Common Eldarin views on the Sindarin pronominal system

The purpose of this article is to demonstrate that the essential features of the Sindarin (S) pronominal system can be predicted from a) the shape of the pronominal system in Quenya (Q) b) the phonetical shifts going back from Quenya to Common Eldarin (CE) and c) the phonetical shifts going from Common Eldarin to Sindarin. This can be done without referring to the attested Sindarin pronouns and their attested use and is hence rather independent of grammatical interpretation (only well-known phonetical shifts are applied). Only in the actual matching of reconstructed forms to their grammatical interpretation one has to make use of the Sindarin corpus. We will refer to The Quenya Pronominal System to justify all attested and hypothetical Quenya pronouns in this article.

Part I - Nominative Pronouns

1) Characteristic consonants and basic elements

Each elvish pronominal form is characterized by one consonant. This consonant is, however, subject to phonetical shifts, hence we will refer to the characteristic consonant of some pronoun as the one appearing in the CE root of this pronoun.

From the Quenya short pronominal verb endings, the shape of the independent non-emphatic pronouns and the Etymologies, we may draw up the following list of characteristic consonants in CE:

1. Pers. Sg.: n (LR:378)
2. Pers. fam. or Sg.: c (evidence from LR:61 and WJ:364 to be t in Quenya => c in CE)
2. Pers. formal. or Pl.: l (e.g. WJ:364 shows this to be l in Quenya)
3. Pers. Sg.: s (LR:385)
3. Pers. Sg.: t (LR:389)
1. Pers. Pl.: m (e.g. from Namárië)

Remarks:

a) We do not distinguish 2nd person in sg. and pl. at this stage: There is evidence that Tolkien changed his mind back and forth about the meaning of the 2nd person pronouns.
b) in the 3rd person, the formation of the pl. does not appear to be realized by a change of characteristical consonant but rather by adding the CE plural marker i. Nevertheless, Quenya shows a preference for using s as verbal ending in Sg. and t in pl., possibly deviating from the original CE use of the forms.

From these characteristic consonants, Quenya short verbal pronominal affixes and non-emphatic pronouns are constructed. Their origin is as follows: The pronous in CE are constructed appending a vowel to the characteristic consonants. This vowel is i in the 1st person, a for 3rd person t. The 3rd person s can apparently discriminate male o, female e and things a. The vowel is e else. In CE, these independent pronouns double as verb conjugational endings, leading to the Quenya short pronominal forms:

This is shown here using the example of linda- (to sing):

CE: ni lindâ or *lindâ-ni => Q: ni linda or lindan (I sing)
CE: *ke lindâ or *lindâ-ke => Q: *ce linda or lindat You sing; Q can't have a -c in final position)
CE: *le lindâ or *lindâ-le => Q: le linda or lindal (You sing)
CE: sa lindâ or lindâ-sa => Q: sa linda or lindas (It sings)
CE: ta lindâ or *lindâ-ta => Q: ta linda (It sings, ?lindat is used in 3rd person Pl.)
CE: *me lindâ or *lindâ-me => Q: me linda or *lindam (we sing; a short verbal ending -m is not attested in Q)

However, these short forms involving only characteristic consonant and a vowel are not the only relevant ones: Quenya also shows three-letter clusters as long pronominal verb endings and uses the same forms for emphatic independent pronouns. Evidently, some of these clusters carry more information than the short endings (see my article cited above). The relevant list (again using the example linda) is:

Q: lindanyë (I sing) inyë linda (even I sing)
Q: *lindatyë (you sing) *etyë linda (even you sing)
Q: *lindaccë (you sing) *eccë linda (even you sing)
Q: lindalyë (you sing) elyë linda (even you sing)
Q: lindallë (you sing) *ellë linda (even you sing) (this denotes 2nd person pl.)
Q: *lindaryë (he sings) *eryë linda (even he sings) (used without gender distinction, derived from *-syë)
Q: lindaro (he sings) *ero linda (even he sings) (used as male pronoun)
Q: lindarë (she sings) *erë linda (even she sings) (used as female pronoun)
Q: lindalmë (we sing) *elmë linda (even we sing)
Q: lindammë (we sing) emmë linda (even we sing)
Q: lindalvë (we sing) *elvë linda (even we sing)
Q: lindantë (they sing) *entë linda (even they sing)

Remarks:
a) We do not discuss the interpretation of the various forms for 'we' here - Tolkien changed his intentions here more than once.
b) The forms -ro and -re might lengthen a preceding vowel.

The question is this if this elaboration on the single characteristic consonants represent a feature already present in CE or if they are of purely Quenya origin. Making the conjecture that they are not present in CE, we may tentatively test this by trying to reproduce Sindarin verb inflection from the simple forms only:

CE: *lindâ-me -> *linna-ve -> *linna-v -> S: **linnaf

This is not observed, however

CE: *lindâ-mme -> *linna-mme -> *linna-m -> S: linnam (a double mm cannot be mutated)

is, so we should include the elaborated version of the characteristical consonants also in the CE verb conjugation.

2) Phonetical changes of the basic elements

Thus, in order to proceed, we have to investigate what shape the individual clusters observed in Quenya would assume in CE and Sindarin (this is before mutation and loss of final vowel in Sindarin):

Q: nye CE: nye S: ne (cf CE: NYEL- (sing), Q: nyellë (bell) S: nell (bell))
Q: tye CE: cye S: ce (cf CE: KYELEP- (silver) Q: tyelpë (silver) S: celeb (silver))
Q: cce CE: cce S: che (cf. Q: rocco (horse) S: roch (horse))
Q: lye CE: lye S: le
Q: lle CE: lle S: lle
Q: rye CE: sye S: se (cf. CE: PIS-, Q: pirya (syrup))
Q: ro CE: so S: so
Q: re CE: se S: se
Q: mme CE: mme S: me
Q: nte CE: nte S: nte (cf. Q: anta- (to give) S: ant (gift))

3) Verb inflexion and pronouns in Sindarin

Hence, to give some examples what the Sindarin verb endings should look like:

CE: *lindâ-nye -> *linna-ne -> *linna-n -> S: linnon (not discussing the shift a->o in 1st person sg. here)
CE: *lindâ-cye -> *linna-ge -> *linna-g -> S: *linnag
CE: *lindâ-lye -> *linna-le -> *linna-l -> S: *linnal
CE: *lindâ-cce -> *linna-che -> *linna-ch -> S: *linnach
CE: *lindâ-sye -> *linna-he -> *linna-h -> S: linna
CE: *lindâ-nte -> *linna-nte -> *linna-nt -> S: ?linnant
Once again, I would like to emphasize that there are good arguments that *linnal and *linnag could rather result in *linnog and ?linnol whereas the longer CE form lle would then create the distinct *linnal - this is however not the main issue of this article.
Furthermore, it is unclear if the elaborated forms -nyë, -lyë, -syë, -cyë would really be responsible for the observed endings in Sindarin. I either case, the result does not change if one assumes CE inflection with the short variants -ni, -le, -se, -ce instead.

In any case, just applying rather trivial phonetic shifts already reproduces the whole range of observed Sindarin verb endings. Note that the 3rd person pl. ending would clash with another form: The past tense. Therefore, there is actually a rather natural reason why this form should be replaced by using the plural marker -r to denote 3rd person plural.

We continue by demonstrating the resulting forms of applying the phonetical changes to the Quenya emphatic and non.emphatic pronouns:

Q: inye CE: *inye S: *in
Q: *etye CE: *ecye S: *eg
Q: *ecce CE: *ecce S: *ech
Q: elye CE: *elye S: l*el
Q: *elle CE: *elle S: *el
Q: *erye CE: *esye S: e
Q: *ero CE: *eso S: e
Q: *ere CE: *ese S: e
Q: emme CE: *emme S: *em
Q: *ente CE: *ente S: ?ent

This is already similar to what is observed - we find a general 3rd person form e with no gender distinction (seen from the fact that ho is glossed "he" and he "she" in the etymologies, it is rather surprising to find e "he" in the King's Letter) and another form which almost looks like im (I). However, this form would clash with the plural definite article (which is in rather heavy use in Sindarin), so this would offer a motive to replace the consonant by the 1st person pl. characteristic consonant, the difference still being manifest from the peculiar choice of the vowel i in the 1st person sg.

Q: ni CE: *ni S: *ni
Q: *ce CE: *ce S: *ce
Q: le CE: *le S: *le
Q: me CE: *me S: *me
Q: so CE: *so S: *so
Q: se CE: *se S: *se
Q: sa CE: *sa S: *sa
Q: ta CE: *ta S: *ta

None of these forms is observed in Sindarin in nominative use, however the entry to the stem S- seems to indicate that Tolkien imagined lenition in nominative pronouns by the time he wrote the Etymologies, allowing to identify *so, *se, *sa with ho, he, ha. The rest of the non-emphatic pronouns seems a solid base for the more reliably attested formation of object pronouns which we will treat in part two of this project.

So, this is about as far as one gets by applying phonetical arguments: We find a) the complete range of Sindarin verb endings b) several emphatic pronouns which can be (partically) matched with attested forms c) several non-emphatic pronouns which seem to form a good starting point for the formation of direct objects. Please note again that no clever guesswork about grammatical interpretation of Sindarin is involved: The underlying concepts involve phonetical shifts which are accessible for some 1000 entries in the Etymologies where both Q and S forms are listed and which are also the root of the mutation tables known to the Sindarin writers.

4) Some comments

Before we proceed to the investigation of the pronouns used as verb objects, let us investigate several of the more obscure points - the formation of 3rd person plural in Quenya and the fate of the general verb forms in Quenya when going to Sindarin:

Quenya apparently does not show a consistent formation of the 3rd person plural: In verbal endings, the short form Q: -t and the longer form -ntë appear, showing the characteristical consonant t. This consonant, however, is not as such associated with the plural as the attested pronoun ta (it) shows. The long affix -ntë can be interpreted as the characteristical consonant in addition with a plural marker n (not to be confused with the characteristical consonant n denoting 1st person sg.) to clarify that plural use is intended. However, the short affix -t does not make this distinction. One may argue however that the attested indepentent 3rd person plural pronoun te is in fact derived from *tai employing yet another (well-known) plural marker i. There is apparently a strong preference to express the 3rd person plural with t and the singular with s - the presence of plural markers however might indicate that this is indeed a development unique to Quenya and not present in CE.

In summary, although Quenya usually associates s with 3rd person sg. and t with 3rd person pl., there no reason to assume that CE does likewise - the presence of plural markers indicates that originally 3rd person pl. was expressed by adding plural markers to 3rd person sg. forms.

In order to discuss the second issue, we have to note that Quenya has a general verb form (to be called QG in the following) which is used if the subject, either pronoun or noun, is not part of the verb ending. This general form can only distinguish between singular (QGS) and plural (QGP). The plural is indicated by the use of the plural marker r (which is often used to form plural for Quenya nouns). (In the following, we will refer to 3rd person forms in Quenya as Q3S (singular) and Q3P (plural).)

Thus, we find
lassë lanta (a leaf falls) (QGS)
lassi lantar (leaves fall) (QGP)
ni lanta (I fall) (QGS)
sa lanta (it falls) (QGS)
lantas (it falls) (Q3S)
emmë lantar (we fall) (QGP)
lantantë (they fall) (Q3P)

for A-verbs and

elda carë (an elf makes) (QGS)
eldar carir (elves make) (QGP)
ni carë (I make) (QGS)
sa carë (it makes) (QGS)
caris (it makes) (Q3S)
emmë carir (we make) (QGP)
carintë (they make) (Q3P)

for simple verbs. Conceptually, it is arguably easier to confuse QGS with Q3S than with any other singular pronominal form ("falls" vs. "he/she/it falls" ), because sentences involving a verb and not expressing the pronoun in the conjugational ending are rather rare. Similarly, the 'natural partner' to confuse QGP with is Q3P ("fall" vs. "they fall" ).

Thus, in Quenya we find two distinct forms which we can put through the phonetic changes to find their possible Sindarin equivalents:

A-verbs:

Q3S lindas yields S3S linna
QGS linda yields SGS ?linn (or maybe ?lind)

Q3P lindantë yields S3P ?linnant
QGP lindar yields SGP linnar

Simple verbs:

Q3S caris yields S3S ?ceri
QGS carë yields SGS *car (this is almost câr)

Q3P carintë yields S3P ?cerint
QGP carir yields SGP cerir

Obviously, in Sindarin, only one of two possible forms can be found in each case, so possibly S3S and SGS have merged, and so might SGP and S3P (meaning dannar may presumably express both "fall" AND "they fall" ). But for three of four rejected forms, we are able to provide a plausible reson why these forms should in fact not be preferred:

SGS for A-verbs clashes with the formation of nouns from verbs, cf. the pattern maetha -> maeth (fight), which in the case of A-verbs is often done by dropping the final -a.

S3P for A-verbs produces the clash with past tense formation mentioned above.

S3S for simple verbs finally clashes with the infinitive formation attested in the Etymologies.

A final remark: From the nature of this general form, it is clear why we may not expect a CE pronoun **re or similar: Verbal forms in -r would only appear if there is some subject mentioned in the sentence. But there is no need for a particle which can only denote that it itself is plural but nothing else in a language.

Part II - Pronouns used as Objects of a Verb

1) The curious use of -n

We are now past the stage where rather mindless application of phonetical shifts would do all the work, so now we have to actually observe the use of pronouns in Sindarin sentences and draw conclusions from these observations.

We may consider the following examples:

A tiro nin, Fanuilos! (O look towards me, Everwhite!) (translated in RGEO:72)
caro den i innas lin (your will be done; lit. persumably 'make it thy will')
edro hi ammen (open now for us) (LotR1/II ch. 4, translated in RS:463)

From the first example, we may extract nin with the translation 'towards me'. However, it is possible that this is a rather liberal translation ('towards' would normally be expressed using the preposition na) and that the literal meaning is more 'O watch me!', hence the phrase is likely to involve an accusative pronoun nin.

The second example is not really unambiguous, but if den represents a pronoun here, it is again likely to be a direct object of caro in accusative.

The third example involves the (rather transparent) compound an + men -> ammen, which is used to express the idea 'for us, for the benefit of us', hence we can only extract the form men but not draw any conclusions about its use (there are several more occasions where this particular compound is found, but those do not change the conclusions).

Although none of these examples is particularly clear, the easiest explanation is that pronominal forms used as a direct object of a verb in Sindarin end with an -n. This is a commonly used assumption and we will not contradict it here. This theory also allows a further clarification of the entry S- in the Etymologies:

S- demonstrative stem. sû, sô he (cf. -so inflexion of verbs); sî, sê she (cf. -se inflexion of verbs). Cf. N ho, hon, hono he; he, hen, hene she; ha, han, hana it; plurals huin, hîn, hein.

Making use of our previous conclusions, ho presumably represents nominative 'he' and hon accusative 'him' (and the other forms can be treated correspondingly). This indicates that hono should in fact not be taken as a variant of ho, although we will postpone its interpretation until part three of this article. It is rather curious that plurals are only given for the forms ending in -n - does this imply that ho, hon, hono all form a common plural huin? Let us postpone this question also.

If we combine our indepentent pronominal forms and combine them with the idea of forming an object by the affix -n, we get:

S: nin
S: *cen
S: *len
S: men
S: *son
S: *sen
S: *san
S: *tan

Taking into account that direct objects of a verb are lenited in Sindarin (we will not prove this here), we again can identify *son, *sen, *san with hon, hen and han from the Etymologies. Thus, this simple idea works remarkably well - only *tan (or lenited *dan) seems to disagree with the attested den. Lacking a fundamental explanation, I will make the assumption that this form kept its characteristical consonant, but the vowel was rather 'regularized' to e in the evolution towards Sindarin in order to agree with the other forms. So we would have *te (it, nominative) and ten (it, object).

One might think of trying the same procedure with the emphatic versions of the pronouns, but such constructions are rarely observed in Quenya and never in Sindarin, so we do not attempt to address this question at all.

So - let's analyze the main difficulties of the otherwise nice hypothesis that -n forms an object pronoun: 1) Sindarin does not have a case inflection and 2) in Quenya, pronouns like nin and men are also found, but they denote dative objects.

It is very likely that the same CE element is responsible for the Quenya case inflection -n and the Sindarin preposition an. Hence, originally the forms in -n might also have expressed dative in Sindarin. However, the distinction between direct object in accusative and in dative is somewhat blurred in Sindarin, as apparent from

Anno ammen i mbas ilaurui vín (Give us today or daily bread) and
Ónen i-Estel Edain (I gave Hope to the Dúnedain)

The first example shows the use of the preposition an to express the Sindarin dative, but the second example indicates that it is also permissible to express that dative as a direct object without preposition, possibly word order alone discriminating between accusative and dative. Thus, the (lenited) form commonly referred to as 'direct object' in Sindarin is also capable of expressing the dative. Reversing this argument, an early dative form ending in -n might easily get confused with the original direct object and replace it in the long run. Regardless of the underlying reasons, however, the attested Sindarin examples show that forms in -n are likely to be used in this 'general' direct object sense.

2) Compound forms with an

In Sindarin texts, we find a second class of object pronouns which are only used in dative and can be easily identified as compounds of the preposition/dative element an with forms we've already discussed. Since we have argued that the -n seen in the object pronouns discussed above corresponds to the Quenya dative case inflexion, prefixing an has to be a genuine Sindarin construction and we cannot expect to draw on any Quenya analogies (except for the use of those forms of course, since the intention to express dative remains the same). In Sindarin texts, we find three different forms:

Anno ammen... (give us...) (VT44:21, 22)
ú-chebin estel anim (I have kept no hope for myself) (LotR Appendix A)
Guren bêd enni (my heart (inner mind) tells me) (VT41:11)

From what has been said before, one would identify ammen as an prefixed to the object form *men, anim as the prefixed form of the emphatic pronoun im and enni as the element an added to the non-emphatic pronoun *ni and subject to i-affection. A first technical question - why is there no i-affection in anim? Obviously, prefixes are sometimes not subject to i-affection, cf. the known infinitives esgeri (from osgar- (cut around)) and ortheri (from orthor- (conquer)). Maybe Tolkien used this freedom to avoid **enim (which probably doesn't bother most of the readers of this article except those who know that this corresponds to a Latin word 'namely' - Tolkien would have known...).

The translations given support the hypothesis that im is an emphatic form and ni is not: Clearly, 'for myself' is a stronger expression than 'me'.

It is curious why the form ammen has both a prefix an and a final -n (so the object form instead of the nominative one was used as the base of the construction here, possibly expressing the case inflection two times). The fact that this form occurs rather often in the corpus indicates that using the object form as base for such a construction is not really uncommon. So - maybe both enni and *ennin might be permissible.

Finally, note that compound forms with an- seem to be the more common way to express the dative and that they are never seen expressing an accusative object.

3) The exception: le

There is one form in the Corpus of Sindarin texts which does not fit in the pattern outlined above: le. It is seen in the following examples:

Fanuilos le linnathon (to thee, Everwhite, I will sing) (LotR:1/II ch. 1 and RGEO:72)
le nallon sí di-nguruthos (to thee I cry now in [lit. beneath] the shadow of death) (RGEO:72)
le linnon im Tinúviel (untranslated, probably 'to thee I sing, I, Tinúviel) (The Lays of Beleriand p. 354)

By saying 'it does not fit the pattern', one does not imply here that it would not show up in the reconstruction procedure outlined above: We do reconstruct a form le, however it is should be a non-emphatic nominative pronoun, whereas the examples above suggest that it is used as a dative (I [will] sing for the benefit of you).

How can we understand this? In RGEO:73, the pronoun is said to be of Quenya origin. While the reconstruction procedure outlined above might indicate that this holds in fact for all pronouns, there is one important conceptual difference: We have used Quenya forms in the analysis above to reconstruct the underlying CE forms, and from those we have constructed the Sindarin forms in turn. However, this pronoun is directly said to be derived from Quenya, hence it is different (Quenya dative would also be len though, so the Sindar must have misunderstood something when they took it over...). Anyway, this information sets the pronoun apart and for whatever reason Tolkien may have chosen this particular form, we will treat it as an exception to the usual use of object pronouns.

So - presumably instead of the more regular forms *len, *alle or *allen one should insert just le.

Part III - Possessives

1) Possessive endings

We will start the investigation of possessive forms with a phenomenon that is rather seldom seen in Sindarin - possessive endings. In VT:41, we learn that 'My heart tells me' is órenya quete nin in Quenya, ôre nia pete nin in Telerin and guren bêd enni in Sindarin (thus, an explicit link of the Sindarin guren to the Quenya órenya (my heart) is made). A second example is found in Gandalfs attempt to open the gates of Moria: here lammen (my tongue) occurs.

Note that a form like lammen would be very hard to interprete in Sindarin: It cannot be an ending -en on lam (tongue), since that would only result in **lamen. But it also cannot be the ending -men, since we do not see **gurmen or **gurven. From the evolutionary point of view, however, all makes perfect sense: In Quenya, this is lambenya and this becomes *lamme-na and finally S: lammen (thanks to Tolkien, I don't need to work out the second example here... he did it already - see above).

The dilligent reader will have no problem in verifying that the Quenya difference between possessive -nya and verb ending -nyë will be lost when going to Sindarin and that the resulting forms are in fact exactly analoguous to the verb inflexion, i.e. 'thy tongue' would become *lammel, 'our tongue'*lammem and so on.

This identification however has profound consequences: It is rather an accident that we seemingly observe the ending **-en in both cases: This is only because this is the original vowel in both cases. A form like Q: arda (realm) would however yield ardanya (my realm) and this is bound to become S: *ardhan. Similarly, Q: osto (strong place) would be Q: ostonya (my strong place) and become S: *oston.

In the case of words originally ending in a consonant, this might lead to an extra complication: Q and CE: atar would lead to Q: atarinya (my father), using the -i- characteristic of the 1st person sg. as a connecting vowel, and this would become S: *ederin (subject to i-affection), but 'thy father' would render Q: atarelya and S: *adarel.

Small wonder these forms are not often found in Sindarin but presumably only survive in archaic patterns of speech.

2) Possessive adjectives

The more common way to express possessives in Sindarin is by using forms which apparently behave like adjectives (i.e. they follow the noun and are lenited in this position). Compared to the rest of the pronominal system, we find numerous examples in the corpus:

mhellyn în phain (all his friends) (SD:128 )
Meril bess dîn (Rose, his wife) (SD:128 )
ered e-mbar nín (the mountains of my home) (UT:40, translated in UT:54)
Adar nín (my father) (VT:44)
i eneth lín (thy will) (VT:44)
i arnad lín (thy kingdom) (VT:44)
i innas lin (thy will) (VT:44)
i mbas ilaurui vín (our daily bread) (VT:44)
i úgerth vin (our wrong-doings) (VT:44)

With the exception of în, all those forms incorporate a characteristic consonant denoting the person, the vowel i and end in -n. Despite three different length patterns, i.e. i, í and î, we will assume that all those forms should be regularized to í: The distinction between í and î cannot be made in tengwar anyway, and í and i could easily get confused in Tolkiens handwriting - we will find arguments later on which show that this í should preferably be long in order to make sense in the overall scenario.

In order to match the attested possessive adjectives with the forms discussed so far, we should provide explanations for the following: a) a lengthening of the vowel b) the shift of the vowel e -> i (except for 1st person sg.) and c) an interpretation of the forms hono, hene, hana which we have not matched otherwise (not many pronouns are left now...).

By making a clever guess, we can reduce these three problems down to one (the last remains unexplained...): Suppose we start from the form hene and for some reason, the first vowel gets long and we have *héne. When we now reach the Old Sindarin (OS) stage, what happens is:

When not final, the quality of three of the primitive long vowels was altered in Old Sindarin: â became ó (this change is explicitly mentioned in LR:392 s.v. THÔN), ê became í and ô became ú. (Fauskanger, 'Old Sindarin - between Primitive Elvish and Grey-elven')

So, we have *híne (because the first e was not final) and the loss of the final vowel then results in hín. Likewise, had we started from *mene, we would have arrived at mín, and this is indeed what we would like to have. In order for this to work, we must assume that Tolkiens entry in the Etymologies refers to an archaic version of the pronouns and that the vowel is indeed long (this connects with the discussion above about the interpretation of the accents).

Can we find independent confirmation of such a scenario? In Quenya, possessives are exclusively attested as possessive endings (we have discussed those above), but there is tana (that) to the stem TA-. This is not a possessive but a demonstrative, but the important point here is that grammatically it behaves like an adjective, i.e. Q: cirya tana (that ship). Hence, the idea that the forms hono, hene and hana might be similar elaborations of the stem resulting in adjectivish forms appears all over sudden not so terribly exotic.

If so, we are able to derive the possessive forms corresponding to hon, hen, han: These would behave as hono -> *hóno -> *húno -> *hún (his), hene -> *héne -> *híne -> *hín (her) and hana -> *hána -> *hóna -> *hón (its).

3) The question of în

It remains to discuss the origin of the mysterious în. This form does not show any characteristic consonant (which is rather strange). Working backwards, we can construct an object like ?ene, but that has no counterpart in Quenya either, so we do not gain much.

Observing its use in actual text, we see that it is a reflexive pronoun, i.e. it refers back to the subject of the sentence: e aníra suilannad mhellyn în phain (he (Aragorn) wishes to give greeting to all his (Aragorn's) friends). Thus, one might speculate that this is a general reflexive form, which therefore does not need any characteristical consonant - it might always refer back to the subject.

Clarifying this speculation: 'I see my face' could then possibly be expressed *cenin níf în in Sindarin, the pronoun referring back to the subject 'I' and indicating that I see 'my own' face. But this is only a speculation based on the lack of a characteristical consonant - from this one form, no firm conclusions can be drawn.

Part IV - Cleaning up

The ideas presented in this last part of the analysis will differ from the previous part by being more speculative. Previously, is has been shown that combining rather well-known ideas, one can understand the structure of the Sindarin pronominal system. But this procedure left some gaps which cannot be filled by Quenya analogies or observation of the use of forms in the Sindarin texts. Thus, in order to fill these gaps, I will provide some educated guesses, tacidly assuming the reader is aware that I do not write these speculations with quite the same confidence as I have done above.

1) Plurals

The formation of 1st person plural is rather trivial: one observes a shift in the characteristic consonant from n (I) to m (we). The second person is rather tricky: Tolkien states that All these 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 (PM:42-43) This, however, was changed later: In WJ:364, Tolkien describes (for Quenya) a reduced pronominal affix -t (sg) and -l (pl), in striking contrast to the above statement. Dependent on what period of Tolkiens works one considers as most relevant, one may therefore identify:

a) c with 2nd person sg. and l with 2nd person pl.
b) c with 2nd person familiar and l with 2nd person courteous
c) c with 2nd person familiar sg., cc with 2nd person familiar pl., l with 2nd person courteous sg, ll with 2nd person courteous pl.
Obviously, this question cannot be settled uniquely.

It remains to discuss the plural formation of the 3rd person. We have already discussed that this is done in Quenya by a change of the characteristical consonant, but we have argued that there is no reason to assume that this has to be so in CE also. Indeed, in Sindarin we can deduce from Meril bess dîn (Rose, his wife) that t is used in singular and the etymologies tell that the forms derived from S- also denote singular. However, we are given the (Noldorin) plural forms huin, hîn, hein. We may tentively sindarinize those to get hyn, hîn, hain. The last form is indeed attested in the Moria Gate inscription Im Narvi hain echant (I Narvi made them), and from this we also learn that the listed plural forms should be associated with the object pronouns and that 3rd person plural in Sindarin is formed by i-affection.

In the nominative forms ho, he, ha, adding the plural marker would result in *hoi, *hei, *hai, which might become in Sindarin *hy, *hí, *hai. For the possessives, the order in which one carries out the vowel-shifts in our little scenario and the i-affection obviously matters - so let's cross our fingers and assume that i-affection only took place as the final vowel vanished, then we find *hún -> *huin, *hín -> *hín and *hón -> *hýn. But this is extremely shaky...

Similarly, we might deduce *te - > *ty , ten -> *tyn, tín -> *tín to cover the group in t.

2) mín and the question of lenition

In the 'Ae adar' (VT:44) we find the phrase sui mín i gohenam, evidently 'like we who forgive'. This mín is not a new form, but we would have expected it to be used as a possessive adjective, but here it is evidently seen in a different function. Note also that it is not lenited here, in spite of the fact that some have argued that sui, ending in a vowel, should cause lenition of the following word. We may speculate that the use of a possessive not associated with any noun would lead to the idea of 'ourselves' here, and this would open a second group of emphatic forms, i.e. *nín câr han (I do it myself, lit. 'myself does it' ).

Be this as it may, the fact that we see an unlenited form despite good reasons to lenit is certainly interesting. We have said before that the entry S- indicates that Tolkien imagined lenition even if a pronoun is used in nominative. But this use of the unlenited form mín might indicate a revision of that decision in the late stages, so it is possible that *so, *se*, *sa (he, she, it) is the correct version in later Sindarin.

3) A possible difference between *sa and *te

We know that the form den denotes 'it' in Sindarin, and we know further that dîn is translated as 'his', therefore the form is, unlike ho, he, ha a general expression for all genders. We have argued that it is not used to discriminate between singular and plural (as it was found in Quenya). So - what is the difference? There is not much to go, but in Quenya one knows the two demonstratives Q: sina (this), denoting things close to the speaker and Q: tana (that) for things away from the speaker. Obviously, these demonstratives are derived from the same stems S- and TA which are also underlying the personal pronouns, so there may be some justification to use their use as a guideline.

We can test this assumption on the corpus: Im Narvi hain echant refers to the doors directly in front of the writer, so one would expect the S- group being used here. Meril bess dîn refers to Sam's wife, who is by the time Aragorn writes (or makes someone write) the letter far away. Finally, caro den refers to a general 'it' which is not specifically mentioned, so there is some reason to assume that should not necessarily be close to the speaker, hence the choice of the TA group can be understood. This, however, may not be a very solid distinction.

Part V - Summary

Why should you believe all of this (well, at least part of it)? Primarily, because especially the first part is based on known things - no fancy grammar hypothesis is actually needed to derive the majority of the Sindarin pronominal verb endings and to get most of the pronouns - it comes almost for free by plain phonetical evolution. Where it can be tested, i.e. in the identification of ni with a non-emphatic pronoun and im its emphatic version, agreement with the given text is found, and in situations like the description of lammen clearly a description not based on the evolution history is bound to fail. All in all, most assumptions have to be made in the interpretation of individual forms, a task that is extremely difficult because Tolkien changed his intentions, and so there is no definite reconstruction. Most of the other assumptions (such as the lengthening of the vowel in the possessive) are based on things which are seen in the corpus, so although we lack an explanation, it is obvious that they're there. So - try to think over the whole scenario - it is capable of explaining quite a lot...

Thorsten Renk
thorsten@sindarin.de

Back to the index