Blog 18 – Secrets of the Words You Know

Blog 18 *s; some secrets of ‘some’, and linguistic analysis from over 800 years ago (Box 8)

Now we turn from the sonorants of PIE to its one fricative (if we exclude the laryngeals), the sound *s

*s is a very frequent initial sound in PIE roots. But it also has a unique feature, in that for many words an initial ‘s’ will be found in a good number of Indo-European languages, while others have words clearly derived from an exactly similar root without the initial *s. We have already seen examples under both *m with the root underlying ‘remember’ and under *n with the root underlying ‘nerve’. The term ‘mobile s’ is used to describe such cases.

Latin shows no examples of ‘s’ before nasals and liquids, and Greek is similar except for a few words beginning with ‘sm’.

We will start with words that unambiguously appear to have had an initial *s in PIE (or where the *s appears to have been preceded by a laryngeal), and then discuss separately cases of an initial ‘mobile s’, and the reason why *s seems to have been particularly prone to this treatment.

‘Unambiguous’ *s at the start of a word

There are many basic words that begin with an initial *s followed by a vowel in PIE, as evidenced by a large measure of identity across daughter languages. Examples include the following:

Salt Saline

Seven September

Sister Sorority

Sit Sedentary

In each of these cases, the inherited English word is on the left and a borrowing from Latin on the right. In many other Indo-European languages, words with the same or very similar meanings also start with ‘s’.

At least three Indo-European languages, Persian, Welsh and Greek, however replace initial *s before a vowel with ‘h’. This is why we have words like:

Related to ‘salt’: Greek-based ‘halogen’ (literally ‘salt-producing’: chemical elements that produce salts when reacting with sodium) and ‘halcyon’ (literally ‘salt-dog’, the word ‘salt’ here used poetically to mean the sea). A ‘sea-dog’ was the Greek expression for a fabled bird like a kingfisher which was said to make a floating nest in the sea in calm midwinter days. So a ‘halcyon day’ is a day of wonderful unexpected calm. We will return in later blogs to the second elements of both of these words.

Box 8, below, gives an insightful piece of analysis of Latin and Greek initial ‘s’ and ‘h’, respectively, set down over 800 years ago. Reading this, it is remarkable that such clues were not followed up before the revelations provided to Western scholars by Sanskrit.

Box 8

Giraldus Cambrensis on ‘Salt’

The Welsh cleric and antiquary, Giraldus Cambrensis (died around 1223), makes an interesting comment on the initial consonant of this and other ‘s/h’words, showing early awareness of relationships across languages.

He writes: ‘Salt is ‘hals’ in Greek and ‘halen in Welsh’ (which he then attributes to the Britons staying in Greece after leaving their mythical homeland in Troy). He continues, ‘It seems remarkable to me that I do not find so many languages agree as much over any other word as they do in this: ‘hals’ in Greek, ‘halen’ in Welsh, ‘halgein’ in Irish, where g is inserted, and ‘sal’ in Latin, where, as Priscian tells us, s replaces the aspirate in some words. Just as ‘hals’ corresponds to ‘sal’ in Latin, so ‘hemi’ [Greek] is ‘semi’ [Latin] and ‘hepta’ is ‘septem’ [‘hepta’ and ‘septem’ are the Greek and Latin for the number 7]. In French the word becomes ‘sel’, the vowel a changing to e as it develops from Latin. In English a ‘t’ is added to make ‘salt’, and in German the word is ‘sout’. In short, you have seven languages, or even eight, which agree completely over this word.’

It’s particularly noteworthy that Giraldus recognises the wider ‘equivalence’ of Greek initial ‘h’ and Latin ‘s’. However, Giraldus seems to have misconceived the Irish word, which is in fact ‘salann’, conserving the PIE initial *s.

Thanks to my sister Rosalind Bolton for bringing Giraldus’ comment to my attention.

Related to ‘sit’: ‘cathedral’, which has the sense of ‘the bishop’s seat’, ultimately comes from a Greek word for a seat, made up of the elements ‘kata’, meaning ‘down’ (a ‘catastrophe’ is literally a ‘turning down’, as mentioned in Technical Note 4) and ‘hedra’, meaning seat or chair, showing Greek ‘h’ for the original ‘s’ which English has conserved. Our word ‘chair’ is borrowed from a French cut-down version of this word, in which the original PIE root *sed- is next to invisible. The Greek word ‘hedra’ is also the heart of the set of words that define straight-sided shapes, such as ‘tetrahedron’, as it also carries the sense of a flat surface.
Greek-derived ‘herpes’ and Latin-derived ‘serpent’, both from a PIE root *serp- meaning ‘to crawl’.
Greek-derived ‘hedonist’ and Latin-derived ‘suave’, both from a PIE stem *sweHdu-, ‘sweet’, the English word inherited from the same root.
Greek-derived ‘helium’ and Latin-derived ‘solar’, both from a PIE root *seHw with a suffix in -el-/-en-, referring to the sun (which English has inherited from the -en- version of the suffix).

A longer shot, but an agreeably paradoxical one, is a possible match between Greek-derived ‘hero’ and Latin-derived ‘servant’, both potentially from the PIE root *ser- meaning to protect (as in our Latin borrowings like ‘pre-serve’). The ‘hero’ could easily be a ‘protector’ of comrades in battle, and it is possible that the ‘servant’ was originally also a much less highly-regarded protector of livestock (say a shepherd), with the word then coming to mean more generally someone who provides a rather menial service, and, in Latin, a slave.

There’s a particularly nice example of the influence of the change in Persian of *s>h. The River Indus flows to the Indian Ocean through the Pakistan province of Sindh. In Sanskrit, which conserved *s, the river as well as the terrain was termed ‘Sindh’, but when the Persians invaded the area, they adapted the river name to their characteristic ‘h’. The Greeks then picked up the river-name as ‘Indus’, losing the ‘h’, and the area through which it ran as ‘India’, a term then used to describe the whole sub-continent.

We have noted elsewhere examples of surprising changes in meaning. A striking one that begins with *s is from a root that has the sense of ‘satisfy’ (a borrowing from the Latin-derived version, as is ‘satisfactory’). The inherited English doublet of these words is ‘sad’ – seemingly perverse, though when you think about it, it is an understandable result of being ‘satiated’ (another borrowing of the same Latin element).

*s also frequently appears before a consonant in PIE and all Indo-European languages. As we saw in Technical Note 2 with ‘spit’ and ‘spew’, both the initial *s and the immediately following consonant tend to be stable. In such cases, Greek maintains the initial *s, so we can see equivalences like inherited ‘stick’ and ‘stitch’, Latin-derived ‘instigate’ and Greek-derived ‘stigma’, all from a PIE root *steiĝ- meaning ‘to prick’.

One of the most widespread PIE roots is *steH-, with the meaning of inherited ‘stand’. This gives English any number of Latin-derived words from ‘constant’ to ‘stay’, ‘status’, ‘statue’, ‘station’ and ‘stable’, Greek-derived ‘stasis’ and ‘system’, and the (Persian-origin) ‘-stan’ endings of (Latin-origin) states – places where one stands – like Afghanistan. Further derivatives include inherited ‘steer’ and the firmest of the metals ‘steel’.

‘To sew’ also goes back to a PIE root *syuH-, which gives us ‘suture’ (from Latin), ‘Sutra’ from Sanskrit, a written work that preserves important teachings, from a word meaning ‘thread’, and the Greek skewer ‘souvlaki’, which is in fact a borrowing from a Latin word for an awl (used to make holes to sew leather).

We can add Celtic and Slavic matches by considering the PIE root *sterH- ‘to spread’, from which English has inherited ‘strew’ and ‘straw’. Matches include the Gaelic term for a broad valley, strath as in Strathclyde, Latin borrowings like ‘structure’ and instruct’ and the Russian term for ‘restructuring’, ‘perestroika’. ‘Street’ (from the Latin ‘via strata’ or ‘paved way’) and ‘Strategy’ (from Greek) are also from this prolific root.

‘Mobile *s’ at the start of a word

We now turn to examples of an initial ‘mobile s’, which is visible in words with and without the initial *s. Such examples are all before consonants (*k, *l, *m, *n, *p, *t and *w), never before vowels.

Examples where English has inherited the form without *s but where borrowings from other Indo-European languages derive from a form with *s include:

‘Thatch’ versus Greek-derived ‘stegosaurus’, from a root *(s)teg-, meaning ‘to cover’. The Greek word means ‘tiled lizard’ as a description of the creature’s heavy plated skin. Here Latin, like English, conserves the form without *s in our direct borrowing ‘toga’ for a covering garment. The same is true for its word for ‘roof tile’, ‘tegula’, literally ‘a little toga’, the ‘l’ suffix signalling a diminutive. French has lost the ‘g’ before the ‘l’, giving the French ‘tuile’. This is seen in the ‘Tuileries’ (the name of a now destroyed royal palace in Paris, which was on the site of a tile works) and is the origin of the English word ‘tile’. The Indian word that gives us ‘thug’ is possibly from Sanskrit ‘sthagayati’ (he/she covers) from the same root. Another form of covering is a deck, a word from the same root, but borrowed into English from Middle Dutch, with the expected equivalence of ‘d’ for an English ‘th’ that we mentioned in Blog post 9 on English and German.
‘Foam’ versus Latin-derived ‘spume’ (that p/f combination again).

Examples where it is English that inherits the form with *s but has also borrowed words from languages that have inherited the form without *s include the many reflexes of the root *(s)ker-, meaning ‘to cut’. This gives English words like ‘shear’, ‘sharp’ and ‘short’ (and as we have seen also ‘shirt’) and visibly Old Norse borrowings like ‘score’ and ‘skirt’ which show that these forms were common to Proto-Germanic. Latin, however, shows the form without *s in borrowings like ‘carnal’, ‘carnivorous’, ‘carrion’ (all going back to a word meaning ‘a piece of flesh’, so containing the idea of cutting), and ‘curt’ with the same suffix as ‘short’.

Other similar examples involving English and Latin are ‘slack’ versus Latin-derived ‘lax’ and ‘slop’ versus Latin-derived ‘lubricate’.

There are also cases where a language group has inherited forms both with and without the initial *s.

Within Germanic, for example, the root *(s)keu-, with a core meaning of ‘to cover or conceal’, is visible in inherited English ‘hide’ and ‘hoard, the evolution of *k>h being entirely normal, as we will see later. The same root, but with the initial *s, is also clearly visible in Old Norse borrowing ‘sky’ (which ‘covers’ the landscape) and later Dutch borrowing ‘scum’.

And English itself has inherited both forms in ‘needle’ and ‘snood’ from the root *(s)neH- meaning ‘to twist, bind or spin’, which as we saw under *n also lies, with a suffixed *u, behind borrowed ‘nerve’ and ‘neuro-‘. A borrowing from Greek from the same root is the thread-like worm, the ‘nematode’.

A similar English example is ‘slime’ and the slimy substance ‘lime’, from a PIE root *(s)lei-.

Why do we see so many examples of roots with and without *s? It is not altogether clear, but it is thought that it may be along the lines of our earlier discussion of false word-division leading to an ‘n’ being sometimes wrongly added or lost. The fact that ‘s’ is an easy sound to combine with following consonants offers a wide choice of situations where this could have happened.

So the sound-code for English s is:

Initial s in an inherited word = Proto Germanic *s= PIE *s = Latin s = Greek h = Irish s = Welsh h = Balto-Slavic s = Iranian h = Indic s

In some cases English initial s is matched by zero in other languages (‘mobile s’) and vice versa.

No other PIE sound yields English inherited initial s.

(ii) *s elsewhere

The combination *rs between vowels develops in Latin to ‘rr, as in ‘terrain’ and ‘territory’, from a root *ters-, which had the original sense of ‘to dry’ (so giving the sense of ‘dry land’ and eventually just ‘land’ or ‘earth’ in Latin). It is also the origin of our inherited words ‘thirst’ and ‘thirsty’, which as you can see conserve the *rs.

Between vowels, *s undergoes changes in some languages.

In Greek, *s is lost between vowels: we will see an example under *dh in a later blog.
In Latin, as we saw under *w in the word ‘virus’ from PIE root *wis-, *s routinely changes to an ‘r’ between vowels. That’s why a sisterly gathering is sometimes termed a ‘sorority’, based on the Latin word for ‘sister’. It’s also why we have from Latin from the root *Heus- (‘to shine’, used particularly of the dawn and so of the East, but in Latin also of the South wind) not just the word that underlies the Southern continent ‘Australia’ with the *s conserved before a consonant, but also the word ‘aurora’, which in Latin referred to the goddess of the dawn, where *s has changed to ‘r’ between vowels. The same root gives us our inherited words ‘East’ and ‘Easter’, and also ‘Austria’, a Latinized version of the German ‘Öster-reich’ or ‘Eastern Kingdom’.
Proto Germanic similarly changes an inherited *s between vowels to ‘r’ in certain conditions, as we have already seen under *r when we pointed out the alternation between ‘was’ and ‘were’. It turns out that the ‘r’ of ‘were’ descends from a ‘z’, which in turn goes back to the same ‘s’ that has survived unchanged in ‘was’. But why did ‘s’ develop into its voiced counterpart ‘z’ in ‘were’ but stayed unchanged in ‘was’? The answer to that question will take us to one of the defining changes that mark out Germanic languages from their Indo-European cousins. We will explore this further in a later blog post.

(iii) Some secrets of ‘some’

An interesting group of words stems from the PIE root *sem-, whose basic meaning seems to be ’one’. This is the origin of the inherited words ‘some’ and ‘same’, Latin-derived words like ‘simultaneous’, ‘sempiternal’, and the many Greek-derived words starting with ‘homo-’ and meaning ‘same’, as in ‘homogeneous’. (Interestingly, Greek-derived ‘hetero-, meaning roughly ‘one of two’ and so also ‘different’, contains the same root, this time in its zero-grade form *sm̥, which gives ‘he-‘ [originally *ha-])

A Greek equivalent needing more explanation is Adelphi, the plural of the Greek for ‘brother’. The word literally means ‘same womb’. The ‘a’ is again from the zero-grade*sm̥ (though this time the expected ‘h’ has been lost, as happened in some Greek dialects), and the ‘delphos’ (as in the place name for the Delphic Oracle) means ‘womb’. A dolphin (also from Greek) is so named from what was perceived as its womb-like shape, or possibly because it gives birth to live young.

The combination of a zero grade *sm̥ and the root *pel- (to fold – yet another example of the f/p relationship), gives Latin-derived ‘simple’ and Greek-derived words in ‘haplo-’. A form with a suffix with *l gives us borrowings from Latin like ‘similar’, and a combination with a root *ker- (‘to grow’) lies behind another such borrowing ‘sincere’, literally ‘of one growth’.

The root *sem- also turns up in Sanskrit – literally. ‘Sanskrit’ is so named as it was considered the ‘perfected’, or more precisely ‘made together’, style of language. The ‘krit’ is a past participle (hence the ‘t’-ending) from a root * kʷer-, meaning ‘to make’ that we will meet again in a later post, while ‘San’ derives from *sem-, with the sense of ‘together’ (ie in one place).

Also from this root are two words imported from Russian

‘samovar’ is literally a ‘self-boiler’ – although it may be a ‘folk-etymology’ (Boxes 3 and 6 in Technical Note 3) based on a similar-sounding Tatar word for a tea urn.
‘samizdat’ literature is literally ‘self-publishing’.

(iv) What’s an inherited English word from another PIE root beginning with *s?

While English maintains both *sp- and *st-, we saw in Blog post 5 that words in *sk- change in English to ‘sh’ as a result of fronting. For a PIE root commencing with*sk-, we can take *skeHi with a suffixed -*d-, which appears to have the basic meaning of ‘to cut or split’ (though in this case I admit this won’t help you much), and appears in borrowings from Latin like ‘rescind’ (with infixed ‘n’) and from Greek like ‘schizophrenia’.

What might we expect in the inherited English descendant of the root? Hint: it has no infixed ‘n’ but it also shows the dental suffix, and it has four letters….