Blog post 25 PIE Stop Consonants *dh and the secrets of dough
Voiced Aspirate: *dh
(i) *dh at the beginning of a word
Just as with *bh, Sanskrit and its daughter languages are alone in conserving the PIE aspirate, which is lost in Germanic and in most other Indo-European groups. And again, Latin and later Greek conserve the sound as a fricative.
In Greek, this is the expected ‘th’, but in Latin *dh yields not ‘th’ but ‘f’ – just as in some English dialects such as Cockney: ‘Fings ain’t wot they used to be’. As a result, an initial ‘f’ in a Latin-derived word may represent either PIE *bh or *dh, and so match either a ‘b’ or a ‘d’ in an English word inherited from PIE. As we will see shortly, a Latin initial ‘f’ may also represent *gʷh, so this is an important (and unusual) example of a single initial consonant potentially representing three original types of articulation: labial, dental or labio-velar.
The PIE root *dhwor-, whose meaning is shown by the inherited English word ‘door’, provides an example of how initial *dh has evolved in various languages.
As you can see, English shows the initial ‘d’ that Grimm’s Law predicts.
The Greek for ‘door’ begins with the expected ‘th’, although the only borrowing into English is, as often with Greek words, a medical term. Your ‘thyroid’ is so named because the celebrated Greek physician, Galen, termed the Adam’s Apple as ‘shield-shaped’ (you may remember from *w in Blog post 16, that the ‘-oid’ ending means ‘-shaped’). This word was then applied to the thyroid gland. The Greek for shield was derived from their word for ‘door’ (‘thura’), because of the tall narrow shape of the typical Greek shield.
There are similar words in many Indo-European languages, almost all conserving the meaning ‘door’, including Persian ‘darbar’, literally the ‘door of the Court’, widely used in princely India for state assemblies, with the typical English spelling of ‘durbar’. As expected, Persian has lost the original aspirate.
Latin however used the unrelated word from which we derive ‘portal’ as its term for a door, and the *dhwor- root was maintained only in a term meaning ‘outdoors’. This begins, as we would expect, with ‘f’. This word in turn formed the basis of several words subsequently borrowed into English, from the still Latin-sounding ‘Forum’ (originally an outdoor market place) to ‘foreign’ – from a land ‘outside’ – and, more surprisingly, ‘forest’, which originally appears to have meant land outside the boundary of say parkland or cultivated fields.
Box 9: False Analogy
If the ‘g’ of ‘foreign’ looks surprising, that’s because this is one of several examples where English spelling fails to represent the etymologically correct spelling of the imported word. This is often the result of false analogy with other words thought to be relevant.
In this case, the Middle English borrowing via Old French from late Latin ‘foranus’ (foreigner) was ‘forein’. However at a later date the spelling was adjusted to that used today, perhaps reflecting the word ‘peregrinus’, which as we saw in Box 7 did indeed also mean ‘foreigner’ or ‘stranger’, or other words ending in ‘-eign’
It’s very similar in spirit and consequences to ‘folk etymology’ (Boxes 3 and 6), but perpetrated by those setting ‘standards’ in spelling. The final ‘b’ of ‘thumb’ and ‘crumb’ are two other examples of incorrect additions, based on words like ‘comb’ where a final ‘b’ made historical if not current sense.
In Technical Note 2 we explained that Latin ‘deus’ was the doublet of Greek ‘Zeus’ rather than of ‘theos’ (which like ‘deus’ is the general word for a god). Now it’s time to ask what is the Latin word that is a doublet of ‘theos’.
The Greek initial ‘th’ represents in this case a PIE *dh, so we need to look for a ‘religious’ word in Latin beginning with ‘f’. Fortunately the words in question are well-represented in English borrowings: the set of words about festivals and feasts, also visible in later borrowings from French (fête) and Spanish (‘fiesta’).
You will see that all these words imply that the PIE root would have included an *s (the French circumflex is a regular mark of an earlier ‘s’). In fact it is *dheHs-, or *dhēs-.
Greek, as mentioned under *s (Blog post 18), loses *s between vowels, which is why we speak of ‘theology’ rather that of ‘*thesology’. Under *s we also noted that in Latin, *s will shift to ‘r’ between vowels, which in turn explains the ‘r’ in our word ‘fair’ in the sense of an outdoor festive occasion, also from the same root, but without the dental suffix of the other Latin-derived words. What is now a final consonant in English would have fallen between vowels in Latin, as for example in the word that gives us the ‘Ferial’ responses sung in some English cathedrals.
The same root, with a nasal suffix and loss of the *s before the *n provides English with two words derived from the Latin for ‘temple’ or ‘shrine’, namely ‘fanatic’ and ‘profane’ (literally ‘outside the shrine’, and so ‘unholy’.
While this root has supplied a good crop of borrowed words, there is no English word directly-inherited from it.
The root *dheH-, meaning ‘to put’, is the origin of a phenomenal number of borrowings by English. This is not surprising, given that the core meaning can be developed in so many directions.
English has inherited from this root the words ‘do’, ‘deed’ and also, with an *m-suffix, ‘doom’, the latter from the sense of something ‘laid down’ in law.
In Latin, it also gives the word ‘to do’, which starts with the expected ‘f’, and appears in English borrowings both with its own *k-suffix in ‘fact’, ‘factotum’ (literally ‘do everything’, so a jack of all trades), ‘defect’, ‘efficient’ and the like, but also in a much more cut-down version visible in borrowed words ending in ‘-fy’ like ‘gratify’ and so many others.
Greek, which on the whole builds on the meaning ‘to put’, also contributes many words from the same root from ‘thesis’ (and ‘antithesis’) and ‘theme’ (with the same m-suffix as ‘doom’) to ‘apothecary’ – shopkeepers ‘put out’ their wares (the preposition/preverb ‘apo’ means ‘away from’).
The same root provides the second element of the Persian word for curtain, which as we saw under *p is literally something ‘put forward’, ‘purdah’.
Similarly, this root is thought to lie behind the second element of the composite root *wi-dheweH-, which gives English ‘widow’ with many similar words across Indo-European. As noted under *w in Blog post 16 , the first element seems to mean ‘asunder’.
We will see other examples of *dheH- as the second element of a word both later in this section and under *k̂ and*gʷ.
A nice example of the similar, but not simultaneous, development of *dh in German and Celtic is provided by a stem *dhū-no- (‘fort’ or perhaps, as we saw with*bherĝh- under *bh, ‘high place’), which lies behind the many Celtic place names with the element ‘dun’ from Dunquin in Ireland to Dunfermline or Dumbarton (the original ‘n’ assimilated to the following ‘b’ in this ‘fort of the Britons’ [of Strathclyde] in Scotland), Verdun in France (‘fort of men’: you may remember that the first element of ‘Fergus’, which we saw under *w, means ‘man’), and the less obvious Maiden Castle in England (the first element means ‘big’ (as in Welsh ‘mawr’) and ‘den’ was originally ‘dun’ before Folk Etymology intervened).
The same stem underlies English inherited ‘downs’ for a range of hills, showing a similar development of *dh to ‘d’. And it also, more surprisingly, underlies the word ‘town’. The argument here is that Germanic borrowed the Celtic version, which would already have lost the aspirate, before the operation of Grimm’s Law. Thus, at that stage, Germanic would have had both an inherited form of the root beginning with *dh- (perhaps with the sense of a high place) and a borrowing from Celtic beginning with *d- (perhaps with the sense of a fortified settlement). After the operation of Grimm’s Law, the first would have developed the initial ‘d’ of ‘downs’ and the second the initial ‘t’ of ‘town’.
Finally, let’s take the root *dher-, which means to hold firmly, or to support.
It’s not evident in inherited words in English, but with an *-m- suffix it is the origin of the Latin adjective which gives us ‘firm’, related words like ‘confirm’ ‘infirmary and also ‘farm’, via the use of the term for a ‘firm’ or ‘stable’ lease or rent, and subsequently transferred to the land so rented.
In Greek, you would expect a word beginning with ‘th’, no doubt with an ‘r’ as the next consonant. And indeed that is what we have – a throne. (In Greek, the word is more generically one for ‘chair’, though also used for the seat of a dignitary.)
Continuing East, King Darius of Persia’s name meant ‘holding firm the good’. Persian has also given the same ‘dar’ element to two important titles in India: ‘Sirdar’, used for a high-ranking military commander, literally meaning ‘head-holder’, and ‘zamindar’ a ‘land-holder’.
In Hindi, the initial aspirated consonant is preserved, for example in ‘dharma’, the term for firmly established law or practice, so also ‘duty’. The same root lies behind the second element of the well-known Indian name Chowdhury or similar spellings. This is a title which means literally ‘holder of four’ (the first element from Sanskrit ‘catur’ meaning ‘four’), and was given to community headmen and the like.
The sound code for English initial ‘d’ can be set out as:
English initial d in an inherited word = PIE *dh = Latin f = Greek th = Indic dh = Celtic, Balto-Slavic and Persian d = d in other Germanic languages except German t
No other PIE initial sound yields English ‘d’
(ii) *dh in the middle or end of a word
In medial position, Latin normally shows ‘d’, as indeed the word ‘medial’ shows (the ‘d’ goes back to a PIE *dh). Another case is the classic dog-name ‘Fido’, from the Latin ‘I trust’, which goes back to the root *bheidh-, which also gives us English ‘bide’.
However, when next to ‘u’ or ‘r’, or before ‘l’, Latin changes medial *dh to ‘b’, giving us another example of how a dental may change to a labial. This explains, among other things, why a man might have his beard trimmed by a barber. The PIE root from which ‘beard’ is derived is *bhardh- (as seen also in the Russian surname ‘Borodin’). Because the *r is adjacent to the *dh, Latin changes *dh>b.
But wait! You might reasonably object that the initial *bh should give Latin ‘f’. Indeed it should: logically we should be going to the ‘*Farber’.
It’s not always possible to offer definitive explanations for sound changes for which we have no documentary evidence. This is an example. There are two possible explanations, one involving assimilation, and one dissimilation:
- Assimilation: once the *dh of *bhardh- had developed to ‘b’, Latin could have simply assimilated the initial consonant to that sound, by losing the aspiration;
- Dissimilation: Latin could have lost the aspirate of the first consonant of *bhardh- through the same kind of dissimilation that we saw in ‘Buddha’ under *bh. In this case, the resulting initial ‘b’ would have been automatically conserved. However, this is not a routine change in Latin, and therefore seems the less likely of the two.
The same phonetic context also explains the difference in final consonants between ‘word’ (inherited) and ‘verb’ (borrowed from Latin), which we mentioned in Blog post 15 (*r). This consonant goes back to the root *dheH-, which we have just been discussing, so when added to the root *werH-, meaning ‘to speak’, a ‘word’ is something like a ‘doing of speech’. Since the *dh is next to the *r in Latin, it shows as ‘b’ in ‘verb’ rather than as the ‘d’ which is conserved in English ‘word’.
It also explains why the (inherited) red gem is a (Latin-derived) ruby. Here the PIE root is *Hrudhr-, so that the *dh is next to both a *u and an *r. The same reason explains ‘rubella’ and rubicund’, likewise borrowed from Latin, versus inherited ‘ruddy’. This still fails to explain one more Latin borrowing, the name ‘Rufus’, with the sense of ‘red-haired’. Here it seems we have an example of borrowing by Latin from another Italic dialect which changed *bh>f between vowels.
Greek conserves medial *dh as a fricative, though English borrowings with the meaning ‘red’ are somewhat obscure medical terms like ‘erythema’. A better-known example from the same root is the name of the country along the Red Sea, ‘Eritrea’, though this is an Italianised version of a Greek original, which would have displayed ‘th’ rather than the Italian ‘t’. As in several earlier cases, the initial vowel in the Greek word for ‘red’ reminds us of an initial PIE laryngeal.
At the end of the word, we can expect *dh to give English ‘d’, as in ‘beard’ or, as we have already seen, ‘bind’ and its fellow related words from *bhendh-.
(iii) A dairymaid in Paradise: the secrets of ‘dough’
The word ‘dough’ goes back to a PIE root *dheiĝh-, which means to form or to build.
The maid doing the kneading was known in Old English a ‘dæge’, also from this root (‘a doughie’), a word that is hiding behind the girl’s other place of work, the ‘dairy’ (with an Anglo-French ‘-erie’ – ‘place of’ – tacked onto it). But the lady of the house was also kneading: the word ‘lady’ – the second element again from this root – means ‘loaf-kneader’, just as ‘lord’ is an abbreviation of ‘loaf-warder’.
You will expect any Latin equivalents to begin with an ‘f’, and indeed they do, from figure to figment and fiction (something made-up), and more surprisingly, via French, also ‘faint’, all from words about making or fabricating something. ‘Faint’ may have acquired its meaning from the sense of something artificial, and so sham, or (of a person) indolent.
More exotically, the second element of ‘paradise’ is from the Persian for a wall, also from this root, then borrowed by Greek and passed on through Latin and French to English. The word literally means ‘a wall around’, the first element being the Persian equivalent of the Greek preposition ‘peri-’ as in ‘periphery’. As Persian is a ‘satəm’ language, the PIE palatal *ĝh of the root evolves to a Persian sibilant, showing as ‘s’ in the Greek.
(iv) What’s an inherited English word from another PIE root beginning with *dh?
We will select the root *dhīgʷ– ‘to stick, or fix’. It is the source of our many borrowings from Latin of versions of ‘fix’ (‘affix, ‘suffix’ etc). Can you come up with a couple of English inherited words from this root? Two hints:
- As we shall see shortly under *gʷh, in West Germanic, including therefore English, ‘w’ is lost after non-initial velars (including the velar element of an inherited labiovelar).
- One of the options shows palatalization of the final consonant.
The answer offers three choices: two nouns and a verb. As usual, it will be sent to Subscribers next week