Etymologies of some English kinship terms

Abbreviations: PIE (Proto-Indo-European), PGmc (Proto-Germanic), OE (Old English), ME (Middle English), NE (New English, i.e. modern English).

father
From OE fæder. The sequence /-dər/ regularly became /-ðər/ after stressed vowels in early ME, which is why we have father rather than fader. The development of the stressed vowel, however, is irregular. The expected development would be into /aː/ by ME open syllable lengthening and then into /ej/ (as in face) by the Great Vowel Shift. However, in this word only the first change seems to have occured, so that the stressed vowel in father is the same as the one in palm and spa, rather than the one in face. In British English, the words rather (< OE hraþor) and lather (< OE lēaþor) have similarly resisted the Great Vowel Shift, although in American English the vowel in these words has been shortened, as it has in all dialects in gather (< OE gadorian) (slather has /a/, but is of unknown origin); that is, these words seem to have resisted the ME open syllable lengthening in the first place. Of course, we can’t rule out the alternatively possibility that they were lengthened and then subsequently shortened, perhaps as part of the same round of shortenings that resulted in short vowels in words like bread and blood. I don’t actually know of any words in -ather where both of the expected changes have taken place, so that the word is pronounced with /-ejðər/.

As for OE fæder, this is a regular development from PGmc *fadēr. And fadēr, in turn, is a regular development from PIE *ph2tḗr.

mother
From OE mōdor. See above on the change of -d- to -th-. The development of the stressed vowel is again irregular. The expected development would be for it to remain as /oː/ in ME and then to develop into /uː/ by the Great Vowel Shift. In this word, however, we have /ʌ/, the usual outcome of ME short /u/. The same outcome exists in brother (< OE brōþor), other (< OE ōþer) and smother (< OE smorian, with -th- inserted perhaps due to influence from the agentive form of the verb, *smorþor ‘suffocator’). There are a couple of words (blood and flood) where ME /oː/ become /uː/ by the Great Vowel Shift, but was subsequently shortened and changed into /ʌ/. However, this change was irregular; for example, it didn’t occur in food. Perhaps the same shortening occured in mother, brother, other and smother after the Great Vowel Shift. I can’t think of any words in -other which aren’t pronounced with /-ʌðər/, so perhaps the shortening was regular in this environment.

As for OE mōdor, this is from PGmc *mōdēr. The changes of unstressed vowels from PGmc to OE are very complicated and I don’t understand them very well. But I don’t know why *mōdēr became mōdor rather than mōder. Perhaps the preceding heavy syllable had something to do with it? PGmc *brōþēr and *duhtēr, with a heavy initial syllable, became brōþor and dōhtor respectively, while PGmc *swēster, also with a heavy initial syllable, had variants in both -er and -or, but PGmc *fadēr, with a light initial syllable, became fæder. The vowel -o- was regularly inserted in OE before postconsonantal r at the end of a word (c.f. OE wundor ‘wonder’ < PGmc *wundrą), so if the -e- in *mōder was dropped due to the preceding syllable being heavy, that would explain it.

PGmc *mōdēr, in turn, is from PIE *máh2tēr. The expected development of this word would be *mōþēr, but the accent appears to have been shifted to the suffix at some point, perhaps due to analogy with *ph2tḗr ‘father’ and *dʰugh2tḗr ‘daughter’, so that -d- occurs by Verner’s Law. Sanskrit mātā́ ‘mother’ shows accentuation on the suffix as well; it is on the basis of Ancient Greek mḗtēr ‘mother’ that accent on the initial syllable is reconstructed.

brother
From OE brōþor. See above on the pronunciation of OE ō as /ʌ/ in NE. OE brōþor is from PGmc *brōþēr. See above on the change of unstressed ē into o. And PGmc *brōþēr is a regular development from PIE *bʰráh2tēr.
sister
From OE sweostor. Variants of the OE word in -er and in swi-, swy- or swu- are also attested. The quality of the final unstressed vowel would not make any difference to the NE reflex, because all OE unstressed vowels merged into /ə/ in ME. However, variants in sweo- would have ended up as NE swester (rhymes with fester), and variants in swu- would have ended up as NE suster (rhymes with muster). So the modern form of the word must originate from a variant in swi- or swy-.

The PGmc form of the word was *swestēr, and the regular development of this in OE would have been *swester, or sweostor if the -e- was dropped due to the preceding syllable, as conjectured above. The other variants probably resulted from a combination of influence of the -w- on the following vowel (c.f. OE wudu ‘wood’ from PGmc *widuz) and influence of the Old Norse form of the word, systir.

PGmc *swestēr is from PIE *swésōr. The expected development of PIE *swésōr would be *swesōr (which would become OE *sweosor and NE *sweaser /swiːzər/), but presumably the word was influenced in PGmc by *bʰráh2tēr ‘brother’, *ph2tḗr ‘father’, *máh2tēr ‘mother’ and *dʰugh2tḗr ‘daughter’, which all ended in *-tḗr or *-tēr in PIE.

son
A regular development from OE sunu. ME open syllable lengthening did sometimes affect short /i/ and /u/ (c.f. week < OE wicu) but not usually. OE sunu, in turn, is a regular development from PGmc *sunuz. PGmc *sunuz is from PIE *suHnús. The expected development of PIE *suHnús would be *sūnus (which would become OE *sūns and NE *sunse /sʌns/). Apparently the accent was retracted to the initial syllable, so that the *-s became *-z by Verner’s Law, and the long vowel was shortened. The shortening of the long vowel also occurs in Italic and Celtic (those branches give no evidence with regards to the accent). Ringe (2006) attributes the change to “morphological resegmentations or reanalyses which yielded roots without a final laryngeal (or its reflex)”, which isn’t very enlightening. The reconstruction with the long vowel and final-syllable accent is based on Sanskrit sūnú.
daughter
A regular development from OE dōhtor, via ME doughter /dowxtər/; although ME /ow/ normally became NE /ow/ (as in boat), it regularly became /ɔː/ (as in thought) before /x/. OE dōhtor is from PGmc duhtēr. For some reason, the vowel was lowered to o and lengthened. The lowering apparently also occured in every other Germanic language, and the lengthening happened Old Norse as well (c.f. Gothic dauhtar /dɔxtar/, Old Norse dóttir, Old High German tohtar). But I have no idea why either change occured. As for the development of the unstressed vowel from ē to o, see above. PGmc duhtēr is from PIE *dʰugh2tḗr. The expected development of PIE *dʰugh2tḗr in PIE would be either *dukþēr or *dukuþēr, depending on whether you believe that interconsonantal laryngeals in non-initial syllables developed into zero or *u in PGmc (opinions differ). The form we actually see is the result of interference from the oblique stem of the word, *dʰuktr̥-ˊ, which became PGmc *duhtur-.
uncle
From Old French oncle ‘uncle’. Old English had two words for ‘uncle’: fædera ‘father’s brother’ and ēam ‘mother’s brother’. The regular developments of these words into NE would have been fathere (rhymes with gather, due to trisyllabic laxing) and eam (rhymes with beam), respectively.
aunt
From Old French ante ‘aunt’. In southern England, Australia (presumably South Africa and New Zealand as well?), New England and Virginia the word is pronounced with /aː/, while in other areas of the world it is pronounced with /a/. I don’t know why this is the case. I also don’t know why it is spelled with au-.

Old English had two words for ‘aunt’: faþu ‘father’s sister’ and modrige ‘mother’s sister’. The regular developments of these words into NE would have been fathe (rhymes with lathe) and mothery (pronounced in the same way as the derived word meaning ‘like a mother’), respectively.

nephew
From Old French neveu ‘nephew, grandson’. The word was originally spelt nevew, and pronounced accordingly. The origin of the spelling with -ph- is kind of a mystery, but perhaps it was due to influence from Latin nepōs ‘nephew, grandson’. A spelling pronunciation with /f/ subsequently emerged and became predominant in American English. The pronunciation with /f/ is now usual in British English as well, although some old-fashioned speakers still pronounce it with /v/.

The native word, neve ‘nephew, grandson’ (rhymes with Eve), survived into ME but is now obsolete. This word is a regular development of OE nefa. OE nefa is a regular development of PGmc *nefô. PGmc *nefô is from PIE *népōts. The expected development of PIE *népōts would be *nefōþs (probably, although I’m not aware of any final *-ts clusters which survived into PGmc), but the noun came to be declined in the same way as *gumô ‘man’, and the ending in the nom. sg. was changed accordingly.

niece
From Old French nece ‘niece, granddaughter’. The native word, nift ‘nephew, granddaughter’ (rhymes with lift), survived into ME but is now obsolete. This word is a regular development of OE nift. OE nift is a regular development of PGmc *niftiz. There does not appear to be a securely reconstructable feminine counterpart of PIE *népōts, although the PGmc form would go back to a PIE form *néptis, and the same form would give Latin neptis.
Advertisements

One response to “Etymologies of some English kinship terms

  1. David Marjanović

    PGmc *swestēr is from PIE *swésōr. The expected development of PIE *swésōr would be *swesōr […], but presumably the word was influenced in PGmc by […], which all ended in *-tḗr or *-tēr in PIE.

    Either that, and/or *-str- was generalized into the nominative from cases with zero-grade in the second syllable: *str is the regular PGmc outcome of PIE *sr in Grimm positions. (In Verner positions it’s apparently just *r with lengthening of the preceding vowel, presumably via *zr.)

    The expected development of PIE *suHnús would be *sūnus […]. Apparently the accent was retracted to the initial syllable, so that the *-s became *-z by Verner’s Law,

    AFAIK, word-final PIE *s always became PGmc *z regardless of where the stress was.

    and the long vowel was shortened. The shortening of the long vowel also occurs in Italic and Celtic (those branches give no evidence with regards to the accent).

    Dybo’s law: long vowels (original or from a lost laryngeal) are shortened before the PIE stressed syllable in, precisely, Germanic and Italo-Celtic. I have no idea why Ringe (2006) didn’t mention it (most of his examples for Osthoff’s law fit Dybo’s as well, and his exceptions don’t…); perhaps he hadn’t read enough Russian literature, or perhaps he didn’t like the (weak) implication that Germanic and Italo-Celtic are sister-groups.

    The expected development of PIE *dʰugh2tḗr in PIE would be either *dukþēr or *dukuþēr, depending on whether you believe that interconsonantal laryngeals in non-initial syllables developed into zero or *u in PGmc (opinions differ).

    Interesting, I didn’t know that *u had been proposed. It’s suspicious, though: *u was the default vowel for epenthesis, used to resolve the PIE syllabic sonorants as well as (somewhat randomly) other consonant clusters all the way to OHG miluh “milk”. So, finding *u in a place where a laryngeal had been earlier probably means that the *u is epenthetic.

    aunt
    […] I also don’t know why it is spelled with au-.

    It was very common in ME to render French an as aun. I don’t know why that was done, but it was at least regular (or close enough, I don’t know).

    I’m not aware of any final *-ts clusters which survived into PGmc

    Me neither. Quite possibly they surfaced as *-s already in PIE.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s