S. verb inflections grammar.
In addition to tense, Sindarin verbs are also inflected with subject suffixes in the first and second person. The full set of suffixes (and their conceptual development) is discussed in the entry on subject suffixes. The treatment of 3rd person is somewhat unusual, however. The 3rd singular inflexion has no pronominal suffix, as in: agarfant beth “he spoke words” (PE17/126). Likewise the 3rd plural inflexion uses the plural suffix -r (PE17/132). However, this suffix is also used for subject/verb agreement in Sindarin. When the subject is plural, the verb must have the suffix -r:
This phrase is unglossed, and the exact meaning of the verb is unclear; Carl Hostetter tentatively suggested it may mean “remember”, though he also noted several other possibilities (VT50/15-17). Regardless of its meaning, it is clear this verb is in the plural to agree with its subject in Ellath. Elsewhere, though, the suffix -r appears without a separate plural subject:
Here there is no plural subject, and perhaps the meaning is more accurately “who they should call Fullwise”, with 3rd plural “they” used in an impersonal expression as it sometimes is in English. Thus in Sindarin, a bare verb (or verb with -r) can be used with or without a separate subject, and if there is no separate subject it represents the 3rd person:
In the 3rd person singular, the ancient final vowels in the present and past tense were lost for basic verbs (and in the past tense for derived verbs): câr “[he/she] does”, agor “[he/she] did”. But when an inflectional suffix is added to the tense, the ancient vowel was preserved, with various vocalic effects. This is most striking for the present tense of basic verbs, where the ancient vowel i is not only preserved but induces i-affection in inflected forms: câr “[he/she] does” vs. cerin “I do”, cerir “they do”.
For the future tense as well as the present tense of derived verbs (but not half-strong verbs), the ancient long ā is partially preserved in uninflected forms, surviving as short final -a: gala “[he/she] grows” or galatha “[he/she] will grow”. Most likely the long -ā first shortened as it usually did at the end of words, but was not lost because it was felt to be an intrinsic feature of the verb. In inflected forms, however, the long ā did not automatically shorten, and where preserved underwent its usual polysyllabic development: ā > ǭ > o. Thus in inflected forms the vowel a frequently changes to o: galon “I grow”, galathon “I will grow”.
For presents of derived-verb and futures in general, the change a → o did not always occur. In particular, it did not occur with suffixal -r, perhaps because this was added directly to the later verb stem and did not preserve its historical developments. In other words, perhaps -r was added only after the final ā shortened in the stem. In addition, a → o did not occur in those cases where the pronominal suffix was a cluster, such as galanc “we grow” using the 1962 1st person inclusive suffix -nc (PE17/132). Likewise with 1969 1st person inclusive suffix, we see athab “we won’t”, which is a strong indication that its ancient form was *athā-kwē. The long ā was shortened before these ancient clusters, and thus did not change ā > o.
The behavior seems to be somewhat different for half-strong verbs like [N.] nimmid(a)- “moisten” and [S.] covad(a)- “(make) meet”. Here the final a was lost in 3rd singular forms, as in nimmid “[he/she] moistens” and covad “[he/she] meets”. This is perhaps because this ancient -a was short rather than long: *ninkwită-, *komată-. If so, I suspect their inflected forms would have the vowel a in all cases, as in nimmidan “I moisten” and covadan “I meet”; I represent the stem of these half-strong verbs as nimmid(a)- and covad(a)- to indicate the joining vowel a. This is speculation, however, since we have no examples of inflected half-strong verbs.
Finally, there are some interesting open questions on the inflected forms of the past tense. In particular, the ancient past tense ended in long ē, but the joining vowel for the past tense of Sindarin basic verbs consistently appears with short e, as in ónen “I gave” (LotR/1061), eniðen “I intended” (PE22/165), agowen “I tasted” (PE22/152). This indicates that the final vowel in the past tense shortened before the pronominal suffix was agglutinated so that akjāwē-n(i) > akǭwĕn > agowen. However for derived verbs there are examples that hint the long ē may have developed into ī and possibly even causes i-affection, as in rithessin or (deleted) rithantin “I tried (intransitive vs. transitive)” (PE17/167). Such i-affection was a feature in the past tense systems for some of the earlier conceptual precursors to Sindarin (see below).
There were additional special developments for verbs whose uninflected past tenses ended in a nasal + a voiceless stop, such as for basic verbs ending in stops (annanc “[he/she] slayed” past tense of dag-) or half-strong and derived verbs (teithant “[he/she] drew” past tense of teitha-). In inflected forms these clusters would presumably undergo their medial developments to long nasals (or to -ng- in the case of -nc) so that their inflected forms would be like annangen “I slayed” or teithannen “I drew”. However, while such examples exist for Noldorin, there are no Sindarin examples of this and in fact there is one counterexample that indicates that the inflected pasts would preserve (or restore) the older spirantal pronunciation: echanthel “*you drew” (VT47/38).
The expected medial development of such a form would be: etkantē-l(e) > echanthĕl > echannel (vs. final developments -ant(e) > -anth > -ant). But the older spirantal pronunciation is still reflected in tengwar spelling (ld]3[lj where 3[ is a nasalized “th”). In notes from the late 1960s Tolkien said the spirantal pronunciation was sometimes restored from the spelling (VT42/27), and in the case of the past tense such a restoration/preservation would have been supported by the uninflected past echant. In such a system, inflected pasts would have nasal-spirant clusters like annanchen “I slayed”, teithanthen “I drew”. One advantage of such as system is that the 1st person past for derived verbs could be distinguished from the passive participle, which otherwise would be identical in form: teithannen “drawn”.
I personally find such a system quite plausible, but I don’t currently recommend it for the purposes of Neo-Sindarin writing for the following reasons: (1) It is based on a single example (echanthel) which conceivably represents an archaic form. (2) While there are no Sindarin examples, there are several Noldorin examples of inflected past forms showing long nasals medially. (3) So far as I know, no other Neo-Sindarin grammar uses this spirantal past tense formation. (4) The aforementioned passive participle formation (which is well attested in Sindarin) is also an argument against it, since the passive participle is probably the result of similar phonetic developments from the ancient past form + -inā: tektantinā > teχthanthen(a) > teithannen.
To summarize the above, using the verbs car- “do” and gala- “grow” as examples:
Conceptual Development: The variations in inflected vs. uninflected forms (and associated vocalic effects) dates back as far as the conjugations in the Early Noldorin Grammar of the 1920s (PE13/129). For example, in the present tense conjugation of the derived verb ᴱN. glathra- “polish” there is the a/o variation in inflected forms with single consonants vs. clusters, such as 1st and 2nd sg. glathron, glathrob vs. 1st and 2nd pl. glathranc, glathrast. Likewise basic verbs often preserved i which in turn could cause i-affection, such as 1st. and 2nd pl. medinc, medist for the verb ᴱN. mad- “eat”. The 3rd sg. forms were uninflected (glathra, mâd), and the 3rd pl. forms used -r.
Things were not entirely the same, however. In these 1920s conjugations, the 3rd pl. forms used -or for both derived and basic verbs, as in glathror, mador (as opposed to Sindarin-style glathrar, medir). The 1st pl. exclusive had the vowel u for both basic and derived verbs: glathrum, madum, no doubt due to vocalic assimilation to the labial nasal m. For reasons that are unclear, the 1st sg. basic verb inflection was madon. In the slightly later set of conjugations for basic verbs (PE13/131), the use of i seems to be limited to plural forms, as in neuter, masculine and feminine singular mâd, madeg, mades vs. neuter, masculine and feminine plural medir, medig, med(a)is.
In these 1920s inflections, past forms used i but without any i-affection, such as 1st and 2nd sg. pasts manthin, manthib. This is as opposed to 1920s aorist forms which used i with i-affection, such as 1st and 2nd sg. aorist mennin, mennib. In the slightly later Early Noldorin word lists, plural forms used simple i without any r, with or without i-intrusion, as in present 3rd sg. gonod vs. 3rd pl. genedi or genyd (PE13/145). Uninflected pasts had i-intrusion whereas inflected pasts had i and i-affection, as in 3rd. sg. neuter medaint vs. 3rd. pl. masculine medenniog.
In the (currently unpublished) Noldorin grammar of the 1930s (P17/44), uninflected past forms had no suffix and no i-affection, whereas inflected past forms had an i and varying degrees of i-affection, as in: 3rd. sg. past tagant vs. 1st. sg. tagennin or tegennin; 3rd. sg. past madant (†mant) vs. 1st. sg. medennin (†mennin). However, in The Etymologies of the 1930s, some inflected pasts used the vowel e and others the vowel i, as in: drammen, holen, sogennen (Ety/DARÁM; EtyAC/KHAL²; Ety/SUK) vs. hemmin, hennin (EtyAC/KHAM; Ety/KHAT).
As noted above, there are hints that this vacillation continued into the 1950s and 60s at least for the derived verb, as seen by r(a)itha- “try” (PE17/167) which had inflected past forms with both vowels: rithessin (intransitive), rithanen (transitive) and deleted form rithantem, rithantin. Thus it seems Tolkien wrestled with e vs. i for the inflected past forms of Sindarin and its conceptual precursors throughout his life.
Regarding the fate of final nasal clusters like -nt, this is another area where Tolkien vacillated. We see examples like manthin (-nth-) in Early Noldorin (P13/129), though in the case of Early Noldorin, the preservation of medial nth was in fact the normal medial development anyway. There are Noldorin examples like mennin (-nn-) (PE17/44) and one example echanthel (-nth-) in Sindarin (VT47/38). Thus, it is hard to tell what Tolkien ultimately decided.