Early Qenya Grammar - what can we learn?
EQ: Early Qenya
CE: Common Eldarin
PE: Parma Eldalamberon
VT: Vinyar Tengwar
The recently published issue of Parma Eldalaberon 14 contains rather complete writings about the grammar of early Qenya, a conceptual (in external timeline) predecessor of later Quenya. Since in these writings we actually get to see Tolkien give a complete description of a grammar system, we might hope to gain enough insights into the underlying concepts and ideas that in the end our analysis of mature Quenya (and eventually Sindarin) texts is better guided. It is the purpose of this article so show possibilities where the manuscript might be beneficial for our understanding of Quenya and Sindarin.
1. Early Qenya is different
Anyone who had hoped that the publication of a grammar by Tolkien would settle all problems, at least for the discussion of Quenya is likely to be disappointed. In general, we see a language which shows a good deal of similarity with mature Quenya but also several striking differences.
To provide an example: The declination of the noun is addressed in PE14:43 for EQ: kalma (light):
Sg: Nom: kalma Acc: kalmat Gen: kalman Dat: kalmar
Pl: Nom: kalmali Acc: kalmalin Gen: kalmalion Dat: kalmalir
Contrast this with mature Quenya: Here, the word Q: calma is rather translated 'lamp, a light' and its declination is strikingly different:
Sg: Nom: calma Acc: calmá Gen: calmo Dat: calman
Pl: Nom: calmar Acc: calmai Gen: calmaron Dat: calmain
What is the regular plural in EQ has become the partitive plural calmali (some, many(?) lamps) in Quenya and case inflection endings are markedly different.
Thus, we cannot a priori expect that a rule found in EQ grammar has a counterpart in later Quenya grammar. However, if we encounter an object in Quenya or Sindarin texts that looks similar to an object in EQ grammar and if the rules outlined in the EQ grammar are able to shed light on the behaviour of this object, then it is a reasonable assumption that we have found an idea that was not altered by Tolkien in the course of his language-creation process.
The following sections do not follow any specific pattern - they just deal with particular points which appear of special interest in rather random order.
2. Impersonal constructions, Passive
Occasionally, verbs marked as 'impersonal' appear in our sources of Quenya and Sindarin.
In Sindarin, the following impersonal verbs can be identified:
boe (Etymologies, entry MBAW-, there appearing as Noldorin bui)
This is not translated, but only marked as impersonal, however, the translation of the root itself suggests that it means something like 'it compels' (with a dummy pronoun 'it' appearing in the english translation - logically, there is not subject since the verb is impersonal)
ail (Etymologies, entry ULU-, appearing there as oeil, eil)
Translated 'it is raining'; marked to be derived from *ulya, the missing pronomial ending (and consequent loss of the ending -a in Noldorin) suggests strongly that this verb is also impersonal.
In Quenya, somewhat more information is available:
'to dream' is explicitly identified as being impersonal.
orë nin caritas (VT41:13)
Translated 'I would like to do so', 'I feel moved to do so'. Literally, '[it] impels for me to do it'
The EQ grammar has two different uses of 'impersonal'. PE14:56 has
Note that the neuter (rem: i.e. the pronoun (h)a-) is never used as as impersonal subject: there is no prefix used at all in that case, as uqe 'it rains', tiqe 'it thaws'..
This adds up to a rather consistent picture that sentences in any form of elvish do not require a subject (quite unlike English sentences which insert 'it' if no agent doing something like 'raining' can be identifid).
The three examples of words in connection with the weather suggest that Elvish in general does not name an agent doing the weather, hence probably 'it snows', 'it is cold', 'it is foggy', 'it is windy' would all lead to subjectless sentences. As the other examples show, impersonal constructions seem also be used to describe conditions that the person subject to them cannot alter (which might often be unconscious decisions): Outside forces compell to do something and an elf experiences his dream, he doesn't consciously shape it. From orë nin caritas we can deduce that the person subject to external conditions is found in dative.
To give some suggestions:
Q: ná ringa ([it] is cold) (not **nas ringa).
Q: ná ringa nin ([it] is cold for me = I'm cold)
S: ring enni ([it is] cold for me = I'm cold)
However, PE14:56 has also a second definition of impersonal:
The inflections of verbs are always pretty regular and consist of (a) no ending for singular (b) -r for the impersonal (distinct from the endingless form, e.g. uqe 'it rains', but tulir 'one goes, somebody goes').
Thus, we have the truly impersonal form discussed above when no agent doing something is present at all and we have a second form in -r which is used when a agent is present, but is not identified or known. In order to draw a distinction between those, I will refer to the second form as 'unspecified' instead of 'impersonal'.
The interesting thing about the unspecified form is that it can be used to express passive: The section quotet above continues:
...this becomes a passive if pronomial elements are added, for these are in the accusative (rarely dative). In the first case (accusative) these still may retain (...) their accusative-position after the verb, but as the passive feeling has increased such expressions as ha-matsir 'it is soiled' are not unusual.
We may decompose the form as follows: matsir (some-one soiles), matsir ha (someone soiles it, pronoun in accusative position), ha-matsir (it is soiled, pronoun moved in nominative position, passive is understood). (to compare, ha-matse would be 'he soiles').
One wonders if this is at the root of S: i sennui Panthael estathar aen (who ought to be called Panthael). Assuming for a moment that the ending -r is relevant for the same purpose in Sindarin, we would get estathar (someone will call), with the relative pronoun in accusative (replacing the pronoun in this particular context) this would be i estathar (whom someone will call = who will be called).
However, the use of -r in later Quenya texts suggests that Tolkien abandoned this form: Laurië lantar lassi is not translated 'someone falls golden leaves' - so in Quenya, -r is just used to mark a verb for plural if an explicit subject is given (if a 3rd person plural subject needs to be expressed in the verbal ending, this is -ntë or -t which is quite different from -r). This also persists in Sindarin: Contrast Q: i carir quettar ómainen (who make words with voices) with S: Dor Gyrth i chuinar (land of the dead that live). In Quenya, a relative pronoun in plural is followed by a verb marked for plural only, the same seems to be true in Sindarin, we have no indication that -r is intended as 3rd person plural or unspecified form here.
It is suggestive to assume that Tolkien in the end combined his two definitions of impersonal: A Quenya verb cenir is in almost the same sense personless as Q: orë or EQ: uqe - it does not contain a subject of any kind. Maybe Tolkien decided later that instead of inserting an unidentified agent, passive is expressed by inserting no agent at all, but (optionally) marking the verb for plural instead (so old constructions using the Qenya ending -r would continue to be valid). Hence, estathar might be translated '[some unspecified persons] will call'.
There is one more piece of evidence to substanciate this view: In the 'Ataremma', there is the line na care indómelya, which has to be 'your will be done'. If one uses the previous line aranielya na tuluva using the literal translation 'wish that your kingdom will come' to identify na with an optative particle similar to nai, we can assume that na care indómelya translates literally as 'wish that [someone] does your will' with an impersonal 'someone' (technically, the sentence is without subject), and such a construction would line up well with the scenario outlined above.
3. Modal particles for conditional and subjunctive, Optative
PE14:59 describes the construction of conditional and subjunctive expressions:
(...)This is not expressed inflectionally but by particles, nai and ki, of which nai represents remoter possibility ('might'), ki nearer 'may'. The pure optative is also often expressed by nai or naike combined, at head of wish.
This seems to connect rather well with later Quenya: Here we find
Nai hiruvalyë Valimar! Nai elyë hiruva! (Maybe thou shalt find Valinor! Maybe even thou shalt find it!)
where the EQ explanation is clearly helpful in resolving the curious ambiguity between the interpretation of the sentence as a wish and the translation 'maybe'. It also explains
Nai tiruvantes i harar... (May they keep it those who sit...)
We also have an example of the second form in mature Quenya:
Lá caritas alasaila ké nauva (Not to do it might turn out to be unwise.)(VT42:34)
Identifying EQ: ki with Q: cé the interpretation of the sentence is straightforward. Thus, we have actual reason to assume that the grammar of these forms is not markedly different in Quenya. So - what are they used for and how? Tolkien provides examples:
For direct use, we have EQ: hi-tule ki (she may come) which we can try to update to mature Quenya as tulis cé.
With mai (if), Tolkien writes nai or ki are usually in apodosis (i.e. appear in the second part of an if-clause). Hence:
EQ: mai ni-tule - tu-tulil (if I come - they come)
Tentatively using the element Q: *ai = 'if(?)' extracted from aiquen (if anybody) (I just need something to form equivalent Quenya sentences, so I don't really justify this choice here), we may form
Q: Ai tulin, tulintë (If I come, they come.)
We have a future example without uncertainty (which is a strange kind of conditional...):
EQ: mai ni-tuluva - tu-tuluval. (Whenever I come - they come.)
Q: Ai tuluvan, tuluvantë. (Whenever I come - they come.)
and we see that the modal particle is needed to form a conditional where the outcome is uncertain:
EQ: mai ni-tuluva - tu-tuluval ki. (If I come, they will come.)
Q: Ai tuluvan, tuluvantë cé. (If I come, they will come.)
There is no immediate relevance for Sindarin. However, we have some reason to assume that modal particles of subjunctive form a generic feature of Tolkiens Elvish languages. The modal particles are almost unchanged from Qenya to Quenya which suggests a certain persistence of the idea. Furthermore, we find in the 'Ataremma' na care indómelya which is hard to translate if one assumes that na is an actual form of 'to be' but makes perfect sense if one sees this na as a variant of nai denoting optative. Now, the same particle also appears in na airë esselya (holy be thy name) which looks structurally very similar to its Sindarin counterpart no aer i eneth lín. Therefore, it is likely that subjunctive is expressed in Sindarin using modal particles as well.
With this conclusion, turn back to S: i sennui Panthael estathar aen. Using the reasoning of the previous section, we can readily translate S: i Panthael estathar as 'who will be called Panthael', or, following the example Q: vanda sina termaruva (this oath shall stand), we might also try 'who shall be called Panthael'. We would precisely need a modal particle to express subjunctive and transform 'shall > should'. Candidates are sennui and aen. Neither of those looks like being related to EQ: nai, ki. However, aen replaced a crossed out ge (TolkLang 40.45), and this we may link to EQ: ki Q: cé, therefore aen is the most likely candidate for a modal particle in Sindarin.
4. The question of 'that'
I don't know how many of the native English-speaking readers are aware of the fact that English has two different types of 'that'. If you are, please indulge me for a moment.
First, we have the relative pronoun, for example in 'The thing that he bought.'. Then, there is the conjunction starting a nominal clause, e.g. 'He said that he went to buy something.' How can I tell the difference? For me, this is easy because my mother tongue (German) uses two different words: The first sentence would see 'that' translated as 'das' whereas the second sentence would lead to 'daß'. But also in English, there is a simple test: If it is a relative pronoun, you can replace it by 'which'. If you can't do that, it is the conjunction. The first sentence becomes 'The thing which he bought.'; this is fine, so it is the relative pronoun, whereas the second sentence gets rather meaningless: 'He said which he went to buy something.' This is not good English.
It is a pecularity of English that these two functions of 'that' coincide. Nevertheless, some writers have used (notably in Sindarin) i (which is attested as definite article and relative pronoun only) also to express the conjunction 'that'. This is mainly based on Tolkiens comments that Q: nai can be decomposed into na + i (be [it] that) (RGEO:68).
In his Early Qenya Grammar however, Tolkien is quite explicit about the distinction between pronoun and conjunction. PE14:54 has:
The indeclinable relative pronoun is ya, which is either to be understood in any relation, or, very frequently, is defined by demonstrative or pronomial or adverbial words inside the relative clause.
The conjunction 'that' is ne and must not be confused.
This basically confirms that Tolkien was well aware of the distinction and meant to make them in his Elvish language by the time he designed Qenya. A further serious problem for the decomposition of nai is found in PE14:57 where explicit forms of the verb 'to be' are given: e (i) (is), il are, ir (one is)\dots Nevertheless, nai is mentioned as a modal particle expressing sunjunctive/optative. It is just the decomposition as 'be [it] that' that does not work in this setting. At least at this stage, nai apparently had a very different origin.
We have some further evidence against the use of i as a conjunction in mature Quenya: First, there is the sentence Q: Merin sa haryalyë alassë (I wish that you have happiness.) Although of doubtful origin, it would be in line with the assumption that Tolkien did continue to make the distinction between the two uses of 'that'. Second, in the 'Ataremma', we find na in a function that is very similar to nai - indicating that the i here is not really needed.
The most plausible conclusion is that Tolkien decided relatively late that the form nai should be composed from the stem 'to be' and the conjunction 'that'. The translation of Namárië still has 'maybe' as translation, denoting more remote possibility than a wish, the explanation of this as 'may it be' seems more an afterthought. We have to conclude that at least late Quenya probably permits the use of i as the conjunction 'that' (or maybe this is only a relict from CE surviving in compounds and i has gained a different role since?), but this may not be not relavant for early material.
Up to now, the only larger number from where we might deduce how numbers are composed in the elvish languages occurs in the King's Letter (SD:128-9) nelchaenen uin Echuir which is translated 'the thirty-first day of the Stirring'. Based on the occurance of erin dolothen Ethuil (on the eighth day of Spring) and the knowledge of toloth (eight) from LR:394, we might identify the ending -en, transforming the number 8 into the ordinal number 8th. If so, we are left with nelchaen.
In PE14:49, we see the following scheme:
Numbers up to 10 are represented as single words which are similar to their mature Quenya equivalents, e.g. EQ: kanta (four), lemin (five). Numbers from 11 to 19 are formed (schematically) by prefixing the first part of the single number to kea (ten), also this may undergo shifts, such as in minqe (eleven). However, the numbers larger than 12 seem pretty regular, e.g. kankea (fourteen), lenkea (fifteen).
The scheme changes again once 20 is reached: The element kea is now placed in partitive case with inflectional ending -inen (this became the instrumental in mature Quenya) and becomes kainen (of tens). Hence kankainen (four of tens; fourty), but kankea (four AND ten). The last digit is then expressed by writing the number in front position, hence *kanta kankainen (44). The numbers are formed quite the opposite way in English and Qenya: While english has four-thousand-five-hundred-thirty-two, Qenya would have * yúyo neldekainen lemin tuksa kanta húmi (two-thirty-fivehundred-fourthousand).
In the course of this building scheme, Tolkien mentiones the form EQ: nel(de)kainen (30). Comparing this with S: nelchaenen, the similarity is striking.
Presumably, the two pieces of evidence can be reconciled as follows: In the King's Letter, erin dolothen is 'on the eighth', however nelchaenen uin Echuir uses a different preposition and is more likely 30 days from the beginning of the stirring (note that suilad uin aran is 'greeting from the king' ), therefore we may guess nelchaen 'thirty'. We know, however, caer (ten), attested in LR:363. The question arises if Noldorin, similar to Early Qenya, uses remnants of an inflected form of caer to count tens. If so, we may form *nelchaer (thirteen), but nelchaen (thirty). Analoguous to EQ, the last digit would then presumably be prefixed: S: *min nelchaen (thirty-one), S: *min tadchaen (twenty-one) and so on.
6. Pronouns - characteristical consonants
The discussion of pronouns does not introduce a great surprise: Basically we 'rediscover' the set of characteristical consonants that can be inferred from mature Sindarin and Quenya:
To quote the verb inflection prefixes (PE14:52):
Sg. 1. ni- 2. ke- 3. m. hu- f. hi- n. ha-
Pl. 1. me- 2. le- 3. m. tu- f. si- n. ta-
1. Pl also has qu-, and we are told si < *ti.
So, apart from qu-, this is the rather well-known set: n and m for 1st person sg. and pl. (cf. S: nin (me, to me) and mín (used as 'we')), k and l for 2nd person (cf. S: le (to thee)), here to make a distinction between sg. and pl., later presumably to make a distinction between familiar and courteous, and finally s and t for the 3rd person (cf. ho, he, ha, LR:385 entry S-), here again to make a distinction between sg. and pl. which still seems to be the case in mature Quenya, but not in Sindarin, cf. sellath dîn (all his daughters). Thus, the distribution of characteristical consonants indicates that the general outline of the pronomial system is rather stable during the long process of language creation, although the details vary a lot.
An interesting side remark: Discussing modificators of root consonants, Tolkien gives in PE14:66 the example
These relationships are equally important in the formative elements: for example
pronomial element ke as (...) kke,
explicitly giving a 2nd person form -kke which would, if it was not abandoned, eventually lead to a verbal ending -ch in Sindarin. Thus, the assumption that -ch is indeed associated with some kind of 2nd person is not unreasonable, although we are not in a position to identify the precise role of this ending. If this is relevant for the infamous Arphent Rían sentence is unclear.
7. Past tense formation
The description of the past tense formation in Early Qenya is interesting, as some ideas seem to persist in Noldorin and Quenya:
The past stem is obtained by the suffix -ye, (...) -ie, or -ne, but -ie (the commonest) is normally accompanied by stem strengthening consisting of (1) a-infixion (2) n-infixion, (3) vowel lengthening (this last perhaps largely an analogical extension from the â in many stems..
In Sindarin (or the Noldorin of the Etymologies), a final -ie would be lost, but we observe the same stem strengthening pattern for the Noldorin verb past tense formation (see 'The Past-Tense Verb in the Noldorin of the Etymologies', [URL]http://www.elvish.org/Tengwestie/articles/Hostetter/noldpat.phtml[/URL]). Here, we have a-infixion hal- -> haul (lifted), n-infixion, dag- -> danc (slew) and vowel lengthening nidh- -> nídh (bruised).
Up to a prefixed augment, the suffix -ie furthermore is very similar to the formation of the perfect in Quenya: Here, we find Q: utúlie (has come) whereas EQ: túlie is denoted as past tense. On the other hand, n-infixion seems to be primarily used to denote past tense in Quenya, so apparently Tolkien later introduced a difference here. (WJ:366 confirms the existence of past tense and perfect tense in Quenya as separate entities: Discussing the verb auta-, one finds The most frequently used past and perfect were vâne, avânie (...).
But in Sindarin, apparently forms with augment and strengthened stem vowel did not end up as perfect: WJ:415 has *ekwê was probably a primitive past tense, marked as such by the 'augment' or reduplicated stem vowel. Past tenses of this form were usual in Sindarin 'strong' or primary verbs, as *akâra 'made, did' > S agor.
Obviously, we encounter the same ideas, i.e. n-infixion or and ending -ne on the one hand and strengthening of the stem vowel with or without connection with an augment and the ending -ie on the other hand in all stages of the language evolution, but the details remain unclear. Judging from the number of examples of 'stem verbs' forming their past tense in -ie in EQ (P14:58) and assuming that those would not be changed to nasal infixion when going to Noldorin and Sindarin, the remark in WJ:415 that those past tenses were 'usual' makes indeed some sense. However, unfortunately it remains quite unclear which of the stem verbs should form their past tense in which way in Sindarin.
8. Comparative and superlative of adjectives
In PE14:47 we learn about the comparative formation of adjectives in Early Qenya:
Adjectives have two comparatives, (a) augmentative, (b) diminuitive:
(a) -lda (...)
PE14:48 reveals somewhat more about how to compare things:
(...) This is shown by the choice of words for 'than': after -lda it is la 'before' ; after -tsa it is no 'after'
This shows some elements present in mature Quenya also: VT42:32 indicates that Q: lá is used to compare things, however it does not seem to be necessary to append a comparative ending in order to use this form. The Quenya form Q: malda (more) however still seems to involve the EQ ending. For all we know, in general these endings seem to be absent in making comparisons in Quenya.
In Sindarin, we find a prefix S: an- (PM:358) to form the comparative. This is probably related to the Quenya intensifying prefix Q: an- (Letters 278) meaning 'exceedingly'.
In PE14:48, we also learn about superlative:
The superlative is expressed by the comparative with prefixed article (...) followed by the genitive plural adjectival or partitive in -înen - the latter especially of collectives, as:
i-ner i-táralda 'n-Noldolion 'the tallest man of the Gnomes'
The idea to form the superlative by a construction involving genitive seems to be valid in Quenya as well: Here we have ancalima elenion (brightest among the stars), using the intensifying prefix instead of a (presumably dismissed) comparative ending.
I would like to thank Aelfwine and Pengolodh for their helpful comments on the draft version of this article.
Back to the index