Thoughts on the etymologies of enn and hinn in Nordic

Abstract

This contribution focuses on Stroh-Wollin’s (2020 in NJL) etymologies of the Nordic definite articles enn and hinn and contrastive hinn/hitt. While I do not contest her central claim that Old Icelandic enn and Mainland Scandinavian hinn have separate historical origins, I do argue that her etymologies should not be accepted over more conventional ones already present in the literature. First, the etymology of enn should, along traditional lines, be connected to Germanic cognates such as Gothic jain-, German jen-, and English yon (rather than derived from an ancient PIE *eno-).¹ Furthermore, contrastive hinn/hitt and definite hinn/hit should be considered a doublet, both ultimately deriving from a distal/contrastive element (rather than the article having separate origins in an innovated Proto-Nordic proximal demonstrative).

1. A recent take on enn and hinn

In recent work, Stroh-Wollin (2016, 2020) puts forth a detailed picture of the history and development of the definite articles enn and (h)inn in older stages of Nordic. Using data from early Norwegian and Icelandic manuscripts, Stroh-Wollin (see especially 2020:215 Table 1) argues that the preadjectival article in Old Icelandic was enn, to be distinguished from preadjectival hinn and suffixed -(h)inn ² found on the mainland. According to her, the insular form enn later developed into -enn and -inn (with vowel i due to lack of stress), in turn resulting in preposed inn and hinn, forms which happened to converge with the article on the mainland (see Stroh-Wollin 2020:219–220).

Table 1. PIE *áno- and *nó-, and their cognates in Indo-European languages, from Dunkel (2014a:247, 2014b:55–58, 386). ^a Reconstructions in parentheses are based on attestations in only one IE dialect; see especially Melchert (2009) on the Hittite deictic elements

^a Forms which Dunkel explicitly rejects are, for obvious reasons, not included in the table.

Demonstrating that there is a real synchronic division between insular enn and mainland hinn is a true service to the field, but when it comes to the etymologies offered by Stroh-Wollin (2020), there are reasons to be skeptical. She posits, on the one hand, that the Old Icelandic article enn is a Nordic retention of PIE *enos ‘that’ (see Table 1 for relevant forms). On the other hand, she claims that the Old Mainland Nordic article hinn/hit developed from a Proto-Nordic proximal demonstrative, PN * ‘this’. This proximal demonstrative is claimed to be distinct from the formally identical PN * ‘the other/former’, which gave rise to the contrastive hinn/hitt/hint ‘the other/former’ found throughout Nordic. Both forms – the proximal and the contrastive – would have had their ultimate origins in PIE *i- ∼ *ei- (see Section 3 for cognates) plus *eno-, ³ but supposedly they emerged at different times: the proximal in Proto-Nordic or Common Nordic, and the contrastive in pre-Germanic.

In this short contribution I will argue that Stroh-Wollin’s reasons for positing these unconventional etymologies are not strong enough to justify abandoning more traditional views, which are perfectly sufficient to explain the origins of enn (discussed in Section 2) and hinn (discussed in Section 3).

2. The etymology of enn

OI enn is commonly associated with Go. jains, OHG jen-, OE geon, forms which go back to PG *jaina- ∼ *jena- ∼ *jana ‘that, the other’ (on which see Hoffmann-Krayer 1897, Cercignani 1984:322–327, Klingenschmitt 1987:178–182, among others). The handbook version of this complex etymology is that *jaina- is needed to account for Go. jains ⁴ and MHG geinir; *jena- is needed to account for OHG jen-, MLG jene, and ON enn; and finally *jana- is needed for OE geon (< unstressed *jon-) (Krahe 1969:§39, Fulk 2018:198). The first element in PG *ja-na- can be identified with the discourse particle ‘yes’ (Go. ja, OHG ja < *i̯ó ‘thereto, and’, possibly cognate with Hitt. -ya ‘and’; Dunkel 2014b:384–385), with the related alternants *je-na- and strengthened *jai-na- (compare Go. jai, WS gēa ‘verily’) (Klingenschmitt 1987:178; see also Dunkel 2014b:387). ⁵

The details of this etymology have been debated from seemingly every angle. Cercignani (1984:323–324, 327), for instance, does not accept <ei> in MHG geinir as evidence of a stem with *-ai-, but sees both the Gothic form and possibly even East OF iēna as continuing PG *jaina-. He also points out that *jena- is strictly necessary only for West OF ien and OE ‘both’, since both OHG jen- and ON enn have alternative explanations, though for ON enn, notably, he considers ‘the postulation of a Proto-Germanic uncompounded stem */ena-/’ in, for example, de Vries (2000:286 s.v. inn) to be ‘quite unnecessary’ (Cercignani 1984:324–325). The root vowel in various West Germanic forms, moreover, makes i-umlaut relevant to the discussion. Cercignani (1984:325, with references) is confident that OHG jen- comes ‘from the new formation /*janiz/’ (vs. *janaz for OE geon); furthermore, WS giend, gind, gynd ‘through, over, etc.’ point to *janði (vs. *janðe for OE geond, giond) (Cercignani 1984:325). Klingenschmitt (1987:178–182), for his part, suggests that umlaut was triggered by the formative -in- in the accusative singular (compare Kjølevik minino ‘my’), i.e. *ja-n-in- > *jẹn-, with a root vowel that was later generalized throughout the paradigm. I-mutation of *jan- could be relevant for North Germanic too, but of course not for Gothic (and besides, the m.acc.sg form in Gothic is jainana, with -an-). In short, while the interrelationships between the forms are not always entirely clear, with various plausible ways of dividing up the data, the postulation of three separate but clearly related stems is unavoidable, and it seems perfectly reasonable to think that OI enn (assuming with Stroh-Wollin that it is an ancient inheritance) should be considered a member of this cognate family.

As an argument against the idea that OI enn is a continuation of PG *jaina- ∼ *jena- ∼ *jana ‘that, the other’, Stroh-Wollin (2020:221) claims that the semantic fit is awkward insofar as the Gothic and West Germanic cognates have a contrastive interpretation available, while OI enn does not. The more appealing option, according to her (see also a tentative Prokosch 1939:273), is to derive OI enn directly from PIE *eno- ‘that’ (with reservations about transmission via any ‘homogeneous’ stage of Proto-Germanic; see Stroh-Wollin 2020:224–226). As we saw above, this option has been called ‘unnecessary’ by Cercignani, and indeed, in order to accept this view one cannot be swayed by the ‘certain reluctance [in the literature] toward etymologies that do not link Scandinavian to other Germanic varieties’ (2020:225). Now, there is of course nothing wrong, in principle, about viewing OI enn as a Nordic retention of PIE *eno-. It is perfectly possible that Old Icelandic was the only Germanic variety that happened to preserve this ancient element. Still, if we have a choice between two formally acceptable etymologies with one presenting few to no Germanic cognates and the other presenting many Germanic cognates, then clearly the latter option must win out.

With that in mind, consider the broader Indo-European cognate situation regarding the elements PIE *áno- and *nó-, both distal stems thematically derived from the particle *án ‘on the other side; possibly’ (Dunkel 2014b:64). See Table 1. It is striking that there are zero Germanic cognates under *áno- and *éno-. Adding OI enn to either the first or second column might then appear to be a reasonable supplement from yet another Indo-European dialect, were it not for the fact that *i̯ó no- (related to the other forms through its stem *no-) – which has only Germanic cognates – is a perfectly viable option for deriving OI enn, too. As Stroh-Wollin (2020:221) recognizes, it is ‘formally … unproblematic’ to derive OI enn from PG *jenaz, with (i) loss of initial j- as expected (e.g. Noreen 1923:§231 on ON enn and Go. jains) and (ii) absence of subsequent breaking of *e considering that OI enn is a western form (see e.g. Schulte 2018:62–63 on the geography of breaking).

The availability of a contrastive reading of the pronoun in East and West Germanic does not justify severing ties to the North Germanic form. In fact, the semantic change that needs to be assumed is not dramatic at all. If there is (something like) monogenesis of Go. jains, OHG jen-, OE geon-, OI enn < PG *jaina- ∼ *jena- ∼ *jana ‘that, the other’, then the only additional assumption needed is that some relatively cheap semantic change has occurred: the contrastive meaning originally present in Proto-Germanic has been retained in Gothic and West Germanic but lost in North Germanic, as part of the semantic bleaching we must otherwise assume in the grammaticalization from demonstrative towards article status. ⁶

3. The etymology of hinn

The demonstrative hinn is composed of the deictic particle PIE *e (Lat. hi-c, hae-c, ho-c ‘this’, nun-c ‘now’, OCS se ‘this’, *i > Hitt. kī(-) ‘this’, among many others) plus the pronominal stem *no- (*áno-) already discussed above (Dunkel 2014b:58, 396–401; see also e.g. Prokosch 1939:273). As a distal and contrastive pronoun, hinn (with n.nom/acc.sg hitt in Old Icelandic and unassimilated hint also attested in older Danish and Swedish) is to be connected to Gr. κϵῖνος ‘that’ < *e e no-, which has undergone a similar semantic development as the Nordic form, where the distal stem *no- has cancelled the proximal meaning of the initial particle (Dunkel 2014b:64 fn. 55), which is to say that the pronoun *no- provides the main deictic force while the particle is merely a (distance-neutral) reinforcer (compare Lat. illi-c = ille ‘that’; see Klingenschmitt 1987:177–178). Indeed, Pokorny explicitly puts it in diachronic terms, stating that *V- was ‘ursprünglich ich-deiktisch, später auch “jener”’ (1959:609 s.v. o-, etc.). That this deictic bleaching happened independently in Greek (e.g. Beekes 2010:397 s.v. ἐκϵῖ, ἐκϵῖνος) and Nordic (and Latin) seems beyond question. ⁷

As for the article OMN hinn/hit, the simplest analysis, given the above information, would be to assume that it too ultimately derives from *e no-, though of course at some point having diverged onto an independent path of development from the contrastive pronoun. On this view, hinn/hitt and hinn/hit constitute a doublet (see also Fulk 2018:198). Stroh-Wollin (2020:220–224) does not, however, opt for a doublet analysis. For her, the definite article hinn/hit, which is specific to the mainland, derives from a proximal PN * (which is considered not to have arisen any earlier than Proto-Nordic, but if back-projected is composed of the same basic elements as the contrastive pronoun, which she writes as PIE *ei and *enos).

One of her main reasons for providing separate etymologies is the seeming unlikelihood ‘that one and the same hin- stem gave rise to all the diverse functions of hinn forms on the mainland’ (Stroh-Wollin 2020:221). The ‘diverse functions of hinn’ in question are: (i) contrastive/distal ‘the other/former’, (ii) antecedent to relative and ‘that’-clauses, (iii) attributive demonstrative with (weakly) proximal deixis or anaphoric reference, (iv) preadjectival article ‘the’, and (v) postnominal article ‘the’ (see Stroh-Wollin 2020:218, her Table 2). Generally speaking, there is nothing peculiar about one element with different functions. In cases where very different functions – the notion of ‘very different’ of course always being up for debate – are at stake, one might argue for separate lexical entries displaying homophony, but the functions mentioned above seem to be closely enough related (see below for discussion of the proximal) that such an analysis is not immediately obvious.

It is instructive to consider the case of the demonstrative den in modern Swedish, which can be used contrastively, anaphorically, with distal deixis, among other uses (see SAG II:322–326). We might also distinguish the article den, which is not only the preadjectival article but can also be used to determine a relative or ‘that’-clause (see SAG II:302–304). The point here is not how similar these functions are to the ones mentioned above for hinn, but rather to point out that Sw. den is highly polyfunctional. While we might distinguish the demonstrative from the article on the synchronic level, no one is tempted to provide separate etymologies for demonstrative den vs. article den. However the rather complex situation of den actually came to be (see Stroh-Wollin 2015a), is there really any doubt that contrastive den (e.g. Ge mig den boken ‘Give me that book’) and preadjectival article den (e.g. den fina boken ‘the nice book’) both ultimately come from *sa/þa-? ⁸

The fact that the n.nom/acc.sg form is systematically distinguished as hitt in the contrastive pronoun and hit in the article might appear to support two separate etymologies, but this is not an unexpected state of affairs: the definite neuter singular form with a single -t can be thought of as the result of analogy with *þat and/or the result of phonetic erosion during the grammaticalization process. ⁹ Even the phonological facts about Sw. den/det here provide some perspective, with /dϵnː/ and /deː(t)/ in the demonstrative but ‘reduced’ /dϵn/ and /dϵ/ in the article. Again, whatever the exact explanation for this difference might be, it is not sufficient to prop up two separate etymologies for these items.

Nevertheless, Stroh-Wollin suggests that the association of both ‘a strong distal deixis and a clear proximal deixis’ with the demonstrative hin- means that more than a single kind of hin- is at stake (2020:217–218). Stroh-Wollin (2016:143–145, 2020:211–212) capitalizes on a handful of interesting attestations to offer a new etymology for the definite article hinn/hit, namely that it derives from a more recently formed proximal PN *hi-na-. I summarize her evidence in (1–3). ¹⁰

There are good reasons to think that these forms do not constitute evidence for a Proto-Nordic proximal. Above all, I take issue with the ‘strengthened variant[s]’ (Stroh-Wollin 2016:145) in (1), which are all, with the exception of (1a), from medieval Gutnish. These forms clearly have the structure of the reinforced demonstrative *sa-si ‘this’ (e.g. hina hinna = ON þenna, hita hitta = ON þetta); the forms with -ss- are especially indicative of the reinforced demonstrative, where the stem ON þess- (e.g. þessu, þessum, þessar, þessir, etc.) has been extended to the m.acc.sg (hinna → hiss-an) and the f.nom.sg (hiss-un, where -un is probably from hun ‘she’; Kock 1895:132–136). The initial consonant is readily accounted for by debuccalization (þ- > h- > Ø) targeting high-frequency items, such as the reinforced demonstrative. Other examples of such debuccalization include Faroese hósdagur ‘Thursday’, har ‘there’, and hesin, henda, hetta ‘this’ (Árnason 2011: 122); Övdalian isn, isų, ittað ‘this’ (Steensland 2021:464); (eighteenth century) Norn ita (< *hitta) < þitta ‘this’ (Barnes 1998:13, 17, 19). See Ralph (1975:129–131) and Barnes (1985, contra Braunmüller 1980); see also Lander (2020:20–21 fn. 13).

All of the forms in (1) can be explained not as forms of a new *hi-na- ‘this’ but rather just as medieval outcomes of the old reinforced demonstrative *sa-si. Note also that even if analogy, rather than sound change, turns out to be the correct way of accounting for the anlaut in these forms, this would not endanger the analysis that the forms in (1) are actually reinforced demonstratives. As for the two forms in (2), these can be viewed simply as the neuter demonstrative (compare OSw. þæn and þæt), with debuccalization of the initial consonant, or as ‘endingless’ variants of the reinforced demonstrative (see Lander 2020:161–163 for discussion).

The forms in (3) do not provide much better evidence for the ‘new’ proximal. The form hino ¹¹ (3a) on the Strøm whetstone is plausibly analyzed as the old proximal PG *hi- (compare Go. m.acc.sg hina, n.nom/acc.sg hita, m/n.dat.sg himma ‘this’). Since there are no word boundaries indicated (Jansson & Wessén 1962:195), stainin on G 110 (3b) does not necessarily have to be read as stain in at all; it could be an early attestation of the suffixed definite article (compare Sö 41 mirkit + mikla ‘monument.the big’). The form hin (3c) on DR 47 (an iron fitting for a door) is a n.acc.pl attributive demonstrative going with dʀr dyr(r) ‘door’ (plurale tantum, usually feminine but sometimes neuter). This example may well illustrate weak proximal deixis associated with *hin-, but the simplest explanation here is that the incipient definite article, having evolved away from its mother form (i.e. the contrastive/distal pronoun), had developed ‘neutral’ deixis, with both weakly proximal and weakly distal readings available, much like Fr. ce, celle, ces. ¹² Finally, the example in (3d) would be amenable to the same kind of analysis: the word skulld(-) is mentioned a number of times before hina skulld, giving hina an ‘article-like’ feel (Stroh-Wollin 2020:212). Indeed, as a neutral demonstrative, hina need not be interpreted with proximal deixis here – distal ‘that debt’ is a perfectly reasonable interpretation as well.

In sum, the evidence for a ‘new’ Proto-Nordic proximal evaporates upon closer inspection, and with this also the alleged source for the definite article on the mainland. It should be mentioned that the more conventional etymology (i.e. article hinn/hit has the same source as contrastive hinn/hitt) also puts the grammaticalization of the article on more typologically natural ground, since definite articles very often develop from distal demonstratives (Greenberg 1978:61). This generalization is not exceptionless, with proximal demonstratives also attested as a source (see Heine & Kuteva 2002:109–111), and Stroh-Wollin (2016:145 fn. 22) emphasizes that as long as we assume that these pronouns are weakly deictic, there is little reason to doubt that such a development was possible. In other words, while the typological argument is not enough on its own to discard the ‘new proximal’ etymology of OMN hinn/hit, it could be considered the final straw.

4. Summary

In conclusion, I have argued in favor of two quite conventional etymologies in this paper. The first is that, if we assume that the definite article OI enn is in fact an old retention, then the simplest etymological explanation is that it comes from PG *jena-. Germanic is the only Indo-European dialect showing the combination *i̯ó no-, and since it is formally and semantically trivial to derive enn from *jena- (∼ *jana-, jaina-), this analysis is more straightforward than deriving enn from PIE *áno or *é no (for which Germanic continuations appear to be absent). Furthermore, I propose that both the pan-Scandinavian contrastive hinn/hitt/hint and the Old Mainland Nordic definite article hinn/hit are descended from the same item, namely distal/contrastive PG *hin- (compare Gr. κϵῖνος), contra Stroh-Wollin (2020), who derives only the former from this item. Put differently, contrastive hinn/hitt and article hinn/hit are a doublet, in exactly the same basic way that modern Swedish demonstrative den/det and article den/det are a doublet. Crucially, just because an old demonstrative evolved into a definite article does not mean that the demonstrative went extinct in the process. Rather, PN *hina- split into two lineages: on the one hand a contrastive pronoun (> N hinn/hitt), and on the other hand a proto-article (> OMN hinn/hit), the latter of which at some point went from being purely distal to being weakly distal or weakly proximal on its way to becoming a true article. The developments I assume are summarized in (4).

The grammaticalization of a definite article from a distal demonstrative should come as a surprise to no one, and (4) does not amount to much more than this. While Stroh-Wollin’s dual etymology of hinn/hitt and hinn/hit makes use of the available evidence in an inventive way, I would contend that the evidence she presents in favor of the more complicated etymology (article < proximal demonstrative) can actually be recast as evidence in support of the etymology in (4), with the neutral semantics of hin- as a missing link in the development from distal demonstrative to definite article. In other words, (4) is simply the less complicated – and therefore the preferable – explanation.