uncouth thane

 

One would expect Greek _phílos_ 'beloved', as a zero-grade passive adjective, to be oxytone.  The simplest explanation of its recessive accent is that it was commonly used as a form of address and very frequently occurred in the vocative.  On the basis of Vedic usage, one can infer that PIE vocatives had no accent unless they began an utterance, in which case their accent was recessive.  Classical Greek does not preserve this rule, but it is highly plausible that Proto-Greek had nom. sg. masc. *philós, fem. *philá: against voc. sg. masc. *phíle, fem. *phíla (cf. _númpha_, the fossilized voc. sg. of _númphe:_).  When the recessive-vocative rule was lost, most oxytone thematic nouns adjusted their voc. sg. to fit the nom. sg. and other cases, but 'beloved' was so commonly used in the vocative that the other cases were adjusted to it.

 

Similar considerations apply to _téknon_ 'child', in origin a passive adjective 'begotten' parallel in formation to _semnós_ 'awful, holy, solemn' from PIE *tjegW-nó- 'revered'.  That is, PIE *tek-nó- 'begotten' should have yielded Proto-Greek nom. sg. masc. *teknós, fem. *tekná: against voc. sg. masc. *tékne, fem. *tékna 'offspring, child'.  Again, when the recessive-vocative rule was lost for thematics, the high frequency of the vocative would result in new nom. sgs. *téknos, *tékna: with matching accent.  Later in Proto-Greek, when terms of endearment became neuter (like English _sweet thing_ and Terence's _Glycerium_, non-speaking heroine of the Andria, based on a comedy by Menander), _téknon_ 'child' displaced the old masc. and fem. forms, with the voc. sg. taking the same form as the nom./acc. sg.  The dying Caesar is reported to have uttered "kaì sú, téknon" to Brutus (quoting from another play by Menander).

 

I do not understand A.L. Sihler's insistence that PIE *g^neh3- 'to know' had no zero grade, and substituted full grade for it in the /tó/-deverbative and /sk^/-present (New Comp. Gr. of Grk. & Lat., New York/Oxford 1995, §52.3, p. 49).  Greek _gno:tós_ 'recognizable, known' and _gignó:sko:_ (Epirote _gnó:sko:_) 'I (come to) know' are compatible with zero grade, provided one assumes that PIE *-n.h3- regularly yielded Grk. -no:-, in parallel with the other word-internal sequences of syllabic resonants plus *h3, to wit:  PIE *str.h3-tó- 'scattered, spread'

> Grk. _stro:tós_, PIE *ml.h3-sk^oh2 'I come' > Grk. _bló:sko:_, PIE

*sm.-h3nogWH- 'single-hooved' > Homeric nom. pl. _mó:nukhes_ (NCG §§106-7, pp. 103-5).  Celtic *gna:to- 'known' (Old Irish _gna:th_, Gaulish _-gna:tos_) is also compatible with zero grade, while Lithuanian _pa-z^ìntas_ 'known' demands it.  Sanskrit _jña:tá-_ 'known' need not reflect inherited full grade, but can represent zero-grade *ja:tá- with new anlaut imported from full-grade forms of _ja:ná:ti_ 'knows':  pass. _jña:yáte_, opt. _jña:ya:t_, perf. _jajñá:u_, etc., thus avoiding homophony with _ja:tá-_ 'born'.  Latin _(g)no:tus_ 'known' and _(g)no:sco:_ 'I come to know' could easily have acquired -o:- from the full-grade stative _(g)no:vi:_ 'I know', thus also avoiding homophony with _(g)na:tus_ 'born' and _(g)na:scor_ 'I am born' from the zero grade of *g^enh1- 'to beget'.

 

In Germanic one would expect PIE *g^n.h3-tó- to yield *kunDa- by Grimm's Law, prevocalization of the syllabic resonant, loss of the following laryngeal in this position, and Verner's Law.  Instead Gmc. *kunþa- 'known' is indicated by Gothic _kunþs_, Old English _cu:þ_, and Old High German _kund_.  As with _phílos_ and _téknon_, the simplest explanation is a high-frequency vocative, commonly occurring in utterance-initial position when members of the same community addressed each other.  I presume that the inputs of this word to Verner's Law were oxytone in all cases but the vocative, and that the vowel-mergers *o > *a, *a: > *o: occurred at a later stage of Proto-Germanic than Verner's Law.

 

Thus, immediately before Verner's Law, Proto-Gmc. 'known' had nom. sg. masc. *kunþós, fem. *kunþá:, nom. pl. masc. *kunþó:s, fem. *kunþá:s against voc. sg. masc. *kúnþe, fem. *kúnþa, voc. pl. masc. *kúnþo:s, fem. *kúnþa:s.  Immediately after Verner's Law, the nom. forms were *kunDós, *kunDás, *kunDó:s, *kunDá:s against voc. *kúnþe, *kúnþa, *kúnþo:s, *kúnþa:s.  Later the PGmc accent became fixed on the root, leading to a paradigm with a vocative stem *kunþ- against *kunD- used in the other cases.  With most originally oxytone nouns and adjectives subject to Verner's alternation, the seldom-used vocative stem was replaced by the other stem.  But the voc. stem *kunþ- not only had a relatively high frequency; it also had a tradition of respect in formal address.  It would have been UNCOUTH to address one's kinsman as *kunDe rather than *kunþe.  As a result the voc. stem *kunþ- prevailed over *kunD-, not the other way round.

 

The parallel PIE deverbative *g^n.h1-tó- yielded the expected Gmc. *kunDa- 'begotten, born, child' reflected in Go. _himina-kunds_ 'heaven-born', OE _heofon-kund_ 'id.', and Old Norse _kundr_ 'son'.  But ON _a:s-kunnr_ 'descended from the Asen' requires Gmc. *kunþa-, which I believe is to be explained somewhat like the accent of Grk. _téknon_.  Immediately after Verner's Law, the various forms of the simplex 'born' were homophonous with the corresponding forms of 'known', with context indicating which was meant.  The voc. *kúnþe 'O known one, kinsman' was used in polite address to adults of equal status, while *kúnþe 'O begotten one, son' was used in familiar address to children or, figuratively, to adults of lesser status.  After the voc. stem *kunþ- had ousted *kunD- from the paradigm of 'known', *kunD- had the upper hand in 'begotten, born' since it involved no ambiguity whatever.  Yet the vocs. *kúnþe, *kúnþa 'O child' persisted long enough in familiar settings to produce an alternate paradigm *kunþa- 'child, descendant', and this is what is preserved in ON _a:s-kunnr_.

 

Grk. _téknon_ cannot plausibly be separated from Gmc. *þeGna- 'vassal, warrior, freeman, thane'.  It will not do to invent a new PIE root *teg(^)H- attested only in Gmc., for this would violate a root restriction and it is too short for *-g(^)H- to be a root extension.  This creates difficulty for the usual view of Kluge's Law occurring during the Gmc. Lautverschiebung.  From PIE *teknó- one would expect Gmc. *þekka-, and from vocatives in *tékn- only Gmc. *þexn-.  The simplest solution is to backdate Kluge's Law to a stage before the Lautverschiebung, presuming that the geminated tenues produced by the former were unaffected by the latter.

 

Apart from Kluge's Law, this presumption about geminated tenues surviving the Lautverschiebung provides a plausible etymology of Go. _attekan_ 'to touch' (st. VII, pret. 1/3sg. _attaitok_).  Most scholars take the simplex _tekan_ as primary, but forms of _attekan_ are thrice as numerous as those of _tekan_ in the remnants of the Gothic Bible.  Assuming that Gothic inherited the simplex from Gmc. *te:kanaN has led to bizarre results.  D. Ringe proposes a "post-PIE *deh1g- ~ *dh1g- 'touch'" (apparently "post-PIE" is free of root restrictions) from which he derives both Go. _tekan_ and ON _taka_ 'to take' (From PIE to PGmc, Oxford/New York 2006, p. 80).  F. Kortlandt proposes no new root, but a unique series of ad-hoc developments in order to get _tekan_ and _taka_ from PIE *teh2g-, which is reflected in Grk. _tetagó:n_ 'having seized', and according to him also in Latin _tetigi:_ 'I have touched' (Old Norse _taka_, Gothic _tekan_, Greek _tetagó:n_, NOWELE 36:59-65, 2000).  Starting with the reduplicated "aorist" stem *teth2g-, he loses the laryngeal to get *tetg-, assimilates it to *tedg-, and then back-assimilates the reduplication to *dedg-.  Then he invokes his preglottalic theory of PIE plain mediae to write the stem as ['de'd'g-].  This is then simplified by loss of dental articulation in the cluster, yielding ['de''g-] (which he does not explicitly write, avoiding the explanation of how one is supposed to articulate a geminated glottal stop).  Finally, assuming that *h1 was still a glottal stop, he rewrites the stem as *deh1g-, which of course becomes *de:g-, and then *te:k- by the Lautverschiebung.  This PGmc "aorist" stem is then supposed to have been altered in North Gmc. to *to:k- in order to conform to Class VI, thus yielding ON _tók_ 'took' and contaminating *þakanaN to *takanaN (ON _taka_) with its anlaut.  Gothic on the other hand allegedly created a new present by adding present endings to the "aorist" stem *te:k-, and formed a new reduplicated "perfect" stem *teto:k-, putting the verb into Class VII.  On the third hand West Gmc. supposedly created sundry weak verbs from *tak-, the stem resulting from contamination of the regular zero grade *þak- (reflecting PIE *th2g-) with the "aorist" stem *te:k-, the latter conveniently vanishing before it could be attested in WGmc.

 

I find Kortlandt's algebraic manipulation of phones and morphological sleight of hand overwhelmingly unconvincing.  Too many ad-hoc phenomena must go exactly right at exactly the right time.  Moreover, both his and Ringe's derivations suffer from the highly implausible semantic development 'touch' > 'seize, take' without affixation or intensive formation.  But if one assumes that Gothic inherited the PREFIXED stem *atte:k-, continuing pre-Grimm's Law *atte:g-, and this reflecting *ad- (PIE *h2ed-) plus *te:g- (PIE *teh1g-) 'to touch' with trivial sandhi, neither a new root nor a convoluted mess of ad-hoc machinations is necessary.  Go. _attekan_ in this view is cognate with its semantic match, Lat. _tango:_ (generalized zero-grade nasal pres. *th1-n-g-), _tetigi:_ (gen. zero-grade perf. *te-th1g-).  Go. _attaitok_ continues the gen. /o/-grade perf. *te-tóh1g- with analogical -t- replacing the *-d- expected by Verner's Law (as in _saiso_ 'sowed' for *sezo:, ON _sera_) after Gmc. reanalysis of the stem as *at-te:k-.  This reanalysis led to extraction of the Go. simplex forms.  OE _þaccian_ and Old Low German _thak(k)olo:n_ 'to touch lightly, pat, stroke' are probably based on a Gmc. fem. abstract *þakko: 'touch', PIE *th1g-néh2.  But the prefixed verb would have produced its own Gmc. abstract *attakko:, with reanalysis yielding *takko: 'touch'.  This alternative noun is probably the base of Middle LG _tacken_ 'to touch'.

 

Grk. _tetagó:n_ requires a different PIE root *teh2g- 'to seize', and ON _taka_ in my opinion is best explained through the /w/-extension (attested in several IE branches) of *deh3-, originally 'to take' (as in Hittite), not 'to give' as generally elsewhere in IE, where the semantic shift 'take' > 'take (for someone else)' > 'give (to someone else)' occurred.  Development of Gmc. *-k(W)- from PIE *-h3w- was probably pretonic only, that is pre-Grimm's Law *-h3w-´ > *-gW-´, but *´-h3w- > *´-w-.  Thus 'living' involved PIE masc. sg. nom. *gWih3wós, voc. *gWíh3we > pre-GL nom. *gWigWós, voc. *gWíwe > post-GL nom. *kWikWós, voc. *kWíwe.  East Gmc. created a new paradigm from the vocative stem *kWiwo- > *kWiwa- > Go. masc. sg. nom. _qius_, acc. _qiwana_, etc.  West and North Gmc. used the other stem *kWikWo- > *kWikWa-.

 

This digression on _attekan_ justifies placing Kluge's Law before Grimm's.  Now if 'begotten, child' underwent Kluge's assimilation, the masc. sg. nom. *teknós and voc. *tékne would become nom. *tekkós and voc. *tékne.  After the episode of assimilation passed into history, the seldom-used vocative of most words with such a contrast would be analogically levelled to agree with the other cases.  But as in Proto-Greek, I hypothesize that the frequently-used vocative influenced the other cases of this particular word.  This time the accentual pattern was not affected, but the consonantism of the vocative was levelled to the other cases, restoring nom. *teknós beside voc. *tékne.  Thus the high-frequency vocative allowed this word to make an end run around Kluge's Law, so to speak.  Grimm's Law produced nom. *þexnós and voc. *þéxne, and Verner's Law gave nom. *þeGnós against voc. *þéxne.  But by now the noun was no longer used primarily in its original sense 'child, offspring'.  Its principal use had passed, perhaps through 'son of a nobleman', into its historical significance of 'vassal, warrior, freeman, thane'.  It was no longer a term of endearment of parents toward children, and had presumably been replaced as such by *kunDós, *kúnþe (from *g^enh1-, not *g^neh3-).  Thus the vocative *þéxne occurred far less frequently than it had formerly, and it was levelled into *þéGne after the other cases.  When the PGmc accent was later generalized to the root, the only stem in use was *þeGna-.

 

Douglas G. Kilday