallic perspective, all of Italy could bear their name. Since then, Italia has made a comeback, while Langobardia has shrunk into the region of Lombardy.

Gallia ~ Francia

‘France’

To end, a real sign of how much things had changed: thanks to the Franks, Gaul had become France!

Beyond the examples

This is only a smattering of the many words that the Reichenau Glossary includes. There’s lots more that I could get into (available here), such as the everyday words of Germanic origin mentioned that therefore must have entered the local language. I also don’t know why some of these words were included; the presence of cheese, bats and France in Jerome’s Vulgate would surprise me, but I admit I haven’t looked for them. The glossary may have been broader in its scope than just that mighty text. I hope at least that the examples above are enough to demonstrate why a load of 8th-century glosses are of such interest and value for historical linguistics.

Yet there is a bigger-picture point to consider. The language of the document is arguably still Latin, not yet French or some other Romance language. The grammar and shape of the words are Latinate, and there is no indication of changes in sound too. The glosses are not translations, matching words up between two languages that were understood to be separate. Instead, they arose from changes in local everyday speech that were then presented in manuscripts in a more ‘proper’ Latin form.

The relationship between that speech and Latin was still one of register, rather than a recognised divide in dialect or language. Education expanded a person’s register range. Latin, a composite of various essential linguistic features, remained the higher, formal register that was worthy of being committed to writing. In other words, Latin was (and could still be considered) a unified language at this time, but it was certainly starting to tear at the seams.

Overall, the glossary is a fascinating witness to the gradual change from one language to another. Out of these countless individual shifts in sounds, vocabulary and grammar within the language of post-Roman France, something that we can call French would eventually be born.

END.

References:

• Alkire, T., & Rosen, C. (2010). Romance languages: A Historical Introduction. Cambridge University Press.

• Engels, J. (1968). Les” Gloses de Reichenau” rééditées. Neophilologus, 52(4), 378.

• Quirós, M. A. (1986). Las glosas de Reichenau. Revista de Filología y Lingüística de la Universidad de Costa Rica, 12(1). 43-50.

Texts:

• Badische Landesbibliothek

• Orbis Latinus

Reworked article from dannybate.com. Images my own, from Wikimedia, or from credited sources. Cover image: the magnificent interior of St George’s Church on Reichenau, with its murals from the 10th century.

Consequently, a label like ‘a European language’ is to be used with a pinch of caution and a proviso or two. This is all the truer in the modern era, when colonial expansion has transported and transplanted European languages, such as English and Spanish, to other continents. Can English be straightforwardly called ‘a European language’, for example? History might say yes, but the present-day distribution of its speakers would argue back. Even long before our own day, certain languages could not be contained by a single landmass. The Greek language offers a rich example.

Greece, Greeks and Greek – today, country, people and language alike all feel securely European in location, identity and outlook. Testifying to its European-ness, the Greek language holds official status not only in the modern countries of Greece and Cyprus, but also in the European Union, to which those countries belong. It appears on European money too, as “EYPΩ” on euro notes.

Brushing over the question of what larger geographical unit we assign Cyprus and south-of-the-Green-Line Cypriots to, the Greek speakers of the Old World are largely concentrated on territory that we can comfortably call Europe. All of this is no wonder, considering that we get the term Europe from Greek.

The territory of Greece today comprises the mainland at the bottom of the Balkans and thousands of islands strewn across the Aegean and Ionian Seas. On this basis, you’d be forgiven (by me, but perhaps not by the Greeks) for thinking that the language has never wandered too far from this Hellenic heartland. Quite the opposite is true, and so much of its historical usage and cultural achievements have happened outside of Europe.

The purpose behind this post is to foreground that linguistic history, to re-introduce Greek to you as a language not just of Europe, but of Asia and Africa too. It can lead us into a world of extra-European speech and writing, as the language was passed on to peoples beyond the borders of a Greek identity, even generously defined. To begin this exploration, we, like Odysseus himself, first have to head east and make land on the shores of Anatolia.

Subscribe now

A Language of Asia

Once upon a time, Asia was a much, much smaller place. The term Asia has ballooned in its designation over the millennia, increasing in size to include regions (and the peoples living in them) far to the east of its earliest reference point. This was the western edge of Anatolia and the eastern coast of the Aegean, today part of Turkey.

The word seems to have been a foreign acquisition by the Greeks; something like it first appears in Hittite sources (as “A-as-su-wa”), and then in the Linear B script that wrote down Mycenaean Greek (as “A-si-wi-ja”). Originally, when Greeks of the archaic and classical eras wrote the term Ἀσία, what they had in mind was the region of Lydia, western Anatolia, or Anatolia in general. This segment of the world’s map can still be referred to as Asia Minor. ‘OG Asia’ might also be historically appropriate. The Romans, who took control of it in the 2nd century BC, maintained the narrower meaning in their province of Asia.

However, there are hints of the world-striding landmass already in the classical era. The historian Herodotus thought of the world’s land in terms of three broad continents: Europe, Asia and Libya (Africa). The second extended pretty far east in his estimation.

“Πέρσαι οἰκέουσι [Ἀσίην] κατήκοντες ἐπὶ τὴν νοτίην θάλασσαν τὴν Ἐρυθρὴν καλεομένην … μέχρι δὲ τῆς Ἰνδικῆς οἰκέεται Ἀσίη. τὸ δὲ ἀπὸ ταύτης ἔρημος ἤδη τὸ πρὸς τὴν ἠῶ, οὐδὲ ἔχει οὐδεὶς φράσαι οἷον δή τι ἐστί”

“The Persians inhabit [Asia], reaching to the southern sea called the Red … As far as India, Asia is inhabited. But after this, desert is to the east, nor can anyone say what kind of land is there.”
(Herodotus, Histories 4.37,40)

The fault-line between Europe and Asia ran through the Dardanelles and the Bosporus, the slim straits that link the Mediterranean to the Black Sea. Looking across the waters from the western side (the European perspective), Asia could easily roll on to the world’s end, and the geographical label could grow as far as ancient knowledge of it permitted.

The ancient Greeks, mind you, were not going to be penned in by a mere sliver of water like the Dardanelles. They could be found on both sides of the Aegean. Up and down the west coast of Asia Minor, the migration of Aeolian, Ionian and Dorian colonists enlarged the definition of Greece.

Though these cities of Asia may have been secondary settlements for the Greeks, they achieved great prestige and cultural weight. Three of the seven wonders of the ancient world were on the Asian side of the Aegean. Herodotus himself (born c. 485) came from Dorian-developed Halicarnassus, famed for its mausoleum, and now entombed by the Turkish city of Bodrum. Tradition says that Homer, the supreme Greek poet and the inspiration of so much European literature, lived and composed in the Asian region of Ionia. For people to the Greeks’ east, Ionia became synonymous with all Greece; Ionia is in fact the origin of the word for ‘Greece’ itself in many languages today: see Turkish Yunanistan, Hebrew יוון (Yavan), and Persian یونان (Yūnān).

The Greek population in Anatolia survived and thrived through the subsequent medieval and modern periods, right up to the aftermath of the First World War. The language prospered in Anatolia, a key part of the Greek-speaking Eastern Roman Empire (or Byzantine, if you prefer). After this state gave up the ghost in 1453, the conquering Ottoman Empire ruled over the substantial ‘millet’ of Greek speakers in Anatolia, until that empire’s final dissolution in 1922. The Greco-Turkish War (1919–1922) and localised massacres of civilians then triggered an exodus of Greeks from the emerging Turkish state; the cataclysm culminated with the 1923 population exchange between Greece and Turkey. Refugees from one country were made official in the other, and thousands more were ‘encouraged’ to leave. So ended centuries of Turkish life across the Balkans and Greek life in Asia Minor.

That said, the categories of Turk and Greek were determined primarily by religion: Sunni Islam vs. Orthodox Christian. Communities of Greek-speaking Muslims were exempt from the population transfers. Greek can still be heard among them today. On the southern coast of the Black Sea, in Trabzon Province, you can find people praying in mosques yet speaking Greek – Pontic Greek, to be precise, also known as Romeyka.

This endangered dialect has many noteworthy features; being outside the confluences and confusions of the Balkan linguistic zone, Romeyka preserves gems of ancient grammar lost from standard Modern Greek, like the infinitive form of verbs.

After crossing the Dardanelles, we need not travel too far to find the Greek language waiting for us on the Asian side. This should be unsurprising, given that it was in Greek that the whole concept of Asia was nurtured. The language once wrapped itself around the coast of Anatolia, but later movements of people and politics cut up that wrapping and left Romeyka stranded as an island of speech.

Another such island, long since submerged by subsequent waves of language and culture, lay much further to the east, in sight of the Himalayas.

In the 320s BC, Alexander the Great and his Macedonian army broke upon Central Asia, pushing his lightning-quick conquests into lands now part of Afghanistan and Pakistan. Alexander was now the length of one major empire away from the Greek heartlands – specifically the Achaemenid Empire of the Persians, which he was in the final stages of destroying. We have to wonder if Alexander was surprised, as he approached the gateway to India, to find Greeks already living in the neighbourhood.

During the heyday of the Achaemenids, under great Persian kings of kings like Xerxes, communities of Greeks had been transferred to the other side of the vast empire. Some apparently went willingly. As Alexander approached, any joyous reunions between Greeks might have been short-lived. One story recounts that the day after Alexander encountered the Branchidae in Bactria (today northern Afghanistan, Tajikistan and Uzbekistan), he had them all put to death.

Alexander’s conquest of the region added new impetus to the Hellenisation of these Asian lands. New Greek-governed colonies were founded, many named after Alexander himself or his horse. He was Alexander the Great, remember, not Alexander the Humble. These colonies needed a religious life, and so classically columned temples were built to honour the Greek gods. The ruins of such temples at Ai Khanoum (Takhar Province, Afghanistan) were excavated in the 1960s, and a foot belonging to the great god Zeus was unearthed.

Long after the ruler of the world had departed and died, Greek successors continued to govern Greco-Bactrian and Indo-Greek kingdoms. These were located respectively north and south of the Hindu Kush. The weight of the Greek presence swung eastwards, towards India and its Indo-Greek state, in the 2nd century BC, as the nomadic Yuezhi took over Bactria.

A resurgent Persia would later sever the connections between Mother Greece and her Himalayan offspring, but the Greek elites there were hardly marooned; their realms were sewn into a patchwork of surrounding cultures and kingdoms. The rulers in India maintain a distinctly Greek culture in our archaeological sources for over a century, but they were clearly integrated into Indian ways too. The pillar set up by the Indo-Greek ambassador Heliodorus, found far away in central India, honours a local god of gods, Vāsudeva.

Their coins are a meeting of local languages with Greek, and their numismatic designs were copied by other states in Bactria and India. Futhermore, the Greeks’ style of sculpture for gods and heroes likely influenced the statues of Buddhism. They make appearances in Sanskrit and Pali texts – a Greek person is a Yavana in Sanskrit, and a Yona in Pali, again derived from far-away Ionia. One particular Yona, the Seleucid king Antiochus II Theos, is even mentioned in an edict of the mighty Indian emperor Ashoka the Great.

Ashoka advertised his munificence (such as his gift of medical expertise to Antiochus) through rock-cut inscriptions across his empire. Some of these are in fact written in Greek, which implies a Greek-reading audience. The Greek alphabet was even applied to local languages; for example, our sources for Bactrian, an Iranian language, are inscribed in its familiar letters.

One detail that the sorry story of the Branchidae includes is that they still spoke Greek, some 150 years after their migration to Bactria from Miletus in Asia Minor. If this transfer took place under Xerxes, then the Greek language arrived in Bactria sometime after the year 480 BC. Both the historical and archaeological evidence point to the vitality of Bactrian and Indian Greek at least until the line of Indo-Greek kings fizzles out in c. 10 AD.

Greek was surely one language among many in those lands, perhaps functioning as a prestigious lingua franca. Nonetheless, we have a period of Greek language in Central and South Asia spanning almost five hundred years. This is not a negligible episode of the language’s life, just an ephemeral anomaly, but rather the story of a newcomer that put down roots and contributed to an altogether different linguistic landscape.

A Language of Africa

Like Alexander’s army beside the Beas River, we should now turn back and head west. The conqueror’s successes had previously included the great prize of Egypt. Alexander had been accepted as pharaoh there in 332 BC. He ordered the construction of another city called Alexandria, although this one, on the north African coastline, has survived until today.

After Alexander’s death, his generals dismembered the empire, and Egypt ended up in the hands of his trusted officer, Ptolemy. The Ptolemaic period of Egypt, from Ptolemy’s pharaonic elevation in 305 BC to Cleopatra’s suicide in 30 BC, cultivated a flowering of art and science. As an example, it was under the Ptolemies that the Library of Alexandria was founded and patronised, although, frankly, the importance of the library and the tragedy of its destruction(s) are greatly exaggerated. Alexandria’s intellectual life continued into the time of the city’s Roman rulers; it was home to Hypatia and the Jewish philosopher Philo. It was in Alexandria that the Hebrew Bible (to Christians: the Old Testament) was definitively translated from Hebrew into – yes, you guessed it – Greek.

The city was a powerhouse for the language, which could unite its multinational and multireligious inhabitants. Greek was what new arrivals would have turned to to navigate the city and Egypt in general. Among my favourite finds is a papyrus phrasebook from Late Antiquity, which provided Armenian visitors to Egypt with essential Greek words and idioms, except written in Armenian letters for ease of use.

True, the tongues of Egypt would have still included Egyptian, specifically the final stage of the language known as Coptic. The corpus of Coptic literature demonstrates that Greek failed to replace native Egyptian speech entirely. Yet even in this regard, the power of Greek can be felt, because Coptic eschewed Egyptian writing systems and was written with Greek letters instead.

Alexandria remains a shining sun in the history of the Greek language. I would now have you recall, for this article’s purposes, that Alexandria is not in Europe.

Yet again, Alexander the Great was not the first person to bring Greek to the land in question; ancient graffiti carved into a colossal leg outside the temple of Abu Simbel attest to Greeks employed in Egypt as early as 591 BC. Nonetheless, in founding Alexandria, he left the legacy of a Hellenised and Hellenophone Lower Egypt. Alexander would be proud of this legacy. I’d wager he also would be greatly surprised by the teachings and the success of the religious movement that would later become synonymous with Greek speech and identity.

Christianity found a base in Alexandria. Tradition says that Mark the Evangelist was its first bishop. His status as an evangelist means that he has historically been identified as a writer of one of the gospels – surely the four most influential Greek texts of all time. I won’t venture an opinion on where the gospels were composed, but their subject matter (the life of Jesus of Nazareth) and context (Jewish and Greek culture in Roman Palestine) certainly do not point to their penning in Europe.

St. Mark was later martyred and laid to rest in Alexandria until that rest was disturbed by Venice’s ‘acquisition’ of his holy remains. As the legalised faith later established its structures and empire-wide hierarchy, Alexandria was one of the five top jurisdictions, officially all equal with each other, in an arrangement known as the Pentarchy.

This set the city up as a religious centre with its own orbit; believers up and down the banks of the long Nile looked to the Patriarch of Alexandria for guidance and judgement. Much of that guiding and judging happened in the medium of the Greek language. Through this, Greek would thrive in regions to the south – even further south than the vandalism at Abu Simbel.

The Egyptians didn’t have exclusive ownership of the Nile. In the south, before the river splits into White and Blue, lies the historical region of Nubia. Today, it straddles the countries of Egypt and Sudan. In Nubia, the ancient Kingdom of Kush jostled for position with its neighbours: Egypt to the north, and later Aksum to the south and east too.

Egypt was an ancient frenemy for the Kushites. Shared culture between them has meant that it is Sudan, not Egypt, that today boasts the most ancient pyramids of any country.

Later on, after Alexander had been and gone, Nubia was in contact with Ptolemaic Egypt. This experience could be positive and negative; religion and art, as well as spears and arrows, might be exchanged. Kush’s capital city, Meroë, was greatly Hellenised under Egyptian influence, gaining essential Greek things like public baths. The story survives of one king in Nubia, Ergamenes, who...

“... μετεσχηκὼς Ἑλληνικῆς ἀγωγῆς καὶ φιλοσοφήσας…”

‘... had partaken of Greek education and loved wisdom…’
(Diodorus Siculus, Bibliotheca Historica 3.6)

Wisdom-loving Ergamenes then went on to do some rather awful things to the priests of Meroë. Instances of sacerdicide aside, the city became “a little Nubian Alexandria” through interactions with Egypt under the Ptolemies, according to the historian M. I. Rostovtzeff. The welcoming of all things Greek in ancient Nubia must also have included the language.

The considerable employment of Greek there, alongside other languages, is perhaps best demonstrated by the downfall of Kush. The kingdom was conquered in the mid-4th century AD by Aksum, and the gloating victors set up inscribed thrones and standing stones across their enlarged kingdom. To ensure Aksum’s subjects could understand their betters’ declarations, such texts are often multilingual. We would expect to see a language like Ge’ez, a historical and liturgical language of Ethiopia, but Greek appears on them too.

In fact, this was the conquest of one African kingdom with a strong Greek tradition, by another African kingdom with a strong Greek tradition. Our fragments of information about the earliest days of Aksum (which would in time transform into the long-lasting Ethiopian Empire) are transmitted to us in Greek. Zoscales, an otherwise unknown 1st-century Aksumite king, was reported to be both ambitious and knowledgeable in Greek. Over the following centuries, his successors minted coins with Hellenised names and titles. The existence of Sembrouthes in the 2nd century is only known from a single inscription, written in formulaic but decent Greek prose, which acknowledges him as Aksum’s ‘king of kings’ – “βασιλεὺς έκ βασιλεών”.

Back in Nubia, the use of Greek was not just a passing fad confined to antiquity. It seems to have withstood the fall of Kush, and then the arrival of Islam in Africa. As late as 1372 AD, we have a Nubian bishop, Timothy, being consecrated in Alexandria and sent back to his flock with a patriarchal greeting in Greek. The greeting is added to official documents testifying to Timothy’s consecration, which were later buried with him.

The documents’ use of Arabic (the language of regional politics at the time) and of Coptic (the language of Egyptian Christianity) is unsurprising, but the inclusion of Greek is unexpected. That is, unless we presume that it was going to be understood back in Nubia, and that, in the words of Stanley M. Burstein, “Greek remained the official language of Nubian Christianity right to the end of its long and remarkable history”.

It seems that that end came at the turn of the 16th century. Another island of the Greek language (in writing, if not also in speech) was finally overcome by the forces of history. If we include the Abu Simbel graffiti, both in terms of chronology and geography, then we have evidence for a millennium of Greek along the southern stretch of the Nile.

To conclude, somehow

Having journeyed first to the Hindu Kush and then to the ruins of Nubian Kush (a strange coincidence), I’m reminded of an archaeologist’s maxim: “pots are not people”. This piece of wisdom is intended to remind scholars that archaeological finds are not securely linked to people and their identities. One type of pot can belong to diverse peoples, and one people can utilise many types of pot.

The same is true of languages. Language is bound up with identity, often contributing to it, and yet also able to separate itself and wriggle free from allegiance to a single nation. Greek is one example.

On arrival, the Greek language may have first reached Bactria and Nubia borne by self-professed Greeks – that is, people calling themselves something like Ἕλληνες (Héllēnes) – but alternatively, in those ancient days, a single identity that matched exactly with all speakers of Greek may not have seemed obvious or important. Smaller identities like Ionian and Macedonian might instead have been foremost in their minds. For the average Alexandrian, I imagine that the Greek language was just a fact of life, not a personality trait that took precedence over being a denizen of that great city. By the time that today’s single, primary Greek identity emerged, an identity that also overlaps with European-ness, the Greek language had already slipped the net, and been given to the world.

What the world beyond Europe did with Greek is worthy of our consideration, study, and even speculation. It may well have taken on distinct new hews as it met other languages in the mouths and minds of Bactrian, Indian, Egyptian and Nubian speakers. It may have given words and sounds to them, and accepted the same in return. Had more texts come down to us today, new dialectal differences might be perceptible, but we only have scraps and fragments.

Nonetheless, one point stands: Greek has been a linguistic ingredient of parts of Asia and Africa, as well as of Europe. The difference in its durability outside of Europe has partly been a matter of concentration; smaller groups are more easily assimilated into a linguistic majority. Like colours on an artist’s palette, a limited language is soon lost in the mixture, but its contribution is still there in the overall picture.

END.

Non-linked references:

• Benedetti, M. (2014). Greek and Indian languages. Encyclopedia of Ancient Greek Language and Linguistics. 59–62. Leiden: Brill.

• Burstein, S. M. (2008). When Greek Was an African Language: The Role of Greek Culture in Ancient and Medieval Nubia. Journal of World History, 19(1), 41–61.

• Hatke, G. (2013). Aksum and Nubia: Warfare, Commerce, and Political Fictions in Ancient Northeast Africa. NYU Press.

• Nikoloudis, S. (2008). Multiculturalism in the Mycenaean world. Anatolian Interfaces: Hittites, Greeks, and their Neighbors. 45–56. Oxford: Oxbow Books.

• Rostovtzeff, M. I. (1957). The Social and Economic History of the Roman Empire. Second edition. Oxford: Clarendon Press.

• Stoneman, R. (2019). The Greek Experience of India: From Alexander to the Indo-Greeks. Princeton University Press.

Texts:

• Diodorus Siculus, Bibliotheca Historica 3 here

• Herodotus, Histories 4 here

Original post for dannybate.com. Images taken from Wikimedia.

Cover image: the 20-stater coin of the Greco-Bactrian king Eucratides I, found in Bukhara, Uzbekistan, and now located in Paris. It's claimed to be the heaviest gold coin minted in antiquity.

Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ Ν …

A glimpse at the Greek alphabet grasps its similarity to the Roman letters your eyes are now reading. There are differences, of course, but we have an A, a B, an E, an I and a K, and most of the characters in between are not too far-removed from Roman counterparts.

There is one big reason behind all these similarities: the Greek alphabet (specifically, an ancient form of it) was a forefather of the Roman one, adapted for the Latin language, and for centuries used for English too.

One major difference, which you can’t tell from the sight of the letters’ shapes, is in their names. It’s well known that the Greek alphabet comes with distinct labels: alpha, beta (more like ‘veta’ in modern pronunciation), gamma, delta, epsilon and so on. These names are ancient, just as old as the symbolic shapes they refer to.

Yet, if the Greek alphabet is ancestral to our own, what happened to its integral naming system? In other words, why and when did the Greek names alpha, beta, gamma, delta, etc. turn into our own beloved a, bee, cee, dee and the rest?

Subscribe now

The Alphabet, from Egypt to Rome

To begin at the beginning: once upon a time, in the ancient kingdom of Egypt, our letters looked like things.

Each emerging letter signified one sound; that sound was determined by the first sound of the words for those things. This is the revolutionary principle at the root of our alphabet’s history, a realignment of old hieroglyphs behind a single sound-based rule.

Note the word word there. Items of vocabulary were essential intermediaries in setting up this longlasting system of ‘one sound for one symbol’. These ordinary words were also summoned to serve as the new letters’ names.

For an example, a glottal stop was one such sound of speech in need of a letter. So, a glottal-starting word in the Semitic language of the alphabetic innovators was required. The everyday noun chosen was ʾalp, meaning ‘ox’. Now, a hieroglyphic head of cattle would spell that throaty consonant whenever required.

The ox’s head was now, in a nascent way, a letter. As such, it needed a name, a means of referring to it in ancient spelling lessons. It was christened ʾalp, unsurprisingly. Much the same way that we today might say ‘oh no, that’s spelled with an A’, the earliest alphabetic writers might critique a miswritten word with ‘you forgot the ʾalp there, my friend’.

If you think this is all irrelevant antiquity, allow me to make it familiar. This ʾalp character is the origin of our letter A. It would make its way up the coast of the Mediterranean, landing in the hands of the Phoenicians. These maritime marvels of the Levant continued to call it ʾalp, and to use it for a glottal stop, but its shape had by then been reduced to a three-lined 𐤀.

They then passed the parcel to the Greeks. They repurposed 𐤀 as a vowel, a drastic change to the sound of the letter. The Greeks also settled on Α, a rotated version of 𐤀, but they did maintain the old names. Α was and is alpha.

You know more old Semitic vocabulary than you think – the ox is one part of the alpha-bet, just as triangular river deltas find an etymological origin in Semitic-speakers’ doors.

From the Greeks, the system would sail west, to Italy. There it would be put into the hands of the peninsula’s regional powers at the time: the Etruscans, the Oscans, the Umbrians, the Veneti, the Gauls, and the then-unimportant settlements that would eventually coalesce into Rome. Within one of the three qualities of the letters (sound, shape, name), there was another break in transmission after the letters arrived in Italy. It was a total redesign of their names.

We don’t know for how long the Greeks remembered that the origin of their letter names was in ordinary Semitic words; regardless, another step further west was another step away from Semitic speakers who could have reminded early alphabetic writers of that fact. As soon as we reach a period in ancient Italy’s history with enough literacy and literature to find the letter names written down, we see that they have been renamed.

Either by the Etruscans or the Romans (more likely the latter), a radical new system had been devised, which dispensed with the old Graeco-Semitic names. In their place was a simple principle: briefly name the letter by the sound that it represents.

With the vowel letters (A, E, I, O, U), the vowel sound could straightforwardly be the name. The Romans knew this quintet of letters as ‘aa’, ‘ehh’, ‘ee’, ‘ohh’ and ‘oo’.

For those letters representing consonants, an extra E-vowel was usually added to the name, to make it a comfortably complete syllable.

A meaningless hodgepodge of ancient names had thereby been streamlined; the Greeks’ beta, delta, lamda, pi, sigma and tau became the Romans’ be, de, el, pe, es and te. From this point, add in a couple of millennia of use and natural sound change, and you arrive at some very familiar names.

The alphabetic status quo that resulted from the renaming was the classical Roman alphabet, as it would’ve been known, taught and used in the days of Caesar and Cicero. This updated version of the already-ancient alphabet is the common ancestor (in sound, shape and name) of many subsequent sub-alphabets used today, including English’s own.

Phonetic Logic

However, despite the attempt to rename the letters according to one rule, there were further factors at work.

Take, for example, the Romans’ names for C, K and Q. These were ce, ka and qu, pronounced ‘keh’, ‘kah’ and ‘koo’. These are recognisably the origins of the English names: ce, kay and cue (isn’t it odd to see the names written out so?).

As we assume, with good reason, that a C was for the Romans always a hard /k/ sound (hear IPA notation here), then this trio of vowels in the three names was a necessity. They separated the names for three letters that all represented the sound /k/. Bear in mind that the English name of C containing a /s/ sound comes from a post-Roman development. The same applies to our softly spoken G, which was once only a hard /g/.

Yet there is an even bigger pattern behind the Romans’ names, and therefore those still current in English. This pattern demonstrates an awareness of phonetics, of the properties of the sounds of speech.

Consider the Romans names for B, D and P. In them, the defining consonant comes first, before the vowel: be, de and pe. From these, via the Great Vowel Shift, come the English names bee, dee and pee.

Now consider F, L, M and S. In them, the defining consonant comes second, after the vowel: ef, el, em and es.

To appreciate the full before/after pattern at play, we can write out the classical alphabet in full:

What we have here is a split along a phonetic line, according to the qualities of the individual sounds.

With those consonants that halt and then release the airflow (known as ‘stop’ or ‘plosive’ sounds), they precede the vowel in the name. These stops were the /b, k, d, g, k, p, k, t/ sounds behind the Romans’ letters B, C, D, G, K, P, Q and T.

With other consonants, made with an uninterrupted stream of air, the defining sound came second in their Roman letter names. These were the nasal sounds /m, n/, the fricatives /f, s/ and the liquids /l, r/, spelled with M, N, F, S, L, R.

The alphabet had originally been named through words for everyday objects. Now, sounds and their properties were leading the way.

There are, of course, some exceptions to the rule. Nothing in human language is ever perfect. The letter H ought to be in the ‘after’ category, denoting as it did the breathy fricative /h/. Inversely, X spelled the duo of sounds /ks/, beginning with a stop sound that should have put it in the ‘before’ category. If following the phonetic principle to the letter, the Romans would've called H and X ‘eh’ and ‘kse’.

However, such arrangements of sounds would’ve been uncomfortable for classical mouths. Spoken Latin had a dislike ending syllables with /h/ or starting them with /ks/, a dislike that English happens to share.

Time also has complicated the picture. As mentioned already, the Romans assigned C and G to the ‘before’ camp because the letters then stood for the stops /k/ and /g/. Regular sound change in later centuries produced the soft pronunciations of C and G that their English names contain today: namely, the /s/ fricative of the C in centre and city, and the /d͡ʒ/ affricate of the G in gell and gist. Such post-classical changes have slightly obscured the original phonetic rule behind the Roman renaming, but the logic can still be recovered.

That rationale leaves me amazed by the ancients’ understanding of the properties of speech, apparently without any scholarly study resembling our modern field of phonetics.

A Personal Phonetic Epilogue

This history of letter names makes up the meat of Chapter M in my book, Why Q Needs U. In the absence of much else to say about the letter M, the chapter offered a chance to discuss this pattern pertaining to the alphabet as a whole.

I must admit, though, that I didn’t address the obvious question: why? Why did Roman (or Etruscan) writers rename the letters according to this phonetic rule? The answer wasn’t obvious to me, and I feared that my readers wouldn’t be willing to take any more of my trademark academic caution and speculation.

However, in November, I threw the question out to an audience of experts: the linguists of the University of Edinburgh. This sundry crew, united in their loveliness, comprised scholars of languages of both the past and the present day, and it was the phoneticians among them who seem to have cracked the puzzle. Gilly Marchini and Benjamin Molineaux in particular recognised that acoustic clarity may be the key.

The letter names had a need to include and indicate their associated sound as clearly as possible. It would be a bad system that leaves its users unclear whether someone had just said F or G.

The thing is, stop sounds are clearest when they precede a vowel, standing at the start of a syllable. When used so, the burst (as the airflow is released) is relatively loud and audible. In contrast, at the end of a syllable or word, the burst and other qualities of a stop are often not so strident.

“If you listen carefully to the sounds at the end of the words nab, mad, nag, you may find that the so-called voiced consonants /b, d, g/ have very little voicing and might also be called voiceless … It would be more normal to say each of them without releasing the final consonants … You could even say cab and not open your lips for a considerable period of time if it were the last word of an utterance. In such circumstances, it is quite clear that the final consonants are not fully voiced throughout the closure.”

(Ladefoged & Johnson 2011: 61)

Meanwhile, the other categories of sound (the nasals, the fricatives and the liquids) might be louder and clearer after the vowel, in what’s known as the ‘coda’ of a syllable. This quality would make them better suited to coming second in the Romans’ system, something that the enduring names ef, el, em, en, er and es attest to.

Now I have the small matter of learning how to travel back in time, to include this insight (with due credit) in the drafts of my book. Failing that, it can wait for the paperback edition.

END.

Subscribe now

References

• Ladefoged, P., & Johnson, K. (2011). A Course in Phonetics. Sixth edition. Wadsworth Cengage Learning

• Wallace, R. (2011). The Latin alphabet and orthography. In Clackson, J. (ed.) A Companion to the Latin Language. 7-28. Blackwell Publishing Ltd.

Images my own or from Wikimedia. Original Substack article.

Christmas imagery is inseparable from low temperatures, what with all its snow, reindeer, and jolly gentlemen in fur-lined red jackets. This imagery finds its origins, of course, in the northern hemisphere, prior to the holiday’s exportation to the south of the globe. I always feel sorry for shopping-mall Santas suffering through an Argentine or Australian summer. For any meridional readers, then, take this article as a mental tonic, a vision of cooler days to come.

In particular, I want to climb into the family tree of the English word cold. It’s a piece of vocabulary with an ancient pedigree, and consequently some interesting connections too. It also offers an example of when the business of etymology runs up against the rules and laws that it itself had laid down.

Cold has been a part of our language since before English was English. It belongs to what is known as the ‘inherited’ part of the lexicon – this contrasts with more recently ‘borrowed’ vocabulary, such as the plentiful words obligingly received from the Normans after 1066. Prior to that fateful year, we find cold in our Old English sources. If you look it up in a dedicated dictionary, it will inform you that its Old English forebear is ceald.

Some caution, though: this is one variant of the word, used in the south of England. Just like the present-day language, English in this un-Norman-ised era was riven with geographical diversity. The folk of Wessex and Kent might have said ceald, pronounced with an initial /t͡ʃ/ sound (as in cheese). Yet further north, the variant was cald with a hard /k/.

We find the ‘Anglian’ version cald in the melancholy Seafarer, preserved in a tenth-century collection of poems. It speaks to us of a lonely sailor battling winter winds on the ice-cold sea.

“… Calde geþrungen
wǣron fēt mīne forste gebunden,
caldum clommum; þǣr þā ceare seofedun
hāt ymb heortan; hungor innan slāt
merewērges mōd…”

‘… By cold constricted
were my feet, bound by frost
in cold clasps; where then cares seethed
hot about the heart; hunger tears from within
the sea-weary soul…’

(The Seafarer)

It was from cald, not ceald, that Modern English gets cold. The latter yielded to the former over the course of the Middle Ages. Southern ceald makes occasional appearances in Middle English as “chald” and “cheld”, including in the conspicuously archaic language of the Ayenbite of Inwyt (dialectally dissected in the previous post).

In the initial sound of the word cold, we therefore unearth a history of divisions within the fabric of English speech.

Cold is not without friends abroad. On the linguistic basis of German kalt, Dutch koud, Norwegian kald, Swedish kall, and other relatives, we can propose that it was a word in their common ancestor, Proto-Germanic. Speakers of that tongue knew it as *kalda-, a word whose prehistoric status is indicated by its prefixed asterisk.

What I like most about *kalda- is that it was a past participle in their language, something halfway between an adjective and a verb, meaning ‘made cool’. In the same way that written corresponds to write, eaten to eat, or played to play, *kalda- was a past participle serving the verb *kalan- ‘to be cold’. The latter limped into our written record of English, occasionally appearing as calan ‘to be cold’. You’d use this to say you were cold in Old English. The verb has since been lost from English, while its offspring, cold, has outlived it by several centuries.

At the heart of cold is a root, a meaningful core, to which other elements could be affixed. It’s the *kal- bit that *kalda– and *kalan- had in common. Today, this root is something that cold shares with the similarly frigid words chill and cool.

Their differences in vowel and starting consonant are products of sound change, and also the phenomenon of ‘ablaut’, something I’ve explored in a previous post. Like how sing can alternate its vowels to become sang or sung, the separation in vowels between cold and cool is a relic of this ancient system of vowel-swapping.

At work behind these Germanic words is a common root. To acknowledgeable its possible vocalic variation, we can refer to this root as *kVl-, in which V stands for the changeable vowel. It’s now bordering on algebra. The consonants on either side of V were what really conveyed coldness in English’s prehistoric forefathers.

Such was the state of affairs in Proto-Germanic, the immediate ancestor of the Germanic languages. To excavate our root more deeply, we must pass through the shifts of Grimm’s law. This serial sound change, a darling of philologists, was a defining quality of Proto-Germanic that separated its branch from the rest of the Indo-European tree.

One of the series of shifts was the change of the voiced sounds *b, *d and *g into voiceless *p, *t and *k. This is how the inherited lexicon of English has knee and corn, while Latin (a non-Germanic language) has genū and grānum. The inverse of that change gives us a rule stating that whenever we have a Proto-Germanic *k, it was previously Proto-Indo-European *g.

This all is relevant to our root; we can propose *kVl- was once *gVl-.

To build the case that this root really was part of PIE, mind you, we need at least one more foundation from elsewhere in the family tree. That is to say, we need examples of the root *gVl-, meaning similar things, from other Indo-European branches. Do we have such findings? If you like to eat gelato, then you’ll know that we do.

In Latin, we find cold words containing the root gel-, including gelū ‘frost’, gelidus ‘icy’ and gelāre ‘to freeze’. This root is a match with the Germanic in both form and meaning.

On the two legs of Latin and Germanic, we can therefore reach up to Proto-Indo-European and grasp a common root for frozen things like *gVl-. Ideally, we would have a wider set of words that would build a surer base. Yet further evidence consists of some suspicious Slavic and a single scrap of Greek. The two pillars of Latin and Germanic will have to suffice to reconstruct *gVl-.

From this single source come English’s inherited words cold, chill and cool. Moreover, via Latin and its descendants (like Italian and French), we then have the acquisitions of gelato, congeal, gelatin, gel and even jelly.

The initial sounds of this set of cousins have changed since the days of Latin and Proto-Germanic. The trio of cold, cool and chill would’ve once been united in a voiceless stop /k/ in the days of Proto-Germanic. Likewise, gelato, congeal, gelatin, gel and jelly we pronounce with a /d͡ʒ/ sound, but this would’ve been a guttural voiced /g/ in their Latin antecedents. Whatever the sounds were and have become, though, the two groups retain an ancient voiceless/voiced split. Grimm’s law continues to be a fault-line in English vocabulary.

So far, so neat. This collection of words connects up at a prehistoric point according to predictable, generalisable rules like Grimm’s law. It’s an instance of when historical linguistics is almost at the level of the mathematical in its systematicity. Yet sometimes, the system crashes. There is in fact another Latin word, which contains the same two consonants found in gelū, gelidus and gelāre, and which has a sub-zero subject matter. This is the noun glaciēs ‘ice’. You may recognise it from English terms like glacier.

Given the good match in meaning and shape, it’s counter-intuitive to say that it’s difficult to graft glaciēs onto the family tree. The sticking point is explaining how *gVl- might also be the root behind glaciēs. While Indo-European languages historically had no problem swapping or deleting vowels, we lack the laws to explain how the vowel within the root *gVl- might be lost and then have an /a/ vowel added after it – in other words, how we get from *gVl- to the gla- in glaciēs.

I’ve read a couple of theories; one is that glaciēs does contain the same root as in cold and gelato, but was also influenced in its form by the unrelated word aciēs ‘sharp edge’. This isn’t convincing, since an affinity in meaning between aciēs and glaciēs feels missing. In trying to connect glaciēs to gelū, our trust toolkit of rules and concepts struggles to fuse the former onto the established word family to which cold and gelato belong. Experts for the moment seem to have given up on finding any family for the etymologically orphaned glaciēs.

So, that was a stroll through the wider world behind our English word cold. Keep warm out there, and if you celebrate it, I wish you a wonderful Christmas. If you don’t, I wish you a December no less lovely.

END.

If you’d like more etymologising on a winter theme, do have a look at my seasonal posts for 2021 (Rockin’ Around Etymology), 2023 (Christmas Trees and Etymologies) and 2024 (A CHRISTMAS Full of Etymology).

References

• bosworthtoller.com

• De Vaan, M. (2008). Etymological dictionary of Latin and the other Italic languages. Brill.

• oed.com

• Pokorny, J. (1959-1969). Indogermanisches etymologisches Wörterbuch. Francke.

Originally posted on dannybate.com. All images either my own or taken from Wikimedia.

Now, when it comes to any sample of language, I think as a linguist first, as a historian second, and maybe fifth or sixth as a wannabe poet. The primary module of my brain was activated while reading Chapter 5 and its starring verse, the lyrics of the thirteenth-century song Sumer is icumen in. Its language is English, or rather the ancestral Middle English, and its subject matter is very light; its chapter in Clarke’s book explores both the text and the context of this “poem about farting farmyard antics”.

Here are the lyrics:

sumer is icumen in

lhude sing cuccu

groweþ sed

and bloweþ med

and springþ þe wde nu

sing cuccu

awe bleteþ after lomb

lhouþ after calue cu

bulluc sterteþ

bucke uerteþ

murie sing cuccu

cuccu cuccu

wel singes þu cuccu

ne swik þu nauer nu

sing cuccu nu sing cuccu

sing cuccu sing cuccu nu

And now, in the English of today:

Summer has arrived,

Loudly sing, cuckoo!

The seed is growing

And the meadow is blooming,

And the wood springs [into leaf] now.

Sing, cuckoo!

The ewe bleats after her lamb,

The cow lows after her calf;

The bullock prances,

The billy-goat farts,

Merrily sing, cuckoo!

Cuckoo, cuckoo,

You sing well, cuckoo,

Never stop now.

Sing, cuckoo, now; sing, cuckoo;

Sing, cuckoo; sing, cuckoo, now!

A reader’s first impressions and questions provoked by this cacophony of Middle English will sift out the historical linguists from the rest. The musician may wonder how these rambunctious lyrics were performed; the historian may wonder why they were written down and preserved until today. These questions came late to my mind; naturally, it was preoccupied by the H in lhude and lhouþ.

Now I’ve brought your attention to it, you may also admit to wondering what that H is doing there. The words that contain it are translated by H-less Modern English descendants: loudly and lows. The wondering may be compounded by the fair assumption that every letter in medieval texts serves a purpose; in a time when ink and parchment were labour-demanding materials, idle letters were grievous wastage. That is to say, the H was not a silent bonus letter, like the H in hour or B in doubt; it had to contribute something to the encoding of speech.

The nerd that I am, I was already aware of this feature of Middle English. But, like a rare bird for the off-guard ornithologist, it was nice to come across it in the wild. So, what does the H offer the words lhude and lhouþ? These two words belong to a wider pattern of spelling, utilised to spell sounds long since swallowed by the mouths of English speakers.

To understand this H, I will transport your thoughts to the linguistic era that preceded Middle English and the language of Sumer is icumen in. That era was the Old English period. In this pre-Norman tongue, the ancestors of the two words under examination are hlūd and hlōwan. Their H had therefore been there for centuries before the song, but it used to precede the L. Curiouser and curiouser.

An H appears at the start of more Old English words whose present-day descendants lack it. In the HL club, we also find hlēapan, hlot and hlāf – today, leap, lot and loaf. We have hrōf, hring and hrēaw for today’s roof, ring and raw. Old English speakers knew the humble nut as a hnutu, and the neck as a hnecca.

The comparative evidence – harvested from English’s wider Indo-European family tree – leads us to believe that the HL, HR and HR in these old words were once pronounced as a sequence of two sounds. Spelling them with two letters was therefore a reasonable practice.

At an earlier stage, the H likely represented a /x/ sound, as heard in a Scottish loch or a German Buch. This fricative was a product of a serial sound change in the prehistory of English, known as Grimm’s law. English’s unaffected relatives, such as Latin, maintained the pre-Grimm /k/ consonant. Hlūd and hlōwan have Roman cousins in Latin cluēre ‘to be called’ and clāmāre ‘to shout’.

However, as documents in Old English gradually become middle-aged, we find examples of these words without the initial H. Such spelling ‘mistakes’ are indicators, gratefully received by historical linguists, that the sound behind the H was starting to be dropped from the spoken word. This view is propped up by poetry, which wordsmiths constructed at the time through alliteration, making the initial sounds of words match. If the composer of the poem Judith (c. 1000) could alliterate “rēocende hrǣw” (‘reeking corpses’), it suggests that their speech lacked, or at least could lack, an audible sound for the H of hrǣw that would block the poetic pattern.

Yet the H did not go quietly. It fused with the sound of the following L, N or R, or rather donated a phonetic feature to it. It thereby created a trio of new consonants for spoken Old English. These were much like regular L, N and R, except that they were voiceless; they were pronounced without vibration in the larynx. This quality was a legacy of the voiceless H sound. It’s hard to convey in text the subtle difference between regular L, N and R and their unvoiced versions, but a comparable difference exists still over the border from England; Welsh speech distinguishes between sounds spelled L/R/N and those spelled LL/RH/NH.

It is one such voiceless sound that our songster had in mind and mouth two centuries later, when they wrote lhude and lhouþ in Sumer is icumen in. The H was not a relic or a taught rule from Old English writing; the tradition of spelling practices from that time had been severed by the events of 1066. Instead, the LH here reflects a sound still distinct enough from L to deserve its own spelling.

The reversal of the two letters reflects the change to the original sequence of sounds; Middle English authors heaped a heavy burden on H, using it to create pairs of letters (digraphs) that could spell sounds beyond what singleton letters could manage. Some digraphs still in use today (e.g. CH, GH, SH) date to this era of English. Alongside LH, we find RH and NH sporadically in Middle English texts. They consistently appear in words derived from Old English ancestors with HL, HR and HN, demonstrating that this was not memory, but an active difference in speech.

This situation, as you can hear for yourself, was not to last. However much you use the archaic verb low today (depends, I suppose, on how much time you spend around cows), the L at the start of low and also loud now sounds the same as an initial L anywhere else. Likewise, roof, ring, nut and neck have the usual voiced Rs and Ns. This is because the pairs of counterparts, one voiced, the other voiceless, went on to merge in speech. They merged in favour of the more common voiced versions.

When that merger happened, however, is not at all straightforward. Scholars adduce spellings and poetry as evidence that it had already begun in the Old English period. Whether or not this is correct, the instances of LH, RH and NH in Middle English indicate that it was yet to happen or ongoing for some post-1066 speakers.

For example, from 1258, we have a proclamation of Henry III, named “king on Engleneloande, Lhoauerd on Yrloande”. This spelling of lord (etymologically a ‘bread-guard’) harkens back to Old English hlāf. Presumably, this records a southern dialect of Middle English, written in England’s political heartland.

Further north, in Lincolnshire, the delightful Orrm spelled the same word as “laferrd”. The two writers differed not only in geography but also chronology; Orrm was composing Middle English in c. 1180, well before Henry’s proclamation. He does, however, spell roof as “rhof”.

& tanne toc þe deofell himm

inntill þatt hallȝhe chesstre

þatt iss ȝehatenn Ȝerrsalæm

& brohhte himm o þe temmple

& sette himm heȝhe uppo þe rhof

And then took the Devil Him

Into that holy city

That is called Jerusalem,

And brought Him onto the Temple,

And set Him high upon the roof

(Ormulum 11347–11351)

For Lincolnshire language, we can take laferrd and rhof as evidence that the two types of L had merged by that time, but not so with voiced and voiceless R.

Whatever trends towards the ultimate merger we detect, an outlier is the Ayenbite of Inwyt, a curious religious text from Canterbury in 1340. Its translator and composer, the monk Dan Michel of Northgate, consistently uses the conservative LH in “lhord” and “lheste” (lord and listen), as well as NH in “nhote” (nut).

þeruore zayþ oure lhord þet þe kingdom of heuene is hare

Therefore says our Lord that the kingdom of heaven is here

(Ayenbite of Inwyt: The Seventh Bow of Mildness)

The possibility remains that this was poetic affectation, a means to make the language sound archaic and venerable. Yet I fail to see how Dan Michel would have thought that about LH and NH, nor how it would have benefited his work. Instead, it’s simpler to see this as a faithful rendition of Kentish English in the fourteenth century, which still had LH and NH as separate in sound from L and N. If he had travelled to innovative northern regions, Dan Michel’s accent might well have sounded alien and strange, possibly old-fashioned. It would be Caxton’s eggs all over again. In poking our noses into lhude and lhouþ, we gain an appreciation of the depth of diversity across medieval England.

Ultimately, though, Modern English pronunciation owes more to Lincolnshire than to Kent, as it was from the East Midlands that standard speech began to emerge. The doom of the voiceless consonants inherited from Old English was certain: to be lost from the relevant dialects of Middle English before LH, RH and NH could make their mark on our standard spelling. They belong now to the dustbin of writing and speech, whispers of Englishes past.

With one exception.

There is a survivor in this group of consonants, artfully unmentioned so far. Circling back to Old English, we find that initial H not only before L, R and N, but also W.

Words with HW (more commonly written HǷ, with the lost letter wynn) include hwæt, hwæl, hwǣte and hwīt – today, what, whale, wheat and white. These noticeably contain an H. While an H in a word like lord or nut might offend our modern eyes with excessive antiquity, it feels perfectly normal in while, whether and when.

The same process that happened with HL, HR and HN happened with HW. However, the voiceless sound that the fusion of HW created lasted for much longer. It can still be heard today. There are flavours of English speech that still keep whether and weather distinct in their first sound. The former begins with a voiceless /ʍ/; the latter with the more common /w/. These accents, still to be heard in Scotland, Ireland and the USA, are fighting the ongoing ‘wine–whine merger’. My accent succumbed long before I was born, but to keep wine separate from whine was typical talk in southern England up to the eighteenth century.

The slow conquest of the wine–whine merger is a present-day chapter in the story of sounds that began life back in Old English times. The loss of the voiceless LH, RH and NH sounds may be a done deal across English speech, but their embattled comrade, WH, connects us to their era.

When we gain our speech, we inherit not simply stable sounds, but a living, changing system. Those inbuilt changes are linked in a chain of causality that, I suppose, began with the first meaningful grunts. For me, reflecting on the wine–whine merger breathes life back into the language of a historical artefact like Sumer is icumen in, making the language of thirteenth-century folk feel familiar, tangibly like my own.

END.

English spelling and many of the topics mentioned in this piece are discussed in detail in my own book, Why Q Needs U, recently reviewed by The Economist, described as “charming” and “a biography of the English language and an introduction to linguistic science, all smuggled into short, lively and entertaining stories”.

References and resources

• Matyushina, I. (2018). On The Diachronic Analysis of Old English Metre. International Journal of Languages and Linguistics, 5(3), 260-270.

• Upward, C., & Davidson, G. (2011). The History of English Spelling. John Wiley & Sons.

• Ayenbite of Inwyt text

• Middle English Compendium

• oed.com

• The Ormulum text

• The Proclamation of Henry III text

Original Substack post. Images my own or from Wikimedia.

Bonus grammar fact

‘Sumer is icumen in’ is in fact perfective in its tense and aspect; summer is already here. The clause is a relic of a time when English used be as an auxiliary verb that built the perfect tense with verbs of motion, just as French and German still do. English at that time also had a prefix to add to past participles (the i- of icumen) which German again maintains (the ge- of gekommen).

The town straddles a river that shares its name; the picturesque waters of the Sázava in time contribute to those of the Vltava, then the Elbe, and finally the North Sea. Consequently, the river is perhaps better known and better enjoyed than the eponymous town (except perhaps for fans of the Kingdom Come: Deliverance games). Buried in a leafy Bohemian valley, Sázava seems an unlikely spot for historical events as momentous as the title of this essay suggests – but such events have to happen somewhere. My argument is that Sázava was the setting for the culmination of the medieval emergence of ‘the West’, a vaguely defined geopolitical unit that has maintained its relevance until today.

There was once a time when ‘the West’ was not, and I cautiously put forward the idea that its process of coming into being ended in Sázava. Naturally, though, no one knew it at the time. The notion may also surprise the modern Western reader who assigns the Czech Republic to the recent grouping of ‘Eastern Europe’. The flimsiness of such designations is exposed with a glance at the map of Europe, with Sázava standing further west than Vienna and Stockholm.

Allow me to make the case, which I approach first and foremost not as a historian, but as a historical linguist. Yet language both reflects past currents of society, politics and religion, and contributes to them. All four facets of the human experience are closely woven in this tale of Sázava.

In 1096, a group of monks were kicked out (again) from their monastery, on account of being monks the wrong way. This is the kind of deeply medieval debacle that leaves modern souls perplexed and yawning. Their home for a few decades had been Sázava Monastery. This had its origins as a riverside cave, home to the holy hermit Procopius. Such was the fame of this Bohemian recluse (who died in 1053) that he was joined by other hermits. See how the eremitic life, done well, is self-defeating, in that the exemplary hermit soon attracts company.

Their society of solitaries was in 1032 regularised into a monastery, where daily life operated according to the Rule of St Benedict. The monastery can be visited today and is well worth the trip. Its single tower greets guests from a distance, and its unfinished, unroofed and unwalled nave is a skeletal sign of the religious destruction that rocked Bohemia in the fifteenth century. This was when the Czechs, ever the trend-setters, got stuck into their own reformation long before Germany’s Martin Luther wrote a single one of his theses.

Yet it wasn’t the Hussites of the 1400s who rang the death knell for Sázava Monastery. It was finally closed in 1785 by the Austrian emperor Joseph II, a Catholic with a talent for closing down Catholic churches.

Within the strata of Sázava Monastery’s millennium, I want to dig into that 1096 expulsion. By most metrics, Bohemia (today one part of the Czech Republic) sits in the middle of Europe. Sázava sits in the middle of Bohemia. A priori, we expect its history to involve ideas and practices imported from both west and east. This is what we find in the eleventh century. Christianity, originally a Mediterranean religion, had been brought to Bohemia from more than one southern source.

Czechs today celebrate the saintly brothers Cyril and Methodius as the stars of their Christianisation, having been sent by the eastern Roman (‘Byzantine’) emperor in Constantinople in 862. Yet the biographical Life of Cyril suggests that Christianity had arrived already in the area; the local ruler, Rastislav of Great Moravia, had written to the emperor:

“Though our people have rejected paganism and observe Christian law, we do not have a teacher who can explain to us in our language the true Christian faith”

Rastislav’s statement records evidence of missionaries at work among the Bohemians and Moravians, apparently unable to speak directly to the Slavic population. (Their language would in time become Czech and Slovak.) The unnamed forerunners may have been Franks from the powerful Germanic empire to the west. What Rastislav wanted was to host preachers and teachers with knowledge of Slavic speech and a gift for scriptural translation (Cyril and Methodius could provide both), and perhaps also to cosy up to Constantinople.

What the brothers set up was an oven-ready Church for the Slavs, with a calendar of saints and services, and all the essential Biblical texts translated. Given their eastern origins, it may have unintentionally incorporated customs that were the norm in Constantinople, but at odds with the Church in the west. They were working close to the border of two segments of the Christian world: the Church of Constantinople and the Church of Rome.

The latter refers to a ginormous swathe of Europe (and Northern Africa) that nominally owed respect and deference to the Bishop of Rome. It was by far the largest of the five zones of episcopal authority set up in Late Antiquity, and known collectively as the ‘Pentarchy’. According to this arrangement, the five top dogs of Christianity (at least within the borders of the Roman Empire) were the Bishops of Rome, Constantinople, Alexandria, Antioch and Jerusalem. They were in principle all equal in authority, but one or two saw themselves as more equal than the others.

Given their geography, the Czech lands of Bohemia and Moravia were closest to the remit of Rome, but might also feasibly be in touch with the powers-that-were in Constantinople too. Both Rome and Constantinople were looking to win different groups of Slavs over to their own team, and in the process get one over each other. There were no tears yet in the fabric of Christendom, but there were networks of personal loyalties, minor differences in the business of worship, and emerging local customs. These include the questions of unleavened vs. leavened bread for the Eucharist, and whether or not to fast on Saturdays. These may seem so trivial to us today, but when Europe was staking its souls on a singular salvation, it mattered to get things right.

Just as Cyril and Methodius came from Thessaloniki to oust previous would-be apostles, their mission in turn was later kicked out. After Methodius’ death in 885, a new prince, a new pope and newly resurgent Germanic missionaries saw their disciples driven from Great Moravia. They fled to friendly Bulgaria. We see how right from the start, these Czech lands were a wrestling ring for missionaries from the south. They were caught between representatives of the Church of Rome and those loyal to Constantinople. Christendom was more or less still unified at this time, but the rise and fall of the Methodian mission foreshadows the enduring divide between Roman Catholicism and Eastern Orthodoxy.

Language was caught up in these medieval competitions. The arrival of the Slavs in Central and Southern Europe (from the sixth century onwards) introduced a new dimension to the existing linguistic landscape. They had not come equipped with a tradition of writing for their distinct speech, so how were they going to fit into the scripture-grounded religion of ‘the Word’? Would they have to put up with written Greek and Latin? Or could they be given their own script, specific to Slavic language? Cyril and Methodius thought they ought to be, and are credited with the creation of the distinct Glagolitic script.

Although later surpassed by Cyrillic, I see in Glagolitic an attempt to raise up Slavic to the level of a prestigious language. A unique writing script could make Slavic a medium for sacred literature, just as scribes of Greek and Latin, but also of Syriac, Gothic, Armenian, Georgian and Coptic had already achieved.

Yet unlike those existing traditions of translation and writing, Slavic writing was met with mixed reviews and some vociferous opposition. More specifically, it was poorly received elsewhere within the orbit of the Church of Rome. Cyril and Methodius took their new culture of writing and liturgy to Italy. There, the Life of Cyril tells us, it was first met with hostility. In Venice, Slavic was scorned by people who believed that only Latin, Greek and Hebrew were the only acceptable languages of worship and scripture. Cyril and Methodius disagreed, labelling their opponents’ view the “trilingual heresy”. I sense in this debate a fight over the linguistic borders of the classical world, and how to include the language of newcomers.

All was far from unanimous; the Bishop of Rome (the pope) in fact welcomed the brothers and their creation:

“Accepting the Slavic Scriptures, the Pope placed them in the Church of St. Mary called ‘Phatne’. And the holy liturgy was celebrated over them. Then the Pope commanded two bishops, Formosus and Gauderich, to consecrate the Slavic disciples. And when they were consecrated, they at once celebrated the liturgy in the Slavic language in the Church of the Apostle Peter.”

Pope Adrian II (d. 872) therefore gave Slavic writing and worship as good a seal of approval as you could get – or at least, that’s what the Life reports. Different popes had different opinions. Adrian’s successor, John VIII (d. 882), changed his mind on the issue in favour of Methodius, but then Pope Stephen V (d. 891) outright banned translations and religious services in Slavic speech. In 855, he wrote to Svatopluk of Great Moravia, enforcing a prohibition that apparently Methodius himself had previously agreed to:

“… The divine services, sacred rites and solemnities of the Mass, which the same Methodius dared to celebrate in the Slavic language (although he had assured by an oath on the most holy body of the blessed Peter that he would not do so, dreading the guilt of his own perjury), it is henceforth in no way by anyone to be undertaken [in the Slavic language]. By God’s and our apostolic authority, under the bond of anathema, we prohibit it, with the exception of the attainment of edification by the simple and uneducated people…”

Strong words from the servant of the servants of God. In disagreement with his bullish predecessor Adrian, Stephen insisted on the use of Latin in formal religious contexts within his territory. This attitude has shaped the modern world; up until the Roman Catholic reforms of the mid-twentieth century, you would witness only Latin worship within the European remit of Rome, despite it containing a large Slavic population of Czechs, Poles and Slovaks. Slavic sermons, chants and texts took root elsewhere, such as in Bulgaria, Serbia and Russia, where Cyrillic writing also remains the norm.

In the ninth century, we see the Bishops of Rome flexing their muscles and beginning to insist not just on preeminence and respect, but on obedience and uniformity. Religious language was both a factor and a symptom of this attitude. An idea of Western European coherence and conformity was emerging, with Rome at its centre, but it would need a couple more centuries to establish itself.

We can at last rejoin the brothers of Sázava in their Bohemian forest. Their religious life in the eleventh century was a good example of the just-about-still-unified shape of Christianity. Their monastic life was organised according to the template of St Benedict, an Italian, yet they worshipped in the eastern Byzantine style. They also maintained the Slavic language of the Methodian mission in their services and holy texts, apparently unbothered by the ban of Stephen V. The Church designed for the Slavs had clearly managed to put down some roots in Bohemia since the expulsion of Methodius’ disciples.

Great Moravia by this time had fallen, but Bohemia had its own independent dukes who might be interested in continuing its legacy. Duke Oldřich I (d. 1034) was in charge when Sázava was made a monastery, and for a while it was a centre for Slavic writing, learning and worshipping. It even exported Slavic texts and established short-lived links with eastern powers like the Kievan Rus. However, in contrast with its idyllic setting, Sázava sat on a rumbling faultline.

By the end of the eleventh century, the brothers and their Slavic rites had been definitively expelled from Sázava. Latin-leaning monks from Břevnov replaced them. That century saw the papacy in Rome reach new heights, as it recovered from the lows of the tenth century, and revived a fortitude glimpsed in popes like Stephen V. It witnessed the reforms of Gregory VII (d. 1085), who fought to clean up the Church of Rome, confirming its clergy and property as independent from secular powers. This may sound like our modern separation of Church and State, but what Gregory was working towards was both the separation and the elevation of the former. John Julius Norwich describes his tenure in The Popes (2011):

“His strength lay, above all, in the singleness of his purpose. Throughout his life he was guided by one overmastering ideal: the subjection of all Christendom, from the two emperors down, to the authority of the Church of Rome. The Church could make them and unmake them; it could also absolve their subjects from their allegiance. But just as the Church must be supreme upon Earth so too must the Pope be supreme in the Church. He was the judge of all men, himself responsible only to God; his word was not only law, it was the Divine Law.”

Gregory’s mission, eagerly continued by his successors, was centralisation. Across his ancient jurisdiction, established through the Pentarchy, Rome gathered all authority unto itself and its bishop. The pope was no longer just a senior colleague to be respected by other western bishops; now they were his lower-ranked employees. Emperors, kings and dukes also had to submit. As you can imagine, there was considerable backlash to this audacious upgrade and programme of reform; the backlash involved one unbelievable episode in which an emperor stood barefoot outside a castle in January snows, waiting for the pope’s forgiveness. Abject humiliation or imperial one-upmanship? It’s hard to say.

The Gregorian elevation of the papacy was somewhat successful, however, and so Western Europe gained a degree of coherence that it had not had before. There had been nothing structural to unite its immediate post-Roman landscape of Italians, Franks, Bavarians, Slavs, Norsemen, Irishmen, Angles and Saxons. Now, at least in theory and in the minds of Gregory’s allies, Western Europe had a single over-monarch (the pope), a capital city (Rome), and a common language (Latin).

However much Gregory and successive popes managed to make this vision politically real, the reforms did at least bring the religious life of the ‘Latin West’ into the Roman hierarchy. Through this, a stronger warp of ideas, institutions, traditions, philosophies, artistic trends and language could be threaded through the territory. With the Church of Rome as its nervous system, the body of ‘the West’ was now in its infancy.

I fully admit that the concept has varying relevance and reality for people today; it doesn’t occupy my own thoughts very much. Moreover, of course, it has changed almost beyond recognition since the days of Gregory VII, in both its nature and geography; ‘the West’ now includes such westerly locations as Australia. Later reformations and rebellions against papal authority, such as those launched by Hus, Luther, Calvin and Henry VIII, also severed many sinews. But nonetheless, if we trace the genealogy of the concept, it emerges in the eleventh-century consolidation of Roman jurisdiction.

Sázava Monastery was a hole in that Roman tapestry. Along with the notable exception of Croatia (which maintained a unique Church identity in its Balkan base), a community like the one in Sázava was an anomaly. It was a relic of a more catholic and less Catholic Church. Though officially within the Roman orbit, its linguistic and liturgical loyalties lay elsewhere. Through its Methodian pedigree, it was a Constantinopolitan intruder in the emerging West. With the 1096 expulsion of the community from Sázava and Bohemia by Duke Bretislav II, the hole was sewn up. The Czech lands were for the next few centuries a seamless component in the medieval West, albeit not too far from its frayed eastern edge.

Finally, in this socio-politico-religious consolidation of Rome’s jurisdiction, I detect a linguistic shift too. In areas further east, the linguistic story is generally one of plurality. There, as mentioned above, separate writing traditions for separate languages grew up around the separate Churches. Meanwhile, within the West-to-be, Latin held a supremacy that transcended the territory’s many linguistic borders. I insist that this is not a story of unidirectionality and homogeneity; the rich corpora of Old English, Old Norse and Old Irish writing quickly put paid to that notion. A revival of Slavic language, literature and liturgy in Bohemia under Charles IV (d. 1378) also bucks the trend.

Latin nonetheless achieved a supranational prestige, and therefore motivated a change of climate: from a time of linguistic practicality, in which incoming languages had a chance at equal status, to a time of linguistic allegiance, in which Latin was a token language that signalled loyalty and orthodoxy. Rather than equip the many peoples of medieval Western Europe with their own translations and put the language of the Church into their hands, we instead find the centralisation of language, and therefore its restriction to a Roman-right clerical class.

Exclusivity in religious language was nothing new, nor necessarily a bad thing; some religions of the ancient world had successfully traded on esoteric language that gave their practitioners an appearance of expertise and mystique. Inclusivity is an ever-present alternative, though, as people want to participate more fully in their faith. In the case of the Latin West, the bloated weight of Latin would eventually burst as the medieval world became the modern; the Renaissance and Reformation together begot a litter of new languages and literatures. The second of those movements in particular demanded access to the language of liturgy, just as the Slavs of Great Moravia had wanted from Cyril and Methodius.

In the linguistic consolidation at work behind the eradication of sacred Slavic from Sázava in 1096, I spy a glimpse of the end of the Middle Ages, when it would all fall apart.

END.

Original website post. References and sources included as a links within the text. Pictures my own or from Wikimedia. Cover image: icon with four Czech saints, St Procopius on the right.

Back in the ancient days of the Middle Kingdom, the letter S began life as a tooth – or possibly, before that, a bow. Egyptian hieroglyphs may have provided the basis for our letter shapes, but the pictographic founding members of the alphabet were wont to change. We find variation in the sound they denoted, in the direction of their writing, and in what object writers thought they most closely represented.

Following the graphic anarchy of the earliest centuries, our letters are somewhat settled after they are put into the hands of the Phoenicians. Based up the coast from Egypt, these Levantine rulers and traders sent their fairly standardised version of the alphabet off around the Mediterranean. It included the letter 𐤔‎, called shin. It stood for a sibilant sound, plausibly the /ʃ/ in shoe and shine.

There must have still been some optionality in the stance of the letter. While the Phoenicians liked to write with a horizontal 𐤔‎ (hence, Hebrew ש and Arabic س today), the Greeks accepted and popularised a vertical version. Shin became the Greek letter Σ.

There was also at first tremendous variety in how exactly the Greeks shaped the letter – sometimes curvy, sometimes jagged, sometimes with four strokes, sometimes with three. With this acquisition, they could write down the sound /s/ (as in Sparta and Salamis). The Greeks called this letter sigma. This is an unexpected name, apparently taken from the wrong Phoenician letter, namely sāmek (𐤎). Sigma sailed off to Italy, beaching on its southern shores in around the year 700 BC. It would in time simplify into a recognisable Roman S. The rest, as they say, is history.

Yet the importing of the alphabet didn’t remove it from where it had just come from. Back in Greece, the alphabet was booming, and sigma was to develop further. No sooner had the letter achieved its iconic, Ionic four-lined form of Σ in general public use, then its shape began to change in private writing. From the fifth century BC onwards, we see the emergence of a simplified sigma: Ϲ. Consisting of only a single curve, this new shape would have been quicker to write in casual communication. We refer to this character as ‘lunate’ – a moon-like letter.

It would take a few more centuries for Ϲ to oust Σ from formal contexts, but Ϲ was the dominant shape of sigma over the subsequent medieval era. We find it in Greek Christian writing, such as in the oldest manuscript copies of the New Testament. You might come across today on the golden icons in Orthodox churches, on which Ἰησοῦς Χριστός (‘Jesus Christ’) is abbreviated to ΙϹ ΧϹ. It may look like an English C, but its alphabetic cousin is in fact S.

The English rendition of the alphabet gets its lower-case letters from developments in writing in Late Antiquity and the Middle Ages. Likewise, the Greek alphabet went through a similar process of creating a minuscule style of writing. This in turn produced the lower-case letters used by Greek today, and lunate sigma provided the basis for lower-case sigma. Specifically, this is a looped-up σ at the start or in the middle of a word, and ς with a final flourish at a word’s end. For its capital letters, modern Greek writing returned to antiquity, and resurrected the old Σ. It was lunate sigma that travelled with the Greek alphabet southwards, and later northwards.

First, having much changed from the old hieroglyphic bow, Ϲ returned to Egypt. It was specifically a post-classical rendition of the Greek script, containing lunate Ϲ, that arrived. Following the effortless conquest of Egypt by Alexander the Great in 332 BC1, the kingdom of the Nile had come under the rule of a Greek dynasty: the Ptolemies. The dynasty’s last member was arguably its most famous (Cleopatra VII). While the Egyptian language kept its prestige and usage, Greek became widely spoken too, especially in northern cities like Alexandria.2

By the third century AD, Greek letters were even being used to write down the Egyptian language. Specifically, this was the final stage of Egyptian before it died out as a spoken language. This we call Coptic.3 Written Coptic is recognisably based on the Greek alphabet, and from a brief glance at Coptic texts, we see how the imported system uses lunate Ϲ for the sound /s/.

That said, Coptic speech had sounds that the Greek alphabet could not spell, so there are some Egyptian additions alongside Ϲ.

The solution to the phonetic need was to borrow some letters from Demotic, a cursive style that in fact directly derived from Egyptian hieroglyphs. The additions include Ϣ and Ϩ, frequent interlopers among the Greekish letters. There they all sit together on the pages of our Coptic texts, Greek and Demotic together, two distant cousins reunited.

Six centuries later, the journey of S took another turn, when the Greek alphabet again provided the basis for the writing system of another language – and in time, languages plural. This development was the Cyrillic script, which has been more ‘successful’ than the Coptic; it provided the basis for the modern alphabets of Bulgarian, Kazakh, Russian, Serbian, Tajik and Ukrainian.

In Cyrillic, the letter for /s/ is the lunate С. This is to be expected, since the script developed out of a medieval Greek milieu. In more recent times, you might have seen this Cyrillic letter in the initials CCCP. This was the official abbreviation for the Soviet Union, rendered in Latin letters as USSR. Although they resemble our C and P, the two letters in CCCP correspond instead to S and R in the English alphabet. Here they refer to the Union (Sojuz) of Soviet (Sovetskich) Socialist (Socialističeskich) Republics (Respublik).

Besides /s/, many languages that use the Cyrillic script also have the consonant /ʃ/. Like English, they have both of the two sibilant sounds in sin and shin. The latter, however, was not a sound that was present in spoken Latin and Ancient Greek. We can only wonder how Romans and Greeks shushed one another. The absence of /ʃ/ in these ancient languages is reflected in their alphabets; neither had a letter for the sound. English’s solution has been to combine two letters together into the digraph SH, but Cyrillic comes with a single dedicated letter: Ш.

Known as sha, the letter Ш is a survivor. It in fact predates Cyrillic, and was a part of the older Glagolitic system. This was the original alphabet created to spell medieval Slavic languages by St Cyril (c. 826–869 AD). To put it briefly, Glagolitic looks utterly different from any scripts around at that time. Its alphabetic order begins ⰀⰁⰃⰄ, and it includes very loopy letters like Ⰿ and Ⱆ. We have plenty of documents from that first era of written Slavic. Their Glagolitic letters give them the appearance of being strange and alien, but that look of unfamiliarity masks a medieval language that Slavic-language speakers today can still understand without much difficulty.

This great difference from existing scripts may well have been deliberate. Cyril and his fraternal colleague Methodius wanted to set Slavic up as a proper language of Christian worship and scripture, on a par with Latin, Greek and Hebrew. These three all had their own distinct alphabets. For Slavic to join this exclusive group, I suppose its champions felt that Slavic had to have its own writing system. The work of Cyril and Methodius pushed back against what they saw as a ‘trilingual heresy’ – namely, that the worship of God could only happen in one of those three languages. The creation of Glagolitic was part of an effort to expand beyond the languages of antiquity, and therefore the current linguistic borders of Christianity.

However, Glagolitic was just that little bit too different. While it did remain in use until the modern era in certain Church circles, it was soon replaced as an everyday script by Cyrillic. Confusingly, later Cyrillic was not the invention of Cyril, but was rather named in his honour. It was comfortably Greek in origin; if you match Cyrillic and Greek letters up against each other, the genealogy is obvious. It was therefore a much more familiar, learnable and perhaps prestigious system for writers in the ninth century. Practicality won out over Slavic pride. However, Cyrillic faced the same problem as Coptic had done; being Greek-based, it could not spell all the sounds of Slavic. The solution was to raid the Glagolitic treasury for desirable letters, and one such gem was Ш.

But where had Cyril got Ш from? While Glagolitic was clearly meant to be a new and distinct script, there nonetheless seem to be (conscious or unconscious) influences from other writing systems. The three-pointed shape of Ш is reminiscent of a letter mentioned above: the equivalent Hebrew letter ש, called shin. It’s a likely source of inspiration, as Ш and ש can stand for the same sound, and Cyril was in fact well acquainted with the Hebrew language.

We are told by the Life of Cyril that he had previously travelled to what is today Ukraine, where he debated with Jews and Muslims in a three-way competition to convert the powerful kingdom of the Khazars. That Turkic nation had written to the emperor in Constantinople, asking that a representative for Christianity be sent to them. Cyril was the man for the job, and it was during this work trip that he learned Hebrew.4 His familiarity with Jews and their holy texts may have later provided the inspiration for his new Glagolitic letter Ш.

Today, Ш coinhabits the Cyrillic script together with lunar Ϲ, its possible distant relative.

Cyrillic and Coptic are two examples of the complexities of the alphabet’s family that all stems from Egyptian roots. Like two overlapping branches, they make use of written symbols that may share a symbol or script of common origin, although their innovators could not have known it. The story of the alphabet is full of linguistic and geographic twists and turns, rather like the curves of an S.

Text adapted from surplus material of Chapter S of my book Why Q Needs U, available here: https://geni.us/WhyQNeedsU

Images taken from Wikimedia or my own. Original Substack post.

‘Every letter?’ you may incredulously query, and fairly so. We have a sense that some are worse troublemakers than others, the better behaved members of the alphabet, like X and Z. Yet, if we rendezvous to watch the upcoming grand prix, even those two fall silent. ‘But those are French words!’ you object. ‘Yes they are,’ I riposte, ‘and now they’re English too.’

Quickly then, silent letters take us out into murky, blending waters, at the confluence of two languages. There we are confronted by questions concerning the permeable borders of a language – at what point adopted words can be said to belong to the recipient. Moreover, as a linguist, I cannot deny that the presence of silent letters in written English is a complication and source of frustration, but I also invite the scoffers to reflect on what exactly silence in spelling may mean.

Take that final word there, mean. Operate on it with your mind’s scalpel. It comprises four letters: M, E, A, N. The first, second and fourth clearly merit their inclusion. Yet the pronunciation of the word bears no acoustic resemblance to the possible sounds of the letter A, to neither the A in ate nor the A in at. Would you then label the A in mean a silent letter? I guess you wouldn’t, and neither would I. We recognise that the E and A function together, spelling a vowel sound found also in meat, meal and meagre. The technical term for such cooperative pairs is a ‘digraph’.

Standard written English relies greatly on digraphs; the letter H is especially imbued with combinatory power. We utilise H to spell the CH, SH and TH sounds (e.g. chin, shin, thin) that were absent from the mouths of the Romans, and therefore from the rendition of the alphabet that they left us. We recognise that the H in a word like their, in tandem with T, pulls its weight much more than it does in heir.

But what about what? The difference between a plain W and the H-augmented digraph in what, whale and white reflects a diminishing difference in speech, as fewer and fewer speakers pronounce whine as separate from wine. Must we wait for their dialects to die out or submit to the merger, before we designate the H in what as yet another silent letter? Even if so, the Y in mayor may remain a further point of disagreement within English, being audible for many (most?) Americans, yet mostly mute for me and my fellow Brits. Silent letters invite us to examine the fragmented state of English speech.

If collaboration with another alphabetic colleague rescues a letter from redundancy, a load more letters evade the charge of idleness. A final E is a common sight at the close of many English words, like cape, kite, code and cute. It would be remiss of us, however, to eject E from these words, lest we be left with cap, kit, cod and cut. This characteristic practice of English spelling, which I was taught as ‘Magic E’, again involves two cooperative letters that represent the single intended vowel sound – meaning that, in a strange way, you can hear the ‘silent’ letter E. This is another example of a digraph, only this time with the two ingredients split from each other in the sequence. No one said it couldn’t work that way.

Since we have extended our defence to written words in which a ‘silent’ constituent ensures a particular pronunciation, we are compelled to hear the plea of similar cases. The GH within everyday words like night seems at a glance like an egregious waste of ink. It’s a relic of a lost sound, dropped from speech during English’s gradual transition from medieval to modern. Over the North Sea, the loss hasn’t occurred in its sister languages, like Dutch (nacht) and German (Nacht). Yet the English GH didn’t go quietly. Its absence left a void that the vowel could fill, which lengthened ever so slightly to bridge the gap. It came to sound like ‘neet’. This new length then rendered the vowel prey to the Great Vowel Shift (c. 1400–1700 AD), through which we arrive at my southern-England pronunciation of night. Consequently, the GH in the word night is in fact responsible for the sound of its letter I. Without the GH, we’d now be left with nit, just as light and fight would be lit and fit. Being jointly to blame, the I and GH work as one to spell a single vowel – a trigraph!

We’re now a far cry from the ‘one sound for one letter’ rule of the alphabet’s inventors, living four millennia ago. Trigraphs are both offenses to that rule and the conclusion of it. Amidst the endless dance of shifting speech, writing can stand still and continue to spell the final result. They are especially hard to rethink when the division of labour seems equal; my one candidate for a quiet Q is the rare word lacquer, but in its trigraph CQU, is there a primary speller of the underlying consonant /k/? Is it the C or the pair QU, or does the whole trio function as one seamless whole? Such a trigraph tests our perception of which sound matches up to which letter. The QU does at least save the C from softness (subtract it and read: ‘lacer’), while ‘laquer’ without the C might be read with a different vowel, as if it started like lake. Both C and QU then contribute to the pronunciation of this French acquisition; can either then be called silent?

But with a common word like night, couldn’t we at least swap the trigraph IGH for a split digraph, with I and Magic E? Wouldn’t that tidy up English spelling a bit? Yes we could, and yes it would. A word of warning, though: we might throw the odd baby out with the bath water, as this would finalise the merger of sight with site, and right with rite. These homophones are for the moment still separate in spelling, which brings us to another use for mute letters; those silent in speech may yet scream at the speedy reader not to mistake their word for another. If this function, as a signpost towards a writer’s right meaning, gives silent letters a free pass, then heck, even the K in knot, knight and know (not not, night and now) deserves to stay.

Spare a thought then for English spelling’s silent letters, many of them conscripted assistants that are there to make other letters shine. Some may still earn their keep in the medium of writing alone, guiding the reader’s eye to a quick recognition of meaning. That said, a few really have no place receiving the same grateful thanks due to the alphabet’s workhorses, the Es and Hs, that keep the engine running. I would have to agree with that frustrated German – the B in debt and doubt needs to go.

This post was inspired by communication between me and James McConnachie, who recently and kindly reviewed my book, Why Q Needs U, for The Sunday Times. The review can be found here.

Original Substack post, cover image taken from Wikimedia.