ATHEMATIC I NFLECTION IN THE LATIN P RESENT S YSTEM by S TEPHEN L. TYNDALL (Under the direction of Jared Klein) ABSTRACT This thesis examines the six irregular Latin verbal paradigms in their present and future tenses, attempting to explain their irregularity. The forms are traced from their Proto-Indo-European roots through Proto-Italic and Old Latin into their classical forms, with considerable analysis based in both phonology and analogy provided. The copula, in particular, is examined in a number of Indo- European languages, and explanations for irregularity in analogy, haplology, and poetic elision are explored. I NDEX WORDS : Latin, Linguistics, Proto-Italic, Proto-Indo-European, Historical Linguistics
55
Embed
Athematic Inflection in the Latin Present System - Tyndall_stephen
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ATHEMATIC INFLECTION IN THE LATIN PRESENT SYSTEM
by
STEPHEN L. TYNDALL
(Under the direction of Jared Klein)
ABSTRACT
This thesis examines the six irregular Latin verbal paradigms in their present and future tenses,
attempting to explain their irregularity. The forms are traced from their Proto-Indo-European roots
through Proto-Italic and Old Latin into their classical forms, with considerable analysis based in
both phonology and analogy provided. The copula, in particular, is examined in a number of Indo-
European languages, and explanations for irregularity in analogy, haplology, and poetic elision are
explored.
INDEX WORDS: Latin, Linguistics, Proto-Italic, Proto-Indo-European, Historical Linguistics
ATHEMATIC INFLECTION IN THE LATIN PRESENT SYSTEM
by
STEPHEN L. TYNDALL
B.A., The University of Georgia, 2004
A Thesis Submitted to the Graduate Faculty
of The University of Georgia in Partial Fulfillment
The Proto-Indo-European athematic verbal category is almost entirely absent in the Italic verbal
system. Six (or nine, counting derived contractions) Italic verbs are generally thought to contain
remnants of Indo-European athematic inflectional morphology, but the structures of these verbs are
often unexpected, given the postulated Proto-Indo-European original forms.
This thesis attempts to provide an accurate diachronic analysis of these forms, in some instances
relying on prior research and analysis while presenting some new solutions to the problems these
unexpected forms offer.
The first part of this thesis provides a description of the present system of the Proto-Indo-
European verb in its various inflectional patterns, an understanding of which is critical for compre-
hension of the Proto-Italic and Latin morphological systems discussed in this work.
The second part of this thesis provides a description of the present system of the Latin verb,
including the division of the Proto-Indo-European types into the four Latin conjugations, as an
understanding of the regular Latin morphological system is crucial for understanding the structure
and history of the irregular forms.
The third part of this thesis describes the treatment of Proto-Indo-European athematic presents
that have been added to the regular Latin verbal system.
The fourth and largest section of this thesis is broken into synchronic and diachronic descrip-
tions of the five irregular Latin verbs.
The thesis concludes with a discussion of the similarities and differences between the irregular
paradigms of Latin, with tentative explanations of the reasons for retention of and similarity in
their irregularities.
1
2
Table 1.1: A List of Possible Athematic Latin Paradigms - Present Tense
Person and Number esse velle ferre ıre edere dareFirst Singular sum volo fero eo edo doSecond Singular es vıs fers ıs es dasThird Singular est vult fert it est datFirst Plural sumus volumus ferimus ımus edimus damusSecond Plural estis vultis fertis ıtis estis datisThird Plural sunt volunt ferunt eunt edunt dant
Table 1.2: A List of Possible Athematic Latin Paradigms - Future Tense
Person and Number esse velle ferre ıre edere dareFirst Singular ero volam feram ıbo edam daboSecond Singular eris voles feres ıbis edes dabisThird Singular erit volet feret ıbit edet dabitFirst Plural erimus volemus feremus ıbimus edemus dabimusSecond Plural eritis voletis feretis ıbitis edetis dabitisThird Plural erunt volent ferent ıbunt edent dabunt
1.1 PROBLEMATIC PARADIGMS
1.1.1 EO ‘I GO’
The paradigm of eo ‘I go’ has four athematic forms, ıs, it, ımus, and ıtis, and two thematic forms,
eo and eunt, and is the least problematic of the Latin irregular verbs, probably because of the lack
of phonotactic problems engendered by the shape of the root.
3
1.1.2 FERO
The paradigm of fero ‘I bear’ has three seemingly-athematic forms fers, fert, and fertis, though
almost all comparative evidence suggests a simple thematic etymon for the paradigm. The irregular
forms are analyzed as reduced, either via a regular phonetic process or via phonetic reduction based
on frequency of use.
1.1.3 edo ‘I EAT’
The paradigm of edo ‘I eat’ contains three athematic forms es, est, and estis, and three thematic
forms edo, edimus, and edunt. A curious alternation in root vowel quantity between e- in the first
person singular and first and third persons plural and e in the second and third persons singular and
second person plural is the focus of this section.
1.1.4 volo ‘I WISH’
The paradigm of volo ‘I wish,’ is irregular in three indicative forms vıs, vult, and vultis, and appears
to have a historically athematic Latin subjunctive (i.e. optative in Indo-European terms) paradigm.
The second person singular vıs is the most anomalous form of the paradigm and is discussed at
length.
1.1.5 sum ‘I AM’
The paradigm of the copula is the least understood and most contentious of the irregular Latin
verbs. This thesis contains a large literature review, and an original, strong suggestion for the
history of this paradigm.
1.1.6 do ‘I GIVE’
While only modestly irregular by synchronic Latin standards, the paradigm of do ‘I give’ shows
remnants of athematic inflection.
CHAPTER 2
THE PROTO-INDO-EUROPEAN PRESENT INDICATIVE
The Present System of Indo-European is composed of the present indicative and imperative, the
imperfect, the active and medio-passive voices, and the subjunctive and optative moods. The
present and imperfect active indicative and the active subjunctive will be treated in this section.
The Indo-European verb is composed of three parts, a root, zero or more suffixes, and a personal
ending.
2.1 THE PERSONAL ENDINGS OF THE INDO-EUROPEAN PRESENT ACTIVE
Proto-Indo-European present active verbs had two sets of personal endings, the primary and sec-
ondary endings. Primary endings were used in the present, while secondary endings appeared in
the imperfect tense.
2.1.1 PRIMARY ENDINGS OF ACTIVE VERBS
Table 2.1 presents the primary verbal endings of the active voice in their Proto-Indo-European
reconstructed forms1 and with representative forms from the copula of Sanskrit and Greek. Only
those forms relevant to the Italic system appear.
2.1.2 SECONDARY ENDINGS OF ACTIVE VERBS
Table 2.2 presents the secondary endings of the active voice in their Proto-Indo-European recon-
structed forms and with representative forms from the imperfect of the root *bher ’carry’ in San-
skrit and Greek. Again, only those forms relevant to the Italic verbal system appear.1I’ve chosen to use those reconstructions mostly closely underlying the Italic system unless otherwise
noted.
4
5
Table 2.1: The Primary Endings of the Present Active
Person and Number Ending PIE Sanskrit GreekFirst Singular *-mi or *-H2 *H1esmi asmi eimıSecond Singular *-si *H1esi asi eıThird Singular *-ti *H1esti asti estıFirst Plural *-me/o(s) *H. 1sme/os smas esmenSecond Plural *-te *H. 1ste stha esteThird Plural *-enti *H. 1senti santi Doric entı
Table 2.2: The Secondary Endings of the Present Active
Person and Number Ending PIE Sanskrit GreekFirst Singular *-m *ebherom abharam epheronSecond Singular *-s *ebheres abharas epheresThird Singular *-t *ebheret abharat ephereFirst Plural *-me *ebherome abharama epheromenSecond Plural *-te *ebherete abharata ephereteThird Plural *-(e)nt *ebheront abharan epheron
2.2 THE PRESENT TENSE
The present tense in Indo-European was built with the verbal root, zero or more suffixes, and the
primary endings detailed in Table 2.1.
2.2.1 THE ATHEMATIC PRESENT
The Proto-Indo-European athematic present consists of either an ablauting root or a zero grade
root with an ablauting infix, followed by one of the primary endings in Table 2.1.
6
Table 2.3: An Example of an Athematic Narten Present
Person and Number Proto-Indo-IranianFirst Singular *st ´aumiSecond Singular *st ´ausiThird Singular *st ´autiFirst Plural *staumas(i)Second Plural *stauthaThird Plural *stauati
THE SIMPLE ATHEMATIC PRESENT
Simple athematic presents are composed of an ablauting root followed by primary endings. The
root takes full grade in the singular and zero-grade in the plural. See the PIE column in Table 2.1
for an example of this type of verb in the singular and plural.
THE NARTEN PRESENT
Narten presents are similar to simple athematic presents save that they take e-grade in the sin-
gular and full grade in the plural [Narten, 1968, 18]. Table 2.3 presents the forms Narten suggests
underlie this type of present, demonstrated with root√
stu ‘praise.’
THE NASAL INFIX PRESENT
Nasal infix presents consist of a root in zero-grade with an ablauting infix of the form
*-ne-/-n- before the final consonant and primary endings. As is typical with athematic formations,
the ablauting infix takes full grade in the singular and zero grade in the plural. Table 2.4 presents
the paradigm of a nasal infix present built to the root *yug ‘yoke.’
7
Table 2.4: The Athematic Nasal Infix Present
Person and Number PIE SanskritFirst Singular *yu-ne-g-mi yunajmiSecond Singular *yu-ne-g-si yunaks. iThird Singular *yu-ne-g-ti yunaktiFirst Plural *yu-n-g-me/os yunjmasSecond Plural *yu-n-g-te yunkthaThird Plural *yu-n-g-enti yunjanti
Table 2.5: The Athematic Reduplicated Present
Person and Number PIE SanskritFirst Singular *bhı-bher-mi bıbharmiSecond Singular *bhı-bher-si bıbhars. iThird Singular *bhı-bher-ti bıbhartiFirst Plural *bhi-bhr. -me/os bibhr. masSecond Plural *bhi-bhr. -te bibhr. thaThird Plural *bhı-bhr-n. ti bıbhrati
THE REDUPLICATED ATHEMATIC
Reduplicated athematic formations consist of the onset of the root syllable followed by either
*-e- or *-i-, the root, and primary endings [Fortson, 2004, 88]. The root has the same ablaut pattern
as the simple athematic formation. Table 2.5 illustrates this type with the root *bher- ‘carry.’
2.2.2 THE THEMATIC PRESENT
The thematic presents of Indo-European are marked by an ablauting *-e-/-o- morpheme that
appears before the personal ending. In contrast to the alternation between full grade and zero grade
in the singular and plural of the athematic formations, the morpheme alternates between *-o- in the
8
Table 2.6: The Simple Thematic Present
Person and Number PIE SanskritFirst Singular *bher-o-H2 bharamiSecond Singular *bher-e-si bharasiThird Singular *bher-e-ti bharatiFirst Plural *bher-o-me/os bharamasSecond Plural *bher-e-te bharathaThird Plural *bher-o-nti bharanti
first person singular and the first and third persons plural and *-e- in the second and third persons
singular and the second person plural.
THE SIMPLE THEMATIC PRESENT
Simple thematic presents consist of a root in some fixed ablaut grade, the ablauting thematic vowel,
and a personal ending. Two types exist, one with the accent fixed on the root, the other with the
accent on the thematic vowel [Fortson, 2004, 89]. Table 2.6 presents an example of the simple
thematic inflection of the root-accented *bher- ‘carry.’ The difference in accent between the two
types just noted can be seen by comparing Sanskrit bharati illustrated above with the third person
singular form tudati ‘thrusts.’
THE *-ye-/-yo- PRESENT
The *-ye-/-yo- present was built with either full or zero grade of the root with the suffix *-yo- in
the first person singular and first and third persons plural and *-ye- in the second and third persons
singular and second person plural [Fortson, 2004, 89]. The suffix was used as the primary verbal
formation for some roots and as a denominative suffix in other cases [Fortson, 2004, 89-90]. See
the second column of Table 2.7 for an example of the inflection of a *-ye-/-yo- present.
9
THE CAUSATIVE-ITERATIVE
The Proto-Indo-European causative-iterative was composed of a root in o-grade, an ablauting suffix
*-eyo-/-eye-, and primary endings. The causative-iterative has two possible meanings. First, it can
mean that the subject of the verb causes something to undergo the action indicated by the root,
e.g. Latin moneo ‘I warn/cause to think’ < *mon-eyo-H2 from the root *men ‘think.’ Second,
the causative-iterative can mean that the subject performs the action indicated by the verbal root
repeatedly, e.g. Greek potH1eomai < *potH1-eyo-m-H2oi2 from the root *petH1 ‘fly.’ Table 2.7
has an example of causative-iterative inflection in its third column.
THE *-ske-/-sko- PRESENT
The Proto-Indo-European *-ske-/-sko- present was built with a root in zero grade, the ablauting
*-ske-/-sko- morpheme, and primary endings [Fortson, 2004, 90]. The Indo-European daughter
languages differ as to the meaning of this suffix. Some indicate habitual action, while Latin, of
concern here, uses the *-ske-/-sko- formation for inchoatives like rubescit ‘becomes red’ < *rudh-
eH1-ske-ti, from the root *rudh ‘red’ [Fortson, 2004, 90]. This type is illustrated in the fourth
column of Table 2.7.
THE REDUPLICATED THEMATIC PRESENT
The Proto-Indo-European Reduplicated Thematic Present was built with some portion of the
root onset plus *-i-, the root in zero grade, the ablauting thematic vowel, and primary endings
[Fortson, 2004, 90]. Table 2.7 provides an example of this kind of inflection in its fifth column.
OTHER TYPES
There exists a number of other types of present formations, but these fall outside the purview of
this work.2This reconstruction is after Sihler, see [Sihler, 1995, 471]
in -s-, and denominatives from u-stem nouns [DeWandel, 1982, 50-54].
THE NASAL INFIX PRESENT
Latin treats nasal infix presents in several different ways. First, the nasal infix can be fixed in the
zero grade, the present root morpheme re-analyzed as containing the infix, and then the verb can be
inflected like a normal thematic third conjugation verb, cf. tango ‘I touch,’ with the perfect tense
tetigı proving the presence of the infix [DeWandel, 1982, 51]. A different treatment is that of tollo
‘I raise/destroy.’ The form comes from *tolno, itself from *tl.-n-H2-o-H2, rebuilt from athematic
*tl.-ne-H2-mi, and has been merged into the Latin third conjugation.
THE ATHEMATIC REDUPLICATED PRESENT
No athematic reduplicated presents persist into Latin. A few thematic reduplicated presents exist
as members of the third conjugation. Latin sisto ‘I set up’ and bibo ‘I drink’ are of this type, as
is reddo ‘I give back,’ where the geminated d reflects an original *di-dH3-, as in Sanskrit dadati
‘gives,’ (which actually shows a full-grade root).
3.3.4 THE LATIN FOURTH CONJUGATION
The Latin fourth conjugation verb is marked by a morpheme ı before the personal ending. An
example of the inflection of the fourth conjugation verb venıre ‘to come’ is given in Table 3.4. The
fourth conjugation is composed of *-ye-/-yo- presents, a few root athematics, denominatives from
16
Table 3.3: The Latin Third Conjugation
Person and Number Present FutureFirst Singular dıco dıcamSecond Singular dıcis dıcesThird Singular dıcit dıcetFirst Plural dıcimus dıcemusSecond Plural dıcitis dıcetisThird Plural dıcunt dıcent
Table 3.4: The Latin Fourth Conjugation
Person and Number Present FutureFirst Singular venio veniamSecond Singular venıs veniesThird Singular venit venietFirst Plural venımus veniemusSecond Plural venıtis venietisThird Plural veniunt venient
i-stem nouns, desideratives, and a single causative [DeWandel, 1982, 55-57]. These formations are
thematic, built with the suffix *-ye/o- [Sihler, 1995, 536].
CHAPTER 4
LATIN dare ‘TO GIVE’
The Latin present paradigm of dare ‘to give,’ shown in Table 4.1, shows athematic inflection from a
Proto-Indo-European root *deH3 ‘give.’ The verb differs from typical first conjugation verbs in that
the first conjugation suffixes in -a are understood to come from PIE *-eH2-y(e/o)-, an underlyingly
thematic formation. The root shows reduplicated athematic presents in Greek dıdomi ‘I give’ and
Sanskrit dadami ‘id..’ Given an etymon identical to that of Greek and Sanskrit, we might expect
a Latin paradigm *dido, *didos, *didot, *didamus, *didatis, *dident < *didH3n. ti or *didont <
*didH3enti. Latin, however, appears to have built a simple athematic present to the zero grade of
the root, *dH. 3- > da-.
In the first and second persons plural, the forms are precisely what we’d expect from a simple
athematic paradigm, PIE *dH. 3-mos > damus, and PIE *dH. 3-te → datis. The third person plural
would likely have syllabified as PIE *dH3n. ti (cf. Sanskrit third plural dadati < PIE *de-dH3n. ti),
which should have resulted in Latin *dent (cf. Latin tentus ‘stretched’ from *tn. -tos [Sihler, 1995,
Table 4.1: The Present Paradigm of dare ‘to give’
Person and Number dareFirst Singular doSecond Singular dasThird Singular datFirst Plural damusSecond Plural datisThird Plural dant
17
18
97]). Latin *dent was leveled to dant for paradigmatic unity. The future and imperfect forms like
dabit ‘will give’ and dabat ‘was giving’ likewise reflect the zero grade form *dH. 3.
The morpheme da- also becomes the base for the singular forms of the Latin present. In the
first person singular, *da-o contracts to do, and in the third person singular, the personal ending -t
is added to the present stem morpheme to yield dat. The second person singular, das could have
gotten its long vowel in a number of ways. Perhaps it was a remodeling of a more original
*(di)-dos → das [Sihler, 1995, 544]. Otherwise, *das could have been remade as das after
homophony with the the singular pattern of the first conjugation. The first conjugation pattern seen
in laudo, laudas, laudat, ‘I praise,’ ‘you praise,’ and ‘praises,’ respectively, becomes the pattern
for analogical lengthening of the second person singular, leaving a matching singular present
paradigm do, das, dat. The match with the first conjugation is incomplete, however, since the stem
of the first and second persons plural is da- rather than *da-.
The future and imperfect tenses to this root also appear to be built to dH. 3- > da-, as in dabit
‘will give’ and dabat ‘was giving.’ Further, the second element of the form cedo ‘give here!’ may
represent a full-grade, endingless imperative *-deH3.
The paradigm of dare ‘to give’ shows a sort of dereduplicated athematic present, and certain
compounds, such as reddo ‘I give back’ show evidence of the reduplication, and the form reddit
‘gives back’ could come from *re-di-dH. 3ti > *redidat, then, with syncope and vowel weakening,
reddit. These compounds have been merged into the third conjugation and appear regular with
respect to synchronic Latin grammar.
CHAPTER 5
LATIN ıre ‘TO GO’
The Latin paradigm of ıre ‘to go’ is the most regular of the Latin athematic verbs, probably because
the root suffers no phonotactic difficulties when coming into contact with personal endings. The
first person plural form of this root is the only first plural of the Latin irregular verbs without
an inserted thematic vowel. Table 5 presents the inflection of ıre in the present. Four forms of
the paradigm remain athematic, though the vocalism of the root has been analogically leveled
throughout.
5.1 THE SINGULAR
The second and third person singular forms of the paradigm of ıre are nearly identical to their
Proto-Indo-European proto-forms, while the first person singular has been thematized.
Table 5.1: The Present of ıre
Person and Number Latin Sanskrit Greek PIEFirst Singular eo emi eımi *H1ey-miSecond Singular ıs es. i ei *H1ey-siThird Singular it eti eısi *H1ey-tiFirst Plural ımus imas ımen *H1i-mesSecond Plural ıtis itha ıte *H1i-teThird Plural eunt (i)yanti ıasi *H1(i)yenti
19
20
5.1.1 eo ‘I GO’
The Latin form eo comes straightforwardly from *H1ey-o-H2. Intervocalic *y was lost everywhere
in Latin, as in *treyes > Latin tres ‘three’ [Sihler, 1995, 188]. The two vowels remain in hiatus
with no further modification. Note that this form is thematic in structure, in contrast to the proto-
form *H1ey-mi that underlies the Sanskrit and Greek forms of the first person singular. The Latin
present form is here identical to the Proto-Indo-European subjunctive form *H1ey-o-H2, cf. San-
skrit aya(ni).
5.1.2 ıs ‘YOU GO’
The Latin form ıs is one of the clearly athematic forms within this paradigm. The proto-form here is
*H1ey-si. Latin first loses the hic-et-nunc particle, and then the diphthong *ey is monophthongized
to -ı, cf. *bheydh-oH2 ‘I am persuaded’ > Latin fıdo ‘I trust’ [Sihler, 1995, 52].
5.1.3 it ‘GOES’
The Classical Latin form it is descended from Old Latin ıt, which comes from *H1ei-ti by the
process detailed in 5.1.2. The reduction of -ı- to -i- is a regular occurrence before every final
consonant except s [Baldi, 1999b, 266].
5.2 THE PLURAL
All plural forms of the paradigm of ıre show a full-grade root that has supplanted the zero-grade
Proto-Indo-European form reflected by the Greek and Sanskrit plural forms in Table 5.
5.2.1 ımus ‘WE GO’
In this form, the full grade form of the root has been used instead of the more original zero grade
form. Thus, PIE *H1i-mos→ Pre-Latin *ey-mos > Latin ımus.
21
Table 5.2: The Future and Imperfect of ıre
Person and Number Future ImperfectFirst Singular ıbo ıbamSecond Singular ıbis ıbasThird Singular ıbit ıbatFirst Plural ıbimus ıbamusSecond Plural ıbitis ıbatisThird Plural ıbunt ıbant
5.2.2 ıtis ‘YOU (PL.) GO’
As in 5.2.1, ıtis has a full-grade root instead of the expected zero grade. Otherwise, this form’s
history is mutatis mutandis like that of ımus.
5.2.3 eunt ‘THEY GO’
The Latin form eunt differs from the Proto-Indo-European reconstruction in two important aspects:
the full-grade root and the replacement of the athematic ending *-enti with the thematic *-onti.
Thus, PIE *H1(i)y-enti→ Pre-Latin *ey-onti > Latin eunt by loss of the hic-et-nunc particle and
intervocalic *-y-, as in 5.1.1 above. Note here, as in the first person, that *ey-ont(i) is identical to
the Proto-Indo-European subjunctive form of this word, as shown by Sanskrit ayan ‘they will go’
< *H1ey-ont.
5.3 THE FUTURE AND IMPERFECT OF ıre
Both the future and imperfect paradigms of ıre are built with a morpheme -b-, an Italic innovation
likely created from the Proto-Indo-European root *bhu ‘be’ [Palmer, 1961, 271]. Table 5.2 presents
the paradigms of the future and imperfect of ıre.
CHAPTER 6
LATIN ferre ‘TO BEAR’
The present indicative paradigm of ferre ‘to bear’ seems to contain three athematic forms, the
second and third persons singular and second person plural. The paradigm of ferre from Proto-
Indo-European *bher ‘carry, bear’, presented in Table 6.1, is anomalous, in that comparative evi-
dence for an athematic present is scanty. The present indicative paradigms for this root in Greek
and Sanskrit, given in Table 6.2, point to a simple thematic paradigm.
Some isolated athematic forms do exist, however. Sihler discusses an occasionally-occurring
Sanskrit bharti [Sihler, 1995, 541], and Lindeman presents Homeric pherte, a plural imperative
attested once, in book nine of the Iliad [Lindeman, 1976, 117].
We would expect a Latin future *ferit if fert were a real athematic present, just as erit ‘it will be’
< *H1es-et is the future to athematic est ‘is.’ The actual future, as seen in Table 6.1, is feret, from
the PIE thematic subjunctive *bher-e-e-t(i), suggesting a present more in line with the paradigms
in Sanskrit and Greek.
Table 6.1: The Latin Paradigm of ferre
Person and Number Present FutureFirst Singular fero feramSecond Singular fers feresThird Singular fert feretFirst Plural ferimus feremusSecond Plural fertis feretisThird Plural ferunt ferent
22
23
Table 6.2: The Present of *bher in Sanskrit and Greek
Person and Number Sanskrit GreekFirst Singular bharami pheroSecond Singular bharasi phereisThird Singular bharati phereiFirst Plural bharamas pheromenSecond Plural bharatha phereteThird Plural bharanti pherousi
Further, syncope of both vowels and consonants after r is frequent in Latin. The *-ro- and
*-ero- stem nouns attest to this tendency, as do a few words in final *-ris. A Proto-Indo-European
form *agros (cf. Greek agros ‘field’) becomes Latin ager ‘field,’ inscriptional Old Latin SAKROS
‘priest’ becomes classical Latin sacer, and *akris ‘sharp’ becomes Latin acer [Sihler, 1995, 315].
The case of *pueros > *puers (> *puerr) > puer ‘boy’ provides the most direct parallel for
the development of fers, since, by a parallel process, we might expect *feris to become *fer, a
form lacking the -s that is everywhere the marker of the second person singular in Latin. Cowgill
suggests a *ferr which speakers then remake into fers, requiring a second person singular form in
-s rather than -r [Cowgill, 1978, 26].
A few verbs make an explanation of regular syncope in fers, fert, and fertis problematic. As
Sihler claims, gero ‘I wage’ comes from a pre-rhotacism form *geso, so our syncope needs to have
occurred before rhotacism took intervocalic *-s- to -r-, i.e. before about 500 BCE [Sihler, 1995,
541]. Another form, ferit ‘strikes,’ from ferio ‘I strike’, was, in Old Latin ferıt, with a long ı- that
makes this form irrelevant to the discussion of the reduction of *ferit to fert ‘bears.’ A third verb,
tero ‘I rub’ disproves Sihler’s claim that fero is the only Latin simple thematic verb ending in
single -r- [Sihler, 1995, 541] and proves that the syncope suggested by Sihler cannot be regular,
since tero carries normal third conjugation inflection.
24
Since regular syncope cannot explain the retention of terit ‘rubs’ and the reduction of *ferit
‘carries,’ we must turn to another possible explanation. The fact that an irregular (by the standards
of the Latin conjugation system) form survived suggests that this form was more frequently used
and more salient than those forms that were part of Latin’s productive morphological system. As
Bybee claims:
High frequency encourages phonetic change, but it renders items more conservative
in the face of grammatical change or analogical change based on the analysis of
other forms (Phillips 2001)1. For example, high-frequency forms with alternations
resist analogical leveling: while English weep / wept, creep / crept, and leap / leapt
have a tendency to regularize to weeped, creeped, and leaped, respectively, the high-
frequency verbs with the same pattern, keep / kept, sleep / slept show no such tendency
(Bybee 19852, Hooper 1976b3). As a result, morphological irregularity is always cen-
tered on the high-frequency items of a language [Bybee, 2001, 12].
We can surmise, then, that the irregular ferre was more frequent than the regular terere. Fur-
ther research on phonological effects of word frequency suggests that the more frequent a word,
the more phonetic (and therefore phonological) reduction it undergoes. Research by Pluymaekers,
et al., has shown that high frequency Dutch words have phonetically shorter affixes than low fre-
quency words [Pluymaekers et al., 2005, 43-46]. This, taken together with Bybee’s more general,
cross-linguistic claim that frequency and phonological reduction are directly related [Bybee, 2001,
60-62], supports the irregular syncopation of *feris, *ferit, and *feritis to fers, fert, and fertis,
respectively, while allowing the retention of regular thematic inflection in the present of terere.
1See [Phillips, 2001] in bibliography2See [Bybee, 1985] in bibliography3See [Hooper, 1976] in bibliography.
CHAPTER 7
LATIN esse ‘TO EAT’
The paradigm of Latin esse ‘to eat’ shows the typical Latin pattern of simple thematic forms in the
first person singular and first and third persons plural and athematic forms in the second and third
persons singular and second person plural. Table 7 presents the full paradigm of esse in the present
tense.
7.1 THE PROBLEM OF VOWEL LENGTH
This Latin paradigm shows two distinct present stems, ed- in the first person singular and first and
third persons plural, and es- in the second and third persons singular and second person plural.
The paradigm of the root *ed ‘eat’ in Slavic, presented in Table 7.3, and perhaps the pres-
ence of full-grade forms like adanti ‘they eat’ in Vedic, suggest that at least some Indo-European
dialects inherited this root with Narten ablaut pattern. The Proto-Indo-European paradigm of the
Table 7.1: The Present of esse ‘to eat’
Person and Number Latin FormFirst Singular edoSecond Singular esThird Singular estFirst Plural edimusSecond Plural estisThird Plural edunt
25
26
Table 7.2: The Root *ed ‘eat’ in PIE and Pre-Latin
Person and Number PIE Pre-LatinFirst Singular *H1 ´ed-mi *edoSecond Singular *H1 ´ed-si *ess > esThird Singular *H1 ´eds-ti *ess→ estFirst Plural *H1ed-mos edimusSecond Plural *H1eds-te *esstis >→ *estisThird Plural *H1ed-n. ti edunt
root, given Narten ablaut, is presented in Table 7.1 with its probable initial Latin outcomes after
the replacement of athematic endings with thematic endings in the first person singular and first
and third persons plural, and the analogical restitution of of -t in the third person singular and the
second person plural after the double dental law took the sequence *-dst- to ss, which becomes s
after a long vowel almost immediately, in Latin.
Note that in the Pre-Latin paradigm, speakers are forced to deal with four distinct root allo-
morphs: ed- in the first person singular, es- in the second and third persons singular, ed- in the first
and third persons plural, and *es- in the second person plural. These roots have alternations both in
vowel length, between the singular and the plural, and in final consonant, wherein the first person
singular and the first and third persons plural share a -d, while second and third persons singular
and the second person plural share a final -s.
Speakers then collapse a system of four root allomorphs into a system of two allomorphs. Thus
*estis→ estis to match the root morpheme of est and es, or perhaps the long vowel results from
lengthening after consonant cluster reduction, and *edo is remade as edo after edimus and edunt.
The preservation of the Narten pattern with modification in the first person singular and second
person plural in this verb mirrors the presence of the typical athematic ablaut pattern in the Latin
copula, with full grade forms in the second and third persons singular and second person plural
and zero-grade forms in the first person singular and first and third persons plural.
27
7.2 THE VERBAL ENDINGS
The thematic forms, edo, edimus, edunt, and edere, the analogically-produced infinitive, are easily
explained by analogy to third conjugation simple thematic forms like ago, agimus, agunt, and
agere respectively. The vowel quantity in the first person singular edo must be analogical to that of
edimus and edunt, since the expected Narten form would be *ed-mi.
7.3 es ‘YOU EAT’
The underlying form of es is likely *H1 ´ed-si > Proto-Italic et-s with the loss of the hic-et-nunc
particle. The s would devoice the preceding d, leaving a form ets which would, by regular devel-
opment, become *ess, which contracts to es after a long vowel, cf. Plautine mıless ‘soldier’ from
*mılet-s [Sihler, 1995, 205]. Plautine mıless appears in Classical Latin as mıles, with a short e, so,
absent an original lengthened grade from Narten inflection, we would expect *es to be the Classical
Latin form of the second person singular of this verb.
7.4 est ‘EATS’
The underlying form of est should be *eds-ti > *ets-ti > *essi > *ess > es by the Latin outcome
of the double dental law of Indo-European, cf. the *-ti-on- abstract noun sessio ‘a sitting’ to root
sed ‘sit.’ This expected form is identical to the second person singular form above and lacks the
-t characteristic of the third person singular, so speakers were unlikely to have tolerated it. Ana-
logical replacement of the third person singular marker -t yielded est. Without recourse to Narten
inflection, the vowel length is without parallel or good explanation1.
1There exists a remote possibility that the length of the vowels in est and estis is due to Lachmann’sLaw, a contentious rule involving the lengthening of vowels before a sequence of PIE plain voiced stop andvoiceless stop. Consult [Jasanoff, 2004] for the controversy surrounding this rule.
28
Table 7.3: The Old Church Slavic Present Paradigm of ‘to eat’
Person and Number OCS FormFirst Singular jamıSecond Singular jasiThird Singular jastuFirst Plural jamuSecond Plural jasteThird Plural jadetu
7.5 estis ‘YOU(PL.) EAT’
Given the proto-form *eds-tes, the development of estis would parallel that of est in section 7.4
above with respect to the -st.
7.6 PARALLELS TO SLAVIC
The Old Church Slavic present paradigm of ‘to eat,’ given in Table 7.3, parallels the Latin paradigm
in a few important ways. First, the paradigm is clearly athematic. Second, the ja in the root appears
to come from *je, with joticization of an initial vowel and the regular change of *je to ja, cf. OCS
stojati ‘to stand’ from *sto-je-tei [Schmalstieg, 1983, 51]. This, together with the Slavic outcome
of double dental sequences, *-st- [Schmalstieg, 1983, 30], suggests a paradigm that has a root
*ed-, probably from an original Narten present with the lengthened grade analogically extended
to the plural. The presence of extended Narten inflection here supports the supposition of Narten
proto-forms underlying the present paradigm of the Latin forms.
7.7 CONCLUSIONS
The paradigm of edo shows the single instance of the Narten ablaut pattern remaining in Latin.
Indeed, only this verb and the copula show any intra-paradigmatic root ablaut in the present
29
active indicative, and they display the same pattern, with the ablaut grade found in the Proto-
Indo-European singular appearing in the Latin second and third person singular and the second
person plural, and the ablaut grade found in the Proto-Indo-European plural found in the Latin first
person singular and the first and third persons plural.
CHAPTER 8
LATIN velle ‘TO WISH/WANT’
The paradigm of Latin volo has a number of interesting features and irregularities. Two sub-
paradigms, that of malo ‘I prefer’ and nolo ‘I don’t want’, contractions built during the early
history of the Latin language, exist alongside the main paradigm. The three paradigms are pre-
sented together in Table 8.1.
8.1 THE ROOT *welH1-
The root of Latin velle, Proto-Indo-European *welH1-, has a number of interesting characteristics.
The root appears to carry athematic inflection across Indo-European dialects, with Old Lithuanian
first person singular pa-velmi and Sanskrit nasal-infix present vr. n. ıte, which points to a root *welH1
instead of *wel [Sihler, 1995, 539]. The presence of Germanic subjunctive, i.e. Indo-European
optative, forms with full grade, like Gothic wiljau ‘I wish’ leads some to suggest that this root had
Table 8.1: The Paradigms of velle, malle, and nolle
Person and Number velle malle nolleFirst Singular volo malo noloSecond Singular vıs mavıs non vısThird Singular vult mavult non vultFirst Plural volumus malumus nolumusSecond Plural vultis mavultis non vultisThird Plural volunt malunt nolunt
30
31
Narten inflection rather than the simple athematic pattern [Sihler, 1995, 539]. Since Latin almost
always levels the ablaut grade of the root between the singular and the plural (cf. eo ‘I go’ < *ey-
o-H2 and ımus ‘we go’ < *ey-mos rather than *i-mos), the Latin paradigm can tell us little about
the ablaut grade of the Proto-Indo-European etymon.
With regard to the final laryngeal of the reconstructed root, Latin presents a problem. The
root *welH1, should have yielded thematic-looking forms like second person singular *volis from
*welH. 1-si via an intermediate *volas. The root, in order to yield the forms attested in Latin, must
have entered the language as *wel- rather than *welH1, and the set. root structure is not completely
certain, as Swiggers and Seldeslachts discuss [Seldeslachts and Swiggers, 2002, 324].
8.2 THE THEMATIC FORMS OF velle
The thematic forms in the paradigm of velle are, as is normal, the first person singular and the
first and third persons plural. There are a few variations in need of explanation. First, the infinitive
velle from Proto-Indo-European *wel(H1)-s-i, the locative to an *s-stem noun to the verbal root
[Sihler, 1995, 610], reveals an original e-grade in the paradigm. The original paradigm for these
forms, *wel(H1)-mi, *wel(H1)-mos, and *wel(H1)-ent were replaced by *wel-o-H2, *wel-o-mos,
and *wel-onti, respectively. The -e- in the root is rounded to -o- by either a rule of *we- >
*wo-, as in Latin vomo ‘I vomit’ from Proto-Indo-European *wemH1-, cf. Sanskrit athematic vamiti
[Whitney, 1885, 154] and Greek emeo [Sihler, 1995, 41], or a rule of *e > *o before a pinguis l,
as in *kwel-o-H2 ‘I go round’ > Latin colo ‘I till’ [Sihler, 1995, 41]. In either case, the result is a
synchronic Latin allomorph vol- in the present indicative.
The first person plural form volumus shows a -u- between the root and the personal ending,
giving -umus rather than the typical thematic -imus seen in the Latin third conjugation. The vowel
could have been conditioned by the strongly rounded environment, but forms like vomimus ‘we
vomit’ preclude the presence of a regular *i > u rule in labial surroundings. The velar l may have
added a tendency toward rounding. Another uncertain possibility is analogy with sumus ‘we are’
which is itself probably an analogical creation.
32
Table 8.2: The Present Paradigm of vı
Person and Number vıFirst Singular vemiSecond Singular ves. iThird Singular vetiFirst Plural vımasSecond Plural vıthaThird Plural viyanti
8.3 vult ‘WISHES’
Latin vult (Old Latin volt) ‘wishes’ is the regular outcome of Proto-Italic *wel-ti. The rounding of
the *-e- in the root is detailed in section 8.2 above, and the the change of o to u before a pinguis l
is a regular development between Old and Classical Latin [Sihler, 1995, 62].
8.4 vıs ‘YOU WISH’
The postulated proto-form of Latin vıs, *wel-si, would have become, by regular development vel, a
form attested, but as a conjunction meaning ‘if you will’ (cf. Latin ter ‘three times’ < *ters, Greek
trıs) [Sihler, 1995, 69]. This form lacks the -s that elsewhere marks the second person singular in
the Latin verbal system.
Osthoff proposes that vis is a suppletive form, with Latin *wel-si having fallen out completely
and been replaced by a form *wei-si, which he equates with the Vedic Sanskrit root vı ‘strive,’
‘approach,’ or ‘enjoy’1 [Osthoff, 1881, 486]. The root takes athematic inflection, and its present
tense paradigm is presented in Table 8.2.
1Whitney calls vı “A perplexing root in its varieties of meaning; divided into two or three roots by BR.and Grassmann” [Whitney, 1885, 162].
33
Osthoff’s explanation is satisfying phonetically, but the semantics are not an exact match. Fur-
ther, this instance of suppletion would be the only instance of suppletion within a proximal verbal
paradigm in Latin.
Warren Cowgill proposes the most likely sequence of events in the creation of the attested form
vıs. He postulates an intermediary stage between *wel-s and *vıs, *well [Cowgill, 1978, 26].
Here, the distinction in the two types of Latin l, the pinguis or velar l and the exilis or palatal l,
becomes important. Sihler states “The distribution [of l] was as follows: l exilis was found before
the vowels -i- and -ı-, and before another -l-; l pinguis occurred before any other vowel; before any
consonant EXCEPT l; and in word-final position” [Sihler, 1995, 174].
The form *well would contain an l exilis, a palatal consonant. Speakers would add the second
person singular ending, -s to this form, creating *wells, which is then reduced to *wels, but with
retention of the palatal quality of the l. This palatal l then loses its laterality, creating a form weis
spelled veis and attested by Priscian [Cowgill, 1978, 30]. The form veis becomes vıs by normal
sound law, cf. Old Latin deico > Classical Latin dıco ‘I say’ [Sihler, 1995, 52].
Cowgill explains the infrequent presence of final -ls in Latin in two other words with the fol-
lowing: “uls was archaic - quoted by Varro, Gellius, Festus, but scarcely used in Classical Latin -
while puls was a form that was practically imposed by the structure of the language, if pult- was
not to lack a nominative singular” [Cowgill, 1978, 33].
8.5 vultis ‘YOU(PL.) WISH’
If the Proto-Indo-European paradigm for the root *welH1-2 ‘want’ was inflected after a simple
athematic pattern, Latin vultis ‘you(Pl.) wish’ would come from Proto-Italic *wl.-te >→ Pre-Latin
*wol-tes. If the present showed Narten ablaut, then the Proto-Indo-European form would have been
*welH1-te, becoming Proto-Italic *wel-te, whence *wel-tes in pre-Latin. The development of Latin
vultis follows the same path as vult, detailed in section 8.3.
2Recall that that Latin uses *wel- instead of *welH1.
CHAPTER 9
LATIN esse ‘TO BE’
The copula of Latin and Italic is the most contentious and least understood of the irregular athe-
matic Latin verbs. Table 9.1 provides the Latin present and future, and Table 9.2 provides Proto-
Indo-European reconstructions of the present and subjunctive.
9.1 A DISCUSSION OF PRIOR WORK
9.1.1 THE THIRD PERSON PLURAL
The reconstructed PIE form for the third person plural is *H1senti, as can be seen in Doric enti,
Attic eisı, or Sanskrit santi. Latin sunt, however, appears to point to a formation with o-grade of
the ending, as opposed to the other attested Italic forms: Oscan sent/set and Umbrian sent sug-
gesting a proto-Italic as well as PIE form *H1senti. The Oscan and Umbrian forms here suggest no
alterations at all, save in the case of set, which seems to show a loss of *-n- before -t [Buck, 1904,
Table 9.1: The Latin Present and Future of esse
Person and Number Present FutureFirst Singular sum eroSecond Singular es erisThird Singular est eritFirst Plural sumus erimusSecond Plural estis eritisThird Plural sunt erunt
34
35
Table 9.2: The PIE Copula, Present and Subjunctive
160]. The Latin form then requires explanation. The first option that presents itself is a remaking
with the thematic ending, such that *senti→ *sonti (cf. OCS sotu), (cf. Latin eunt ‘they go’ from
*ey-onti← *yenti, cf. Sanskrit yanti ‘they go’). The resulting form sunt would have the benefit of
homoioteleuton with the third conjugation verbs such as dicunt ‘they say’ as a reinforcing pressure.
9.1.2 THE FIRST PERSON PLURAL
The Italic first person plural is so far unattested in languages other than Latin [Buck, 1904, 160].
The Latin form sumus is generally taken to come from a form *somos, built analogically for any
of several reasons. An initial thought, and one that seems reasonable, is that -sm- is a phonotac-
tically disallowed sequence in Latin. This is generally true, at least in Classical Latin, but more
archaic Latin appears to allow this sequence to stand, at least medially. The Duenos inscription, for
instance, contains COSMIS for later Latin comis [Baldi, 1999b, 200] and IOVXMENTA for iumenta
[Dunkel, 1998, 90]. However, initial sm- is unattested in these old inscriptions, and appears to be
lacking in Latin entirely, though evidence either for or against the possibility of such a sequence
is scanty [Sihler, 1995, 214]. If Sihler is correct in connecting Latin mıror with Vedic smi- and
English smile, then such initial phonotactic constraints might be responsible for some refashioning,
since, if the rule is actually *sm- > m-, the Proto-Indo-European form *H1smos would become
*mos in Latin, a form lacking the s that seems to be the morphological core of the paradigm in syn-
36
chronic terms. Thus, while we can’t use phonotactic constraints to analyze the outcome of *esmi
(as detailed below), such constraints are at least plausible, if not certain, in dealing with *smos.
The reduction of *smos to *mos by regular sound law would be problematic for speakers
learning the paradigm. A number of phonological or analogical shifts could produce sumus. One
possibility is a Lindemann’s law sequence in which *H1smos > *sm. mos, which developed to
*semos. An enclitic form of this word would under raising of *e to i or u, yielding simus or sumus
after the regular development of final *-os to -us. Another possibility better explains the presence
of a biform simus which appears “in the idiolect of Augustus Caesar,” in several lost texts, and in
an inscription SECUTI SIMUS ‘we followed’ [Dunkel, 1998, 87]. A sort of anaptyxis, breaking up
the initial *sm-, giving a short vowel, could cause such biforms, as Sihler explains in his discussion
of the -i-/-u- alternation, saying “Quintilian and others state that the sound in question was inter-
mediate between u and i, and the emperor Claudius is said to have proposed adding a letter to the
alphabet in order to write the sound unambiguously” [Sihler, 1995, 64]. The introduction of such
an intermediate short vowel would serve to preserve the s that appears to characterize the present
tense of the copula while accounting for the variant forms appearing in Latin. The existence of the
biforms suggests another option, however. Simple thematic verbs in the first plural in Latin end in
-imus, and certain other verbs, such as velle ‘to wish’ and its derivatives, have first plurals of the
type volumus. The form *mos could have been discarded entirely and rebuilt by attaching either
-imus or -umus analogically after these other verbs, retaining the initial s- after analogy with sunt.
Another important possibility for the creation of sumus is that of thematization. The Latin verb
velle, ‘to wish, want’ shows thematic forms where other languages show athematic forms, i.e. Latin
replaced *wel-mi with *wel-o-H2 and *wel-mos with *wel-o-mos, yielding volo and volumus by
regular sound change (see [Dunkel, 1998, 97] for explanation of the presence of -u-, rather than
-i-, in volumus). If *smos were thematized like *welmos, the resulting *somos would then undergo
regular sound change in an enclitic form as detailed above, yielding sumus.
37
9.1.3 THE SECOND AND THIRD PERSONS SINGULAR
On the surface, the forms es and est appear to require only loss of the hic-et-nunc particle to be
established. However, much scholarly ink has been spilled over the presence of enclitic forms
-s(s) and -st that appear in certain contexts in Plautus. The following lines give a survey of the
distribution of the forms:
Contracted Forms:
non iratass? for non irata ess? ‘you’re not angry?’
quis homo sit magis quam tuss?’ for quis homo sit magis quam tu es? ‘what man
is greater than you are?’
idem mihist. . . vitium for idem mihi est vitium ‘I have the same fault.’
quod factost opus for quod factum est opus ‘what work was done’
Uncontracted Forms:
mulier ess, audacter iuras ‘You’re a woman, you swear boldly.’
quid est id, soror? ‘What is it, sister?’
quod tibi nomen est, fecit mihi ‘What your name is, he gave to me’ (All cited in
Nyman, 1977, 42).
38
Nyman presents the rules for the alternation thus: ”ess, est after consonants, -ss, -st after vowels”
[Nyman, 1977, 42].
However, the rules aren’t exactly that, as shown by factost for factum est above. The rule must
assume the presence of poetic elision. The elision found variously in Horace, Ovid, and Vergil
functions as follows: “...when a word in a verse ends with a vowel or a diphthong or a vowel
followed by -m, and the following word begins with a vowel or a diphthong or a vowel preceded by
h-, the final syllable of the first word and the beginning syllable of the next were usually collapsed
or ‘elided,’ into a single syllable. The quantity of the single syllable resulting from such ‘elision’
was usually that of the second syllable (suggesting that, in some cases at least, the preceding vowel
or diphthong was in pronunciation actually dropped altogether, as the term ’elision,’ from elidere
‘to compress or strike out’ might suggest)” [LaFleur, 1999, xxii].
If this sort of poetic elision began by the time of Plautus, then such forms might merely be early
examples, especially given the frequency of occurrence in past passive participles of the neuter in
-um or the feminine in -a.
Also of note here is that SCRIPTUM EST appears several times in the Senatus Consultum de
Bacchanalibus, and the sequence SACRI EST appears once, showing a lack of the type of contrac-
tion that the works of Plautus display in a text contemporary with the playwright [Baldi, 1999b,
210]. The lack of contraction in a high-style prose text affirms the status of -st and -s(s) as real
enclitic biforms. However, forms such as opust for opus est also appear [Nyman, 1977, 45], forcing
us to revise our suspicions about the postulated enclitic -st.
Further, Nyman postulates a loss of final -s such that the sequence of contraction goes opus
-st > opu’ -st > opust [Nyman, 1977, 45]. However, the Senatus Consultum de Bacchanalibus
shows that contemporary Latin orthography didn’t require the writing of the double s, including as
it does forms such as ESENT for essent, where an orthographic single s represents an etymological
and actual double s [Baldi, 1999b, 200]. It is likely therefore that the sequence of contraction in a
situation like this is opus -st > opust. In instances like this, haplology, though Nyman claims to
have discredited the possibility, still seems to be a reasonable explanation, especially if est can be
39
enclitic and thus contain an unstressed vowel. The sequence *sVs > ss under conditions of weak
stress is far from implausible.
9.1.4 THE SECOND PERSON PLURAL
In the case of estis, the explanation is quite straightforward. The Proto-Indo-European form *H. 1ste
would regularly appear as *stes in Latin and requires only a contamination with est (based on the
shared -st sequence) to generate the form estis < *estes with raising of *e in final syllables.
9.1.5 THE FIRST PERSON SINGULAR
The story of the Italic first personal singular forms is a complex one. In Latin generally the athe-
matic first person singular ending *-mi has been replaced with *-o-H2, as in volo ‘I wish’ < *wel-
o-H2← *wel-mi. However, sum shows no such replacement. A list of Italic forms will prove useful
here. In Latin, we have the basic form sum, but additionally esom, attested on the Garigliano bowl,
dated to the 5th century BCE [Baldi, 1999a, 169]. In Oscan, sum is present, but also attested are sim
from the area of Saticula, and esum from 6th century BCE Campania [Joseph and Wallace, 1987,
675,692]. Additionally, South Picene shows a form esum [Joseph and Wallace, 1987, 683]. The
earliest forms attested all point to a proto-Italic first person singular form *esom, but how are we
to understand the origin of such a form in the light of PIE *esmi?
Many earlier treatments, such as that in Palmer suggest that sum is an entirely analogical cre-
ation, built after sumus, but the presence of esom contradicts this possibility [Palmer, 1961, 269].
Analogy on the proto-Italic level is not entirely out of the question, but the analogical source form
cannot be sumus.
Another possible explanation is partial thematization of the original athematic form. The intro-
duction of the thematic vowel -o- before the ending could have resulted in a sequence like *esmi
→ *es-o-mi > esom with the loss of the hic-et-nunc particle. However, this particular style of
thematization is otherwise unattested in Latin. The usual thematization process, as seen in eo ‘I
go’ and volo involves the complete replacement of the athematic *-mi ending with *-o-H2, which
40
would result in *es-o-H2, a form identical with the Latin future, ero. We likewise cannot postu-
late a thematicization of the form *esmi→ *esomi based on grounds of disambiguation from the
future, since the Oscan and Umbrian future is not based on the PIE subjunctive of *es-, but is rather
derived from a different IE root, *bhu [Buck, 1904, 160]. The ambiguity would occur only within
Latin, leaving the other Italic branches to use the form normally.
Phonotactic difficulty is another tack tried in explaining sum, but as shown previously, medial
*-sm- can stand in more archaic Latin, and -sN- clusters are present in both Oscan and Umbrian
[Joseph and Wallace, 1987, 691]. A final *-sm, following apocope of the final hic-et-nunc -i, might
well produce considerably more difficulties. The sequence would be likely to develop an anaptyctic
vowel, resulting perhaps in *esem (cf. The Latin treatment of Proto-Indo-European *m. , as in Latin
septem < Proto-Indo-European *septm. ‘seven.’ Even though the final *-sm sequence of ‘I am’ is
generated long after the era of Proto-Indo-European, a similar outcome is plausible.) or perhaps
esom with a rounded vowel from the labial character of the m, which produces the attested form.
The enclitic behavior of the copula can then explain the loss of the initial e-, or one can turn to
analogy with sumus.
Another possible method for deriving sum appears in the existence of subjunctive biforms that
appear in Sanskrit. The basic Indo-European subjunctive appears in Latin ero (cf. Sanskrit asa(ni)),
but beside this subjunctive form (*H1es-o-H2), there appears in the Yajurveda a form asam, a sub-
junctive built with the *-m ending instead of the -o-H2 (i.e. *es-o-m), the exact form that appears
in proto-Italic [Dunkel, 1998, 95]. Sanskrit asam is likely secondary, however, an analogical sub-
junctive built to a subjunctive like bharam, a form that itself is an analogical construction, probably
after the imperfect of the Sanskrit paradigm1, and the form is therefore unlikely to underlie any
Italic verb form. The concept of semantic fading of subjunctive forms to indicative forms is con-
ditioned by politeness, and is well attested in various IE languages [Dunkel, 1998, 91]. We must
presume, then, that however this Proto-Italic*esom was built, it replaced whatever the natural out-
1The proportional analogy would look something like imperfect indicative abharama : subjunctivebharama :: imperfect indicative abharam : X, X = new subjunctive bharam, instead of regular first sin-gular subjunctive bhara(ni).
41
Table 9.3: The Greek Thematic Verb
Person and Number Greek Form Thematic VowelFirst Singular phero oSecond Singular phereis eThird Singular pherei eFirst Plural pheromen oSecond Plural pherete eThird Plural pherousi o
come of *esmi was in Latin, giving rise, through enclisis, aphaeresis, or analogy to sum, as detailed
above.
9.2 A NOVEL AND UNITARY APPROACH TO THE PARADIGM
The above methods perform reasonably well in showing possible derivations from the Proto-Indo-
European forms individually. However, an approach looking at the paradigm as a whole, with much
interaction between the forms at an early stage of Latin, can produce the paradigm without resort
to many of the ad hoc methods discussed above.
The paradigm of Latin sum displays the basic vowel pattern seen in the Indo-European thematic
verb, i.e. *o in the first person singular and the first and third persons plural and *e in the second
and third persons singular and second person plural, a pattern setting up a sort of identity in the
minds of speakers, connecting the first personal singular with the first and third persons plural, and
then the second and third persons singular with the second person plural. This mental connection
is the critical fact on which the following analysis of the structure of the sum paradigm hinges. For
a clear example of the vowel pattern in the Indo-European thematic present, see Table 9.3 for the
Greek thematic paradigm of phero ‘I bear.’
42
Proto-Italic speakers would have perceived the o-e-e-o-e-o pattern before the eventual reduction
(cf. Latin third conjugation) to o-i-i-i-i-u2.
I’ll now discuss the formation of the Classical Latin copula form by form.
9.2.1 sunt
The Proto-Italic form *senti gets rebuilt only in Latin (recall that analogy is an irregularly applied
process in human language) with the thematic ending -onti, cf. eunt and volunt, two athematic
forms that get rebuilt in Latin with -onti, giving a form *sonti, which loses the hic-et-nunc particle
and then undergoes vowel raising to -u-, a process supported by homophony with third conjugation
verbs.
9.2.2 sumus
The Proto-Italic *smos is phonotactically uncertain, if not surely disallowed, and was probably ana-
logically altered to *somos based on the vowel pattern shown above, supported by a proportional
analogy of the type *legont : *sont :: *legomos : X, X = *somos.
9.2.3 sum
The Proto-Italic form *esmi loses the hic-et-nunc particle at the same time that *esi and *esti do,
yielding *esm. . The normal outcome of Proto-Indo-European final *m is -em (cf Latin septem <
*septm. ), but here, we need a development from a proto-Italic *-m. , and a reasonable hypothesis
is -om, giving esom, which is directly attested by the Garigliano bowl and the various other Italic
forms. The form then undergoes raising of *o > u in final syllables, yielding the form esum, which
Varro claims was an archaic form of sum [Dunkel, 1998, 88]. To understand the change from esum
to sum, we need to return to the analysis of elision in section 9.1.3. The critical data point for under-
standing this is the form opust for opus est, showing an instance of haplology of the type *-sVs-
2A o-i-i-u-i-u stage of the Latin third conjugation seems likely, based first on the Proto-Indo-Europeanetymon of the first person plural and second on the presence, even in Classical Latin, of third conjugationfirst plurals in -umus.
43
> -ss- <-s->. In a nearly identical process, a periphrastic perfect like *secutus esum ‘I followed’
with an enclitic copular form would be *secutusum or secutussum, given a reduction of the type
*-VsVsV- > *VssV. This process leaves two first singular forms, a stressed esum and an enclitic
sum. Rhotacism, however, causes a change of esum > *erum, a form lacking the s that, in syn-
chronic terms, marks the present of the copula in Latin, and has the -r- that marks the Latin future
and imperfect forms. The stressed erum is then discarded, and the enclitic sum takes over all first
person singular present active indicative copular duties3.
9.2.4 estis
Proto-Italic *stes is remade as *estes after a comparable paradigmatic identity with the second and
third person singular forms (in Old Latin ess and est). The vowel reduction to estis provides a
reinforcing paradigmatic identity in these three forms, giving a unitary stem es- followed by the
standard Latin Personal endings (-s, -t, and -tis).
3Recourse to analogies such as those suggested for sumus in section 9.2.2 are possible for sum as well,but they do not explain the presence of esom and Varro’s esum.
CHAPTER 10
CONCLUSIONS
The six irregular verbs of the Latin present system have substantial similarities in their structures.
Three of these verbs, ferre, velle, and edere have Latin futures that appear to be derived from Indo-
European thematic subjunctives, and therefore fall at least partially into the Latin third conjugation.
Only esse ‘to be’ has a future paradigm directly cognate with the Proto-Indo-European athematic
subjunctive. Other anomalies associated with this verb are that it has a first person singular form
that hasn’t become thematic, and it alone retains part of the Indo-European simple athematic ablaut
pattern between the singular and the plural. Only edere ‘eat’ retains the Narten ablaut pattern. The
paradigm of dare shows athematic inflection, with endings appended directly to a root *dH. 3-
> da-. Elsewhere, intraparadigmatic ablaut is leveled out of the Latin present tense.
These facts suggest that the Indo-European athematic inflectional type was near extinction by
the literate period of the Italian peninsula. Indeed, some of the forms discussed in this thesis, such
as es ‘you eat,’ est ‘eats,’ and estis were replaced by the simple thematic forms edis, edit, and editis
respectively, regularizing the paradigm and drawing it into the third declension.
The fact that the forms discussed herein survived as long as they did is a testament to their
frequency and saliency, and further research into the relationship between irregularity, frequency,
retention, and reduction is warranted in explaining the structure of the Italic system. A paucity of
Vulgar Latin data will likely prevent the creation of a solid, usage-based understanding of the Latin
system, but some usage effects may be observed and explained.
44
BIBLIOGRAPHY
[Baldi, 1999a] Baldi, P. (1999a). Observations on Two Recently Discovered Latin Inscriptions. In
Embleton, S., Joseph, J. E., and Niederehe, H.-J., editors, The Emergence of the Modern Lan-
guage Sciences: Studies on the Transition from Historical-comparative to Structural Linguistics
in Honour of EFK Koerner. John Benjamins, Philadelphia.
[Baldi, 1999b] Baldi, P. (1999b). The Foundations of Latin. Walter de Gruyter, Berlin.
[Buck, 1904] Buck, C. (1904). A Grammar of Oscan & Umbrian. The Atheneum Press, Boston.
[Bybee, 1985] Bybee, J. (1985). Morphology: A Study of the Relation between Meaning and Form.
John Benjamins, Philadelphia.
[Bybee, 2001] Bybee, J. (2001). Phonology and Language Use. Cambridge University Press,
Cambridge, United Kingdom.
[Cowgill, 1978] Cowgill, W. (1978). The Source of Latin vıs ‘Thou Wilt’. Die Sprache, 24:25–44.
[DeWandel, 1982] DeWandel, N. (1982). The origins and development of the Latin present
system. Unpublished Ohio State University Dissertation, University Microfilms International,
Ann Arbor, Mich.
[Dunkel, 1998] Dunkel, G. E. (1998). On the “Thematisation” of Latin sum, volo, eo and edo
and the System of Endings in the IE Subjunctive Active. In Jasanoff, J., Melchert, H. C., and
Oliver, L., editors, Mır Curad. Studies in Honor of Calvert Watkins, pages 83–100. Institut fur
Sprachwissenschaft der Universitat Innsbruck, Innsbruck.
[Fortson, 2004] Fortson, B. (2004). Indo-European Language and Culture: An Introduction.
Blackwell Publishing, Malden, Ma.
45
46
[Hooper, 1976] Hooper, J. (1976). Word frequency in lexical diffusion and the source of mor-
phophonological change. In Christie, W., editor, Current Progress in Historical Linguistics,
pages 95–105. North-Holland Pub. Co., New York.
[Jasanoff, 2004] Jasanoff, J. (2004). Plus ca change. . . lachmann’s law in latin. In Penney, J.
H. W., editor, Indo-European Perspectives: Studies in Honor of Anna Morpurgo Davies, pages
405–416. Oxford University Press, Oxford and New York.
[Joseph and Wallace, 1987] Joseph, B. and Wallace, R. (1987). Latin sum, Oscan sum, sim, esum.
American Journal of Philology, 108:675–693.
[LaFleur, 1999] LaFleur, R. (1999). Love and transformation: an Ovid reader. Scott Foresman-
Addison Wesley, Glenview, Ill.
[Lindeman, 1976] Lindeman, F. (1976). L’apophonie radicale au present-imparfait actif des verbes
athematiques en indo-europeen. Bulletin de la societe de linguistique, 71:113–121.
[Narten, 1968] Narten, J. (1968). Zum ‘proterodynamischen’ Wurzelprasens. In Heesterman,
J. C., Schokker, G. H., and Kuiper, V. I., editors, Pratidanam: Indian, Iranian and Indo-
European Studies Presented to Franciscus Bernardus Jacobus Kuiper on his 60th Birthday,
pages 9–19. Mouton & Co., The Hague, Netherlands.
[Nyman, 1977] Nyman, M. (1977). Where Does Latin “Sum” Come From? Language, 53:39–60.
[Osthoff, 1881] Osthoff, H. (1881). Zu der altlateinischen dvenos-inschrift. Rheinisches Museum
fur Philologie, 36:481–489.
[Palmer, 1961] Palmer, L. (1961). The Latin language. Faber and Faber, London.
[Phillips, 2001] Phillips, B. (2001). Lexical diffusion, lexical frequency, and lexical analysis. In
Bybee, J. and Hopper, P., editors, Frequency and the Emergence of Linguistic Structure, pages
123–136. John Benjamins, Philadelphia.
47
[Pluymaekers et al., 2005] Pluymaekers, M., Ernestus, M., and Baayen, R. (2005). Lexical fre-
quency and acoustic reduction in spoken Dutch. The Journal of the Acoustical Society of
America, 118:25–61.
[Schmalstieg, 1983] Schmalstieg, W. (1983). An Introduction to Old Church Slavic. Slavica,
Columbus, Oh.
[Seldeslachts and Swiggers, 2002] Seldeslachts, H. and Swiggers, P. (2002). A propos de la
flexion de Latin velle. In Sawicki, L. and Shalev, D., editors, Donum Grammaticum: Studies in
Latin and Celtic Linguistics in Honour of Hannah Rosen, pages 317–328. Peeters, Sterling, Va.
[Sihler, 1995] Sihler, A. (1995). New Comparative Grammar of Greek and Latin. Oxford Univer-
sity Press, New York.
[Whitney, 1885] Whitney, W. (1885). The roots, verb-forms, and primary derivatives of the San-