The Quenya Pronominal System - a summary

Abbreviations used:
LOTR: The Lord of the Rings
SIL: The Silmarillion
LR: The Lost Road
WJ: The War of the Jewels
SD: Sauron Defeated
RS: The Return of the Shadow
PM: The Peoples of Middle-Earth
PE: Parma Eldalamberon
VT: Vinyar Tengwar

1. A typology of Q(u)enya pronouns

The pronominal system is always one of the more obscure areas in the study of Tolkien's Elvish languages. Yet in Quenya this is not so because we lack attested information, it is rather that the known material is full of contradictions and puzzles, giving evidence for Tolkien's habit of continuously developing his languages. Quite evidently, there is no such thing as the 'right' Quenya pronominal system, all we can hope to do is map out the different ideas Tolkien employed over time. Having said that, quite a few features (and forms) of the pronominal system are remarkably stable from Early Qenya right to the latest texts dating from the 1970's.

At the core of the Q(u)enya pronominal systems is a set of 'characteristic' consonants. Each of these consonants signifies person (and often number). For example, n is the characteristic consonant of the 1st person singular, which in fact implies that all pronominal forms of 1st person sg. involve this consonant in some way. However, these consonants are features of Common Eldarin, which means that they may not appear in their original shape but rather, dependent on the phonological environment, be the subject of sound shifts and blendings.

Quenya phonology doesn't permit single consonants as independent words, however single characteristic consonants may still appear as pronominal verbal endings, cf. e.g. -n in utúlien 'I am come' (LOTR), the -n signifying 'I'.

Characteristic consonants can be elaborated by appending a vowel. This yields the set of non-emphatic pronouns in nominative. Such a form appears e.g. in ni véla 'I see' (The Father Christmas Letters).

A part of the consonants (some combinations are not permitted by Quenya phonology, other don't seem to make sense) can be elaborated further by appending -ye to the characteristic consonant. Thus, we arrive at e.g. tye from A yonya inye tye-méla 'And I too, my son, I love thee' (LR:61) which is presumably derived from *kjê with subsequent word-initial shift kj > ty (cf. KYEL > Q: tyel- 'come to an end' (LR:366)).

While the forms derived with -ye are also seen as long verbal endings, cf. -lye in hiruvalye 'thou shalt find' (LOTR), there are other long verbal endings where two consonants combine with a vowel to form the ending. Clearly, for phonological reasons these long endings cannot stay alone (unlike the -ye group). As an example we quote -nte from tiruvantes '*they will keep it' (UT:305).

Obviously related to the long pronominal verb endings are the possessive endings (being applied to nouns) - they only differ in the vowel. While long pronominal verb endings include e, the possessive endings are formed with a, cf. esselya 'thy name' (VT43:12).

The next elaboration apparently consists in appending the long verbal endings to a vowel. This yields the set of emphatic nominative pronouns. We find e.g. elyë 'thou' (LOTR) from -lye.

The longest form in which pronouns may appear (not counting case inflection) are independent possessives. These seem to involve a repetition of the characteristic consonant around a vowel in addition to the ending -ya, cf. ninya 'my' (LR:72).

2. Complete paradigms

The easiest way of getting an idea of changing and stable features of the Quenya pronominal system is to focus on complete paradigms. For independent pronouns, we have three at our disposal: The Early Qenya Grammar (PE14:52-54) allows a thorough study of the pronominal system in this early incarnation of the language (1920-1925). VT43:29 contains a paradigm of pronominal forms with the prefix ó-, from the known independent pronominal forms we can conclude that removing the prefix allows a reasonable guess about the non-emphatic set of pronouns as it was in the 1950s. Finally, notes dating from the same time as the essay 'Eldarin Hands, Fingers and Numerals' in VT47:37 (about 1967-70) contain a paradigm of pronouns with the prefix im- which likewise allows to glimpse the underlying system of non-emphatic pronouns. In addition to these tables of independent pronouns, there is also a table of pronominal verb endings found in VT49:16 which is contemporary with the im- paradigm. The ideas underlying the assignment of characteristic consonants between the last two table agrees well.

In addition, paradigms of pronominal verbal endings dating from around 1960 can be found in PE17:57,75 and PE17:132. Even later paradigms dating from 1964 and 1968 are reproduced in VT49:48ff. We will present those as well in the following.

The pronominal system of Early Quenya thus has the following shape (PE14:52, the forms appear hyphenated because Early Qenya has pronouns prefixed to a verb, not as endings):

  Sg. Pl.
1st ni- me-
2nd ke- le-
3rd m hu- tu-
3rd f hi- si- < *ti-
3rd n ha- ta-
The difference between me- and q- is that me- excludes the person(s) addressed, qe- includes them. The ó-paradigm looks as follows:

  Sg. Pl.
1st onye) óni óme
2nd olye) óle óle
3rd m óse óte
3rd f ósa (ós) óta (ót)
3rd n ótar ótari
The last line presumably refers to the form tar 'thither' (LR:389). Finally, the im-paradigm contains the following forms:

  Sg. Pl.
1st imne, imni imme
2nd intye inde, *imde
2nd imle  
3rd inse, imse, insa inte, *imte
Taking into account that the last two paradigms contain etymological forms and late analogical formations, the most likely shape of the underlying pronominal system looks like

  Sg. Pl.
1st #ni #me
2nd #le #le
3rd mf #se #te
3rd n #sa #ta
for the ó-paradigm and

  Sg. Pl.
1st #ni #me
2nd #tye #le
2nd #le  
3rd mf #se #te
3rd n #sa #te
for the im-paradigm.

From these, we can identify the following groups of characteristic consonants: n for 1st person singular, m,q,w for 1st person plural, k(c), k > t, l and d for 2nd person, s, h for 3rd person singular and t for 3rd person plural. Details of the distribution, most notably in the 2nd person are different and comparison with other forms suggests a distinction between formal and familiar use, but the general structure is remarkably stable.

A table of verb forms of care with columns for singular, plural and dual is found in PE17:57. Extracting the pronominal verb endings from this paradigm, we find:

  Sg. Pl. Du.
1st incl.   -lme -ñgwe/-nque
1st excl. -n/-nye -mme -mme/-nwe
2nd fam. -tye -nce †-xe/-cce
2nd form. -lye -lde/-lle -lle/-ste
3rd animate -s/-ste -nte -tte/-sto
3rd neuter -s/-ssa -nta  
impersonal - -r -t
This table does not only contain clearly marked forms for inclusive (i.e. including the person spoken to) and exclusive 1st person plural and dual and between formal and familiar 2nd person pronouns, but also a distinction between animate and inanimate use in the 3rd person. As such, it is the paradigm which makes the largest known number of distinctions.

Two tables of bare verbal endings are found in PE17:75, of which only one is reproduced here, as the second does not add much information. While person and number are identified, an assignment of which 1st person forms are inclusive and exclusive has to be made in comparison with the table on PE17:57, the distinction is not made explicitly. The table reads:

  Sg. Pl. Du.
1st incl.   -lme (-ngwe) -ngo/-lmo
1st excl. -n/-nye -mme -mmo
2nd fam. -t/-tye -ntye -lko
2nd form. -l/-lye -lle -llo
3rd -s -r -t
refl. -kse -kser -kset
While singular and plural are rather similar to the paradigm of verb forms found in PE17:57, the dual forms are remarkably different in the sense that Tolkien changed the way of marking dual by introducing the ending vowel -o. In addition, the table shows a reflexive form in -kse, which is probably to imply a usage like *tirikse 'he watches himself'.

Yet another paradigm is found in PE17:132 where verbal endings can be read off from the verbs care and ala. While number and person are again clearly identified, familiar and formal usage as well as inclusive and exclusive usage has to be deduced by comparison with PE17:57. The table reads:

  Sg. Pl. Du.
1st incl.   -lme -lmo
1st excl. -n/-nye -mbe -mmo
2nd fam. -tar -ltar -star
2nd form. -l/-lye -lle -llo
3rd -s/-sse -r/-nte -t/-sto
While retaining the idea of marking dual with -o, this table contains the not otherwise attested idea that the 2nd person familiar could be marked with a -tar.

The latest known table of verbal endings is found in VT49:16f (in VT49:16 forms are given appended to the verb care). The assignment of formal and familiar and inclusive and exclusive forms has to be deduced from the context. The table reads:

  Sg. Pl. Du.
1st incl.   -lwe -ngwe
1st excl. -n/-nye -lme -mme
2nd -l/-lye -lde -ste
2nd -tye    
3rd -s -lte †-ste
3rd - -r -t
The most striking changes are the reassignment of the inclusive and exclusive 1st person forms in plural and dual. It is also apparent that Tolkien in the end decided to abandon the idea of dual forms being marked with -o.

Several interesting questions can be studied from comparing all these paradigms: 1) The change in the meaning of characteristic consonants 2) the meaning of the associated vowels and 3) the shape of the inflected non-emphatic forms.

Changes in characteristic consonants

There are few surprises in the first person singular - n remains the characteristic consonant throughout all phases.

As for the 1st person plural, here we struggle with the problem that Quenya could have up to four different shades of meaning, all of which are seen in the verb paradigms: a) exclusive (we, but not you), b) inclusive (we and you) c) dual inclusive (the both of us) and d) dual exclusive (the two of us, but not you). The most common consonant of the 1st person plural is m, in Early Qenya apparently excluding the person addressed while q included.

We can study some of the changes by taking a look at other forms. In RS:324 we find the possessive -mma as inclusive, this dates to 1938. However, in Quendi and Eldar (about 1960), the verb ending -mme is exclusive (WJ:371), this is confirmed by the verbal paradigms in PE17 where -mme and its variant -mbe are identifiable as exclusive as well, whereas in a late restructuring (before 1965) -lve became the inclusive, -mme the dual and -lme the exclusive (VT43:6). Presumably this situation still holds for the verb paradigm shown in VT49:16 (the forms are not explicitly identified), thus -lwe (corresponding to -lve would be inclusive, -lme still exclusive, -mme would be exclusive dual and -ngwe inclusive dual). Such a dual form in -ngwe is also mentioned in VT43:36.

Thus, it is entirely conceiveable that #me in the ó- and the im- paradigm do not refer to the same meaning. It is plausible to identify the ending -lve with #we in the late paradigm, if that holds then #we signifies the inclusive form. From Namárië we can read of the dual met, since this was published, Tolkien was likely to stick with it, the remaining forms therefore have to be inclusive and exclusive. Unlike the 1st person dual forms in the verb paradigm VT49:16, met contains a dual marker -t. Apparently the, distinction between the different forms did not receive a lot of attention preceding 1960 since we only find #me in the ó- paradigm which may cover both forms.

The 2nd person has undergone some of the most pronounced conceptual changes. Tolkien struggled with two different decisions: a) should there be a distinction between singular and plural and b) should there be a distinction between formal and informal mode of address.

In PM:42-43, Tolkien describes the idea: 'All these [Elvish] languages...had, or originally had, no distinction between the singular and plural of the second person pronouns; but they had a marked distinction between the familiar forms and the courteous'. Obviously he didn't feel bound by this statement in a number of occasions. Early Qenya has k for 2nd person singular and l for 2nd person plural, i.e. a marked distinction between singular and plural and none between familiar and courteous, this idea being introduced later. Yet the ó-paradigm has no distinction between either singular and plural or familiar and courteous. In 'Quendi and Eldar' (WJ:364) we recover the Early Qenya idea of k marking singular and l plural: 'it often appears in the forms hekat! sg. and hekal! pl. with reduced pronominal affixes of the 2nd person (the -t is what a word-final -c in Quenya would be transformed to, see e.g. filit pl. filici (LR:381) from PHILIK). Finally, the im-paradigm shows a distinction between both familiar k and courteous l and a (general) plural d (which would however be shifted to l in an independent pronoun since Quenya phonology doesn't permit word initial d-). An element de is also described in WJ:363 where Tolkien refers to de and its variant le as 'pronominal elements in the 2nd person', though here no distinction between singular and plural seems to be implied.

A full distinction is upheld in all verbal paradigms where, based on a characteristic consonant k -tye (sometimes with variant -t, in one case also -tar) marks the familiar usage in singular and, based on the characteristic consonant l, -l/-lye marks formal usage. A distinct plural is formed by use of plural markers n,l (where l has a double role as both characteristic consonant and plural marker), leading e.g. with the k and plural marker n to -nce or with l used both as plural marker and characteristic consonant to -lle. A 2nd person pl. characteristic consonant d is observed in both PE17:57 and VT49:16, leading with plural marker l to -lde.

There is hardly anything remaining constant about the assignment if one considers the full development from early to late forms, the only safe statement is that l always denots some kind of 2nd person.

The 3rd person is again comparatively easy. Apart from a shift h>s from Qenya to Quenya, s remains the characteristic consonant of the 3rd person singular whereas t in all conceptual phases stands for the 3rd person plural (in the verb paradigms again supplemented with plural markers l,n to result in -lte/-nte, in dual with an apparent dual marker s to yield -ste. This, however, has to be a peculiarity of Quenya and is presumably not a feature of the underlying Common Eldarin forms, since in Sindarin we can find 3rd person plural pronouns based on s, cf. hain 'them' (LOTR) and singular pronouns derived from t, cf. den '*it' (VT44:21, 22) or dîn 'his' (SD:128) (note that the Sindarin forms are subject to lenition where they occur, hence unmutated forms would involve s,t). tar 'thither' and its inflection in this context seems to confirm this picture - it can be interpreted as a relic form from a time when in Common Eldarin t-based pronouns could also be singular. Yet singular pronouns based on t don't seem to be completely irrelevant for Quenya - in VT42:34 we find tai 'what' which can be interpreted as a compound of a demonstrative with a relative pronoun ta + i 'that which' - so in a demonstrative sense, t apparently remains relevant in Quenya 3rd person singulars, only in personal pronouns it marks plural.

The verb paradigms also show impersonal forms of the verb (these appear when the subject is not expressed in a verbal ending or when the sentence has no subject). The corresponding endings are zero in singular and the plural marker -r and the dual marker -t in the other numbers. By definition, there are no corresponding impersonal forms of personal pronouns.

The choice of vowels

The choice of the vowel which is used to form the simple independent pronoun forms is remarkably constant. We find i in the 1st person singular, e in 1st person plural and 2nd person and gender determines the choice in the 3rd person. Early Qenya has a full three gender distincion between masculine u, feminine i and neuter a, the later paradigms show a simpler concept in which presumably e stands for both masculine and feminine and a covers the neuter. This idea is also seen in the verbal paradigm in PE17:57 where the animate 3rd person forms show an ending -e but neuter 3rd person endings show -a.

During a specific time period, Tolkien did experiment with the idea that o is a marker of dual forms. This is clearly observed in the verbal paradigms PE17:75 and PE17:132, e.g. in the comparison between 2pl -lle and 2d -llo. However, ultimately this idea was abandoned, and it is unclear what effect it would have had on the shape of independent pronouns.

There are also variant forms in which a long vowel is observed - for example VT49:51 shows in the 1st person sg, tyé and lyé in the 2nd sg. and in the 3rd sg. with a separate neuter form .

The shape of the inflected forms

In Early Qenya, pronouns in accusative are unhyphenated but otherwise identical in form with their nominative counterparts (PE14:53). However, in late Quenya, a form tye makes its first appearance in Atarinya tye-meláne. 'My father, I love thee' and A yonya inye tye-méla 'And I too, my son, I love thee' (LR:61). Apparently tye translates as the accusative 'thee' in these examples. This leaves three distinct possibilities, and Tolkien may have toyed with all of them.

tye may simply be an alternative form of the ke of Early Qenya. If so, *inye méla ke would likewise be acceptable grammar for 'I love thee'. The ó-paradigm seems to confirm this idea, mentioning #nye and #lye in what seem to be alternative forms of #ni, #le.

tye may have replaced the former ke - this is what can be argued from the im- paradigm as well where we find #tye alongside with #le and no alternative form is given.

Finally, tye may be the form ke takes when used as object (or possibly when inflected for case). In this case, we would have *ni méla tye 'I love thee' but *ke méla nye 'thou lovest me'.

We do not know for sure if forms like *mye or sye > *rye derived from other characteristic consonants would exist. As far as we know they could comply with Quenya phonology, but it has to be stated that such forms are not in fact observed. Similar forms occur as verbal endings, but all verb paradigms only list -nye, -lye and -tye which would argue against the idea.

A closer look at inflected non-emphatic pronouns doesn't help to find a decisive answer regarding a possible distinction between tye and ke. We find olesse 'with you' (VT43:26) where the case endings are appended to #le, but this wouldn't be surprising since in the ó-paradigm le/lye seem to be alternative forms. On a postcard in posession of Carl F. Hostetter, we find the form lyenna (Lambengolmor 758) where #lye acts as the inflected element (a discussion of the form is now also found in VT49:40). In orenya quete nin (VT41:13) the dative inflection -n is added to ni, not nye. The earlier versions of the Ataremma (I - IV, VT43:9ff) have men in dative 'for us, to us', in the later versions (V, VI) this merges with the imperative in ámen anta 'give to us'. Likewise, Ataremma I-IV contain the form mello 'from us' with the pronoun me in ablative. Ataremma V and VI contain tien, apparently dative 3rd person plural 'for them'. This form is interesting as it seems to indicate that case endings are not appended to te (the base for in the paradigm) but rather to tie-. However, in the Ambidexter Sentence variants AS4-7 the variants ten/téna are found in the same role (VT49:7f). A short sentence quoted in VT49:14 antanen hatal sena 'I cast a spear at him' contains 3rd person se 'he' with a variant dative inflection -na. Finally, in VT44:12 omesse 'on us' can be found, involving both a prefix o- and the locative -sse with the pronoun me.

All in all it seems as if usually the inflectional endings are appended just to the base form of the pronouns, but Tolkien toyed with both variants. We may also study the (uninflected) accusative. In fact, some people have argued that forms like inte would involve the accusative form of the pronoun and not the nominative based on the fact that the underlying elements are only observed in accusative in Quenya. While this is possible, there is however a more likely explanation: Unlike Qenya, Quenya grammar involves pronominal verb endings, so there is no real need for a non-emphatic pronoun in nominative any more. When a verb ending is not sufficient, usually emphatic pronouns occur, making the non-emphatic pronoun in nominative to some degree obsolete in the actual use of the language.

Several non-emphatic forms apart from tye in accusative are found: We find te 'them' in A laita te, laita te! 'Bless them, bless them!' (LOTR), which seems identical with the nominative form as extracted from the im-paradigm and doesn't change to *tye - just possibly to avoid the clash with the 2nd person, but other reasons are more likely. The Ataremma I-IV contain me 'us', in Ataremma V and VI this again shows up in the compound forms álame tulya 'do not lead us' and áme etelehta 'but deliver us'. Note that the earlier versions of the Ataremma combine a 2nd person marker -lye with the imperative marker to indicate to whom the phrase is spoken, this is different in the late versions of the Ataremma, being the chief reason why Ataremma I-IV and V,VI form two distinct groups in this investigation. Compare e.g. alye anta men (Ataremma II-IV) with the later ámen anta (Ataremma V,VI) where the 2nd person remains implicit. This conceptual change is a possible source of confusion. Namárië has the interesting dual form met 'the two of us'. A brief phrase in VT49:15 contains melin sé 'I love him' with a long variant form of the 3rd person object form.

Thus, as far as we can judge from both the paradigms and the attested forms, no good case can be made that it would be necessary to use the longer forms in general for inflection. However, it is quite likely that it is permissible to use at least some of them as variants in all cases, thus both lyenna and *lenna may be permissible.

3. Pronominal verb endings

Early Qenya had a system in which verbs were inflected for person by prefixing the non-emphatic pronoun to the verb (PE14:52,56), we find e.g. ha-tule 'it comes' (PE14:52).

However, by the time of Fíriel's Song (about 1940) the picture has changed - we encounter forms like meláne 'I love' (LR:61) with a 1st person ending -ne or the more complex antaróta 'he gave it' (LR:72) from which we may infer a 3rd person *-so > -ro 'he' and -ta 'it'. Conceptually, this is almost the system of late Quenya - pronouns are expressed by verbal endings, and two pronominal endings in a row denote subject-object. The only difference is that late Quenya has short as well as long pronominal endings, and in subject-object constructions always a long ending denotes the subject and a short ending the object, cf. utúvienyes 'I have found it' (LOTR) involving long 1st person sg. -nye and short 3rd person sg. -s, laituvalmet 'we will praise them' (LOTR) with long 1st person plural -lme and short 3rd person d. -t, tiruvantes 'they shall keep it' (UT:305) with long 3rd person pl. -nte (note the plural marker n here as opposed to -lte in the verb paradigm VT49:16) and short 3rd person sg. -s, leltanelyes 'you sent him' (VT47:21) and camnelyes 'you received it' (VT47:21), involving long 2nd person -lye and a short -s, in the same paragraph denoting both male and neuter.

Interestingly, not all persons in singular show a short and a long form in all verb paradigms. For example, in VT49:16 the 1st person has both -nye and -n, the second person formal -lye/-l, but the informal only shows the long -tye whereas the 3rd person only shows the short -s. Thus, in this conceptual phase of Quenya, there were some combinations of subject-object which could not be expressed without the help of independent pronouns. On the other hand, the verb paradigm given in VT49:48 shows variants -tye/-t but in 3rd person still only -s.

Tolkien apparently also oscillated back and forth between personal prefixes and endings: In SD:56 we find a change from maruvan 'I will abide' (using a verbal ending) to nimaruva (employing a prefix). However, especially in later texts pronominal endings seem to be the usual choice.

There is not an extremely wide variety of attested forms in Quenya texts, but as far as we can tell there is no difference in meaning if a long or a short verbal ending is used (as long as both variants exist) and all known paradigms support this: We see hamil 'you judge' (VT42:33) and hiruvalye 'thou shalt find' (Namárië) involving both the short ending -l and its longer variant -lye. Likewise, short endings used as subject or object seem to be identical, we have eques 'said he' (WJ:415) involving a 3rd person sg. -s which agrees with the object form quoted above assuming the short verbal ending in 3rd sg. would be generless. The generalization of these findings seems to be justified: We do not have an attested 1st person sg. ending -nye in subject form only, instead we find the short ending -n in e.g. maruvan 'I will abide' (LOTR) and the long ending combined with a short object in utúvienyes 'I have found', but there is no reason to assume that forms like *maruvanye 'I will abide' or *utúvienyen 'I have found me' would not be valid. Variants melinyes / melin sé 'I love him' are found in VT49:15.

Based on this observation, we can combine attested subject and object forms to make a list of possible short verbal endings (as the verb paradigms show, not all of them were realized in each conceptual phase of Quenya). In 1st person sg. we have -n as apparent from e.g. maruvan 'I will abide' (LOTR). A 1st person plural form would logically be -m, but such an ending cannot happen in Quenya (see discussion of allowed final consonants in Letters:425). The usual phonetic shift would lead to -m > *-n (cf. talan pl. talami (LR:390)), but since this clashes with 1st person sg. this particular short ending is probably not realized at all, given that in addition the distinction between exclusive, inclusive and dual 'we' cannot be made. The 2nd person is attested in hamil 'you judge' (VT42:33) and in the pronominal elements -t, -l mentioned in WJ:364 - the assignment of the underlying consonants to person/number is variable as discussed above. In the 3rd person sg., -s is attested both for 'it' (caritalyas 'your doing it' (VT42:33)) and 'he' (eques 'said he' (WJ:415)). In plural, -t is known from Ataremma VI emme apsenet 'we forgive them' and as object in laituvalmet 'we shall praise them' (note that in the latter case, Tolkien insists in PE17 that -t is actually a dual, but clearly at some earlier conceptual stage it was a plural). There is a clash between -t in 2nd person sg. and 3rd person pl./d. - some writers have argued that the forms cannot coexist. In reality, that hardly seems to be a problem given that e.g. the German 'sie' can denote 2nd person sg./pl., 3rd person sg. female and 3rd person pl. which still doesn't render the language useless. The power of context to resolve ambiguities is often not sufficiently appreciated. In fact, the verbal paradigm in PE17:57 clearly proves that Tolkien did not view the clash as a problem, as -t is allowed to stand in two different roles in the table.

The systematics of the long verbal endings is more complex. We can separate them into two distinct groups - one in shape identical with the variant forms of the non-emphatic pronouns, including -nye, -lye, -tye and possibly (though increasingly unlikely) *-rye. The other group has no counterpart in the non-emphatic pronouns.

The first group of long endings just employs a vowel e, the characteristic consonant and an additional y. In the second group, the y is replaced by some additional information-carrying element. For example, the long 3rd person plural ending -nte involves in addition to the 3rd person consonant t a plural marker -n (which is also seen in plural case endings -ron, -llon, -ssen, cf. the Plotz Letter). This makes the ending distinct from the 2nd person sg. long ending -tye Tolkien's different ideas about the role of the plural marker then result in endings such as -lte or -lde instead.

The long 1st person plural endings likewise seem to have a straightforward interpretation in one conceptual phase before l was reassigned to a plural marker and the shift in meaning of the 1st person plural/dual forms mentioned above occurred: -lme involving the elements for 'you' and 'we', hence inclusive 'we', -mme repeating the 'we' (we and we only), thus being inclusive and *-lve (deduced from the possessive ending) possibly involving a dual u/w element. However, as already mentioned, if that interpretation was ever Tolkien's true intention, it was subsequently changed.

Tolkien's latest ideas as observed in VT49:16 seem to involve a plural marker -l which is used with the characteristic consonant to find the plural forms. Thus, an independent exclusive 1st person plural pronoun #we becomes the verbal ending -lwe in this way.

This also agrees with the 1st person pl. ending -lle (discussed also in VT38:6f) as seen used in various places. In SD:47 the verb laitalle 'you praise' can be found with probably a plural 'you' ending -lle. The same may be found in VT24:5 where there is a draft version of Namárië given as nai hiruvalle Valimar (instead of nai hiruvalye Valimar as in LOTR).

Tolkien's ideas with regard to the formation of dual forms are, however, less clear. There seem to be at least three mechanisms at work: 1) a doubling of the characteristic consonant as seen in -mme, -cce, -lle and -tte in PE17:57, 2) the final vowel -o as discussed above and seen in e.g. -mmo, -llo and 3) a dual marker s leading to -ste or -star where the corresponding plurals are -lte/-nte and -ltar. In VT49:33 however -ste is also derived via reduplication of a characteristic consonent from dd > zd > st and tt > st. Finally, -ngwe/-nque apparently exhibits yet another way of forming dual.

4. Independent emphatic pronouns

The idea of emphatic pronouns goes back as far as the Early Qenya Grammar - the forms described in PE14:53 are quite different from the emphatic pronouns in late Quenya. In PE14:53 we find:
  Sg. Pl.
1st nîmo melmo
2nd kêto lelko
3rd m húyo tûto
3rd f hie sîse
3rd n hea atta
Both the formation and the inflection of independent emphatic pronouns in later Quenya seem relatively straightforward given what we know about verb endings. The pronouns are formed by appending the long verbal ending to a vowel, this vowel is i- in the first person sg. and e- otherwise. Case endings seem to be directly appended to these forms.

We find quite a number of uninflected emphatic pronouns which do not show substantial changes from the Qenya of 'The Lost Road' on: elye 'thou' (Namárië), elle 'thou' (VT24:5), inye 'I' (LR:61), emme 'we' (VT42:8ff, Ataremma I-VI). We lack attested 3rd person forms, but possibly we might have *esse, *ente/*elte, *este.

Case inflection can be seen in emmen 'for us' (VT43:12, Ataremma V,VI). Given that often the emphasis is on the subject, we should probably not be surprised to find this set of pronouns mainly in non-inflected situations, but it doesn't seem very far-fetched to inflect e.g. to *inyenna 'to me'.

5. Possessive pronouns

The system of possessive pronouns outlined in Early Qenya (PE14:54) already has a distinction between independent and enclitic forms. In fact, the system of possessive pronouns appears rather more stable than the system of independent pronouns - especially the 1st person singular seems to be created in almost its final shape. However, rather than being endings, the possessive pronouns in Early Qenya (like the verb inflectional elements) are prefixes:

  Sg. Pl.
1st nya- mea-
2nd tya- lea-
3rd m hwa- tua-
3rd f hya- sia-
3rd n ha- ta-
The distribution of characteristic consonants agrees (naturally) with the system outlined for the nominative pronouns. In addition, there are independent forms of the pronouns:

  Sg. Pl.
1st ninya menya
2nd ketya lelya
3rd m húva túva
3rd f íva síva
3rd n áva táva
These forms are described as being fully declined adjectives.

By the time of 'The Lost Road' we find that the picture has slightly changed - like in the case of verb inflection, the possessive forms have become endings as apparent from forms like Anarinya 'my sun' (LR:72) or atarinya 'my father' (LR:61). Note the use of -i- here as a connecting vowel. The preference of i instead of e as both connecting vowel and 'filling' vowel (cf. inye but emme, ni but le) seems to be peculiar of the 1st person sg. throughout most of Tolkien's writings.

The independent pronouns are likewise realized in this conceptual phase: Fíriel's song has indo-ninya 'my heart' (LR:72). For all we know, independent possessives occur rarely in later sources but are not obsolete: The Ataremma I-IV involve menya 'our' as an independent form (this is absent in V and VI as the text has been rephrased and doesn't involve a possessive at all, hence this doesn't indicate that the possessive pronominal form as such has become obsolete). The close agreement of these forms with the ones in Early Qenya makes it likely that *cetya, *lelya would also be realized in the later stages of the language. The shape of 3rd person independent possessives doesn't seem to be straightforward.

Possibly the system of Sindarin possessive adjectives can shed some additional light on this question as the forms may well be related. In Sindarin, we find nîn 'my' (VT44:21f) , lîn 'thy' (VT44:21f), #tîn 'his' (SD:128), #mîn 'our' (VT44:21f) and in addition a reflexive în 'his own' (SD:128). This may point to Quenya forms ninya 'my', *lenya 'thy', *tenya 'his', menya 'our' and *enya 'his own'.

Let us now discuss the attested forms of possessive endings in late sources. Apart from the assignment of the 1st person plural to inclusive, exclusive and dual meaning the rest of the system seems hardly changed over time. As far as we can see, most of the time the system of possessive endings can be deduced from the long verbal endings by making the vowel replacement e > a, but VT49:16 shows one important counterexample: While the 3rd person sg. ending is just -s, the corresponding possessive ending is given as -rya, the counterpart of an unrealized long ending **-rye. Interestingly enough, the paradigm also shows impersonal possessives -ya (sg) -rya (pl) and -twa (dual) which are however somewhat outside the scope of this article.

Complete tables of forms are provided in PE17:57 and PE17:132, here we only reproduce the first of those:

  Sg. Pl. Du.
1st incl.   -lma -nqua
1st excl. -nya -lma -mma
2nd fam. -tya †-nca †-xa,-cca
2nd form. -lya -lda,-lla -lla
3rd -rya -nta -twa
A comparison with PE17:57 reveals that these forms follow from the verbal endings indeed by the replacement of the final vowel e > a, with the exception of 3rd person sg. -rya and 3rd person dual -twa. PE17:132 which contains verbal forms where dual is marked with -o show that in this case the corresponding possessives can be found by the replacement o > a. The novel possessive forms in this table are the 2nd person familiar sg. -taro and pl. -tarion. These appear to be genitive inflection in singular and plural applied to the verbal ending.

Apart from VT49:16, the 1st person sg. -nya is attested in a post-Lost Road source in Hildinyar 'my heirs' (LOTR) (again with -i- as connecting vowel). 2nd person forms can be found frequently in the Ataremma or the Aia Maria. A non-inflected example would be e.g. esselya thy name' found in Ataremma I-VI. No possessive ending -tya is attested in a text. The 3rd person sg. can be read off from máryat '(both) her hands' (Namárië) as -rya. The various 1st person plural forms are apparent as -mma from ataremma 'our father' (VT43), #-lma, #-lva from omentielmo/omentielvo (LOTR, first/second edition and discussion in Letters:447) 'of our meeting', here inflected for genitive.

Usually the case inflection endings are appended after the possessive ending. In the case of short endings like the genitive -o which changes the last vowel or dative -n no other choice seems possible, but in the case of the longer allative -nna, ablative -llo and locative -sse, in principle the case ending could come before the possessive. In singular this is never seen - we have e.g. tielyanna 'upon your path' (UT:22) or ortírielyanna 'to thy patronage' (VT44:5). However, in plural we have two attested examples where the case ending precedes the possessive ending which is in turn followed by a plural marker. These examples are rocindillomman 'those who sin against us' (VT43:11) and sangiessemman 'in our needs' (VT44:5). Note that ortírielyanna and sangiessemman occur in fact in the very same text, thus a conceptual change is not an option. It may be that in plural this is indeed the preferred order of these three endings in this conceptual phase, or it may be that the order is optional - but it seems hardly likely that this inverted order is realized in general in plural, examples like dative plural ?ciryainenya(r) which require additional connecting vowels seem hardly likely. In the later Ambidexter sentence (VT49:6ff) the form símaryassen 'in their imagination' occurs in which the possessive -rya precedes the case ending as seen in the majority of examples.

6. Final remarks

If you have read through this article to find at this point a complete reconstruction of the Quenya pronominal system, you are going to be disappointed. Such a thing is impossible without specifying a timeframe, as all the complete paradigms given by Tolkien himself should have made abundantly clear. Instead, I hope to have mapped out how Tolkien's ideas about the Q(u)enya pronominal system changed over time, with a constant re-interpretation of already published forms and invention of new ones. I trust the dilligent reader in need of a complete pronominal table is able to read off the principle of construction of the forms from the discussions above. Armed with this set of tools, it is relatively straightforward to construct the set or pronominal forms corresponding to a given assignment of characteristic consonants.

7. Acknowledgements

Observations made by Carl F. Hostetter, Helge Fauskanger, Helmut W. Pesch and Roman Rausch and others have made me aware of some intersting bits of evidence. Im am grateful to the editorial team of Parma Eldalamberon and Vinyar Tengwar - without their efforts a study like this would be impossible.

Thorsten Renk

Back to the index