ON THE LOSS OF FINAL - M : PHONOLOGICAL OR MORPHOSYNTACTIC CHANGE?

Summary: This paper intends to show that when grouping the various kinds of omissions of final - m in Väänänen’s study on the Vulgar Latin of Pompeian inscriptions, the subcategories in his category b) (‘ m omis sans raison apparente’ i.e. where - m is omitted due to a phonetic process) as “Accusatifs en - a(m) ” like Succesus amat ancilla(m) and ad porta(m) Romana(m) or “Accusatifs en - e(m) ” such as qu(a)e amas Felicione(m) and ante aede(m) must be rearranged in the following two subcategories: 1) cases after prepositions like ad porta(m) Romana(m) and ante aede(m) etc. where besides the phonetic interpretation a parallel morphosyntactic explanation of case confusion cannot be ruled out; and 2) cases with the ob- jective use connected to verbs like Succesus amat ancilla(m) and qu(a)e amas Felicione(m) where, due to the preference of the phonetic interpretation, the morphosyntactic explanation seems to be less probable or even unlikely.

In his study on the Vulgar Latin of Pompeian inscriptions, Väänänen dealt with the problem of the dropping of final -m in a peculiar way. 1 He grouped the omissions of final -m in three categories as follows: a) cases where the omission of -m can be explained by non-phonetic, i.e. technical reasons (e.g. by the lack of space at the end of a line such as ad ampitheatru(m) |, 2 or by a potential abbreviation such as plurima(m) salut (em) or before a word-initial m-such as cu(m) media) or morphosyntactic reasons (by confusion of the cases such as the use of the nominative for the accusative, e.g. halica(m) … palmas or the ablative for the accusative, i.e. in conventu (m) veni with a facultative explanation by the dropping of final -m); b) where -m is omitted without any reason ('m omis sans raison apparente'), i.e. due to a phonetic process, such as inqua (m) or collegioru (m) and Succesus amat ancilla (m) or ad porta (m) Romana(m) etc.; and c) where a final -m is added hypercorrectly such as peperit die{m} Iovis or sine dulcissima{m} etc.
Väänänen's categorisation is problematic on more than one account, however. Several items of category b) ('m omis sans raison apparente') could be inserted in category a) as well, since the reason for an omission is not necessarily phonetic; in many cases it can also be morphosyntactic. A considerable part of Väänänen's examples consists of either "Accusatifs en -a(m)" like Succesus amat ancilla (m) and ad porta (m) Romana(m) or "Accusatifs en -e(m)" such as qu(a)e amas Felicione(m) and ante aede (m), which could all be theoretically interpreted also as examples of confusing the cases, and -contrary to items like inqua (m) or collegioru(m) -could therefore be excluded from a purely phonetic analysis.
With the help of the Computerized Historical Linguistic Database of the Latin Inscriptions of the Imperial Age, 3 the present paper tries to show that in cases like ad porta (m) Romana(m) and ante aede (m) or Succesus amat ancilla (m) and qu(a)e amas Felicione(m) the potential influence of morphosyntactic changes cannot be left out of consideration, since in Vulgar Latin the merger of the accusative and ablative cases was a general process 4 affecting all declensions both in the singular and in the plural, occurring in prepositional phrases as well as without prepositions. Thus it seems unfounded to leave morphosyntactic explanations out of consideration concerning all or at least some of Väänänen's example types cited above. 5 1. First we will discuss the prepositional phrases of the type ad porta (m) and ante aede(m) of Väänänen's category b) ('m omis sans raison apparente'). In these items Väänänen admits the phonological explanation by dropping the final -m exclusively, but as we shall see a morphosyntactic explanation by case confusion between the ablative and the accusative seems to be very likely for the following reasons. 6 Firstly, there are several examples of the confusion of the cases after prepositions in singular which cannot be explained by phonological changes at all. Here I do not mean cases like POST FVNERE for post funus , which is to be explained by case confusion only and is therefore exclusively coded by abl. pro acc. in the LLDB Database, i.e. the use of the ablative instead of the accusative. FVNERE could theoretically also be explained by the dropping of the final -m, provided that we assume a previous change from the neuter to the masculine gender due to the process of the disappearance of the neuter gender, so this way funere(m) could be interpreted as a masculine accusative by the dropping of -m. I mean rather those cases where a phonetic explanation is impossible, such as CVM QVEM for cum quo (LLDB-26800: acc. pro abl.) or can hardly be assumed, as in CVM MARITVM for cum marito (LLDB-46788: acc. pro abl.). 7 For this type of case confusion in the singular after a preposition, there are a total of 137 data forms in the LLDB Database: 77 items concern the preposition cum (40 times occurring before quem as CVM QVEM for cum quo LLDB-26800: acc. pro abl. and 37 times before other words of the 2nd declension such as CVM A|MICVM for cum amico LLDB-67505: acc. pro abl. etc.). This use of the accusative instead of the ablative can also often be found with other prepositions like in ( To these 137 occurrences in the singular, which can hardly be explained by phonetic processes, are to be added those 21 items where in the neuter gender the accusative singular, formally identical with the nominative, is used instead of the ablative singular, such as EX VOTV|M for ex voto, LLDB-1235: nom./acc. pro abl. and RECESSIT DE SECV|LVM for recessit de saeculo LLDB-16108: nom./acc. pro abl., or SINE CRIME|N for sine crimine, LLDB-66279: nom./acc. pro abl.
All these 158 items in the singular are to be explained exclusively by case confusion, i.e. by the use of the accusative instead of the ablative, while a phonetic explanation is impossible or improbable.
Secondly, we shall discuss here the confusions in the plural after prepositions, since a phonetic explanation is impossible here, too. For this type in the plural we have 56 items in all where the accusative is used for the ablative, on 27 occasions after pro ( These 56 items are complemented by those 27 occurrences where, conversely, the ablative plural is used instead of the accusative after prepositions (in the LLDB Database coded by dat./abl. pro acc. because of the formal identity of the ablative with the dative), often (16 times) after ob (14 times as in OB MERITIS SVIS for ob merita sua LLDB-72741: dat./abl. pro acc., but we also have OB () AD|FECTIB for ob () adfectus LLDB-70282: dat./abl. pro acc. and OB HO|NORIBVS for ob honores LLDB-58148: dat./abl. pro acc.), six times after inter (e.g. INTER EIS for inter eos LLDB-50612: dat./abl. pro acc. or INTER NOBIS for inter nos LLDB-47039: dat./abl. pro acc.), four times after per (e.g. PER ANNIS for per annos LLDB-42548: dat./abl. pro acc., or PER QVIBVS for per quos, LLDB-63362: dat./abl. pro acc.) once after ad (VADAM | AD FRAGIS for eamus ad fraga LLDB-47111: dat./abl. pro acc.) and once after post (POST DIES () DATIS for post dies datas, LLDB-17314: dat./abl. pro acc.). 8 Finally, to the 83 items for case confusions in the plural, we have also to add those 21 items where the accusative plural, formally identical with the nominative and therefore coded as nom./acc. pro abl. in the LLDB Database, is used for the ablative plural 7 times after pro as P|RO NOS for pro nobis (LLDB-68148 nom./acc. pro abl.), 4 times after cum such as CVM () SACERDOTES for cum ()  This way we get exactly (56+27+21=) 104 examples of case confusion in the plural after prepositions that are not to be interpreted as having any phonetic reason. By adding the (137+21=) 158 items in the singular to those in the plural, we have a total of 262 occurrences for the substandard use of the cases after prepositions that are all to be explained exclusively by case confusion, mostly (235 times, 90%) by the use of the accusative instead of the ablative or conversely, by the use of the ablative instead of the accusative (which is rare, occurs only 27 times, 10%): in all these 262 cases a phonetic explanation is impossible or improbable.

) and QVEM ABVIT IN [CON]|IVGIO () ANNIS TRES ET M[EN]|SES SEPTE for quem habuit in coniugio () annis tribus et mensibus septem or quem habuit in coniugio () annos tres et menses septem
In this context also those few cases have to be taken into consideration where all constituents of an ablative absolute clause are undoubtedly put in accusatives and are therefore called an accusative absolute construction like VIVENTES | CARDILIVM | ET AVITAM | FELIX TVRRE for viventibus Cardilio et Avita felix turris ( From the above it is clear that case confusion between the accusative and the ablative happened due to the functional merger of the two cases that eventually resulted in the formation of a merged accusative-ablative case. 15 This has to be taken into consideration for all types discussed here, especially the occurrences after prepositions and also all structures without prepositions and connected to verbs of no objective function since case confusion occurs also in the plural of all these types. 3. Finally, in the third part we deal with the problem of the type Succesus amat ancilla(m) and qu(a)e amas Felicione(m) recorded by Väänänen exclusively as the result of the dropping of final -m. Theoretically, a facultative morphosyntactic explanation by case confusion between the ablative and the accusative can also be taken into consideration here, especially regarding the direct objects ancilla (m) and comparavit for arcas comparavit etc. in the inscriptions. Based on the rates after prepositions, where the ablative was used instead of the accusative in 10% of the records, we would expect to have at least in ca. 10% of all plural phrases like memorias posuit etc. occurrences of faulty phrases like memoriis posuit etc. However, this is not the case, as we will demonstrate below.
This potential counter-argument based on the relative rarity or even lack of confusion between the accusative and the ablative as for the objective use of the accusative (of the type curam egit, memoriam posuit, arcam comparavit etc.) in the plural as evidenced in inscriptions has a weak point. While it could be an argument against the morphosyntactic explanation, it could probably be offset by the frequency analysis to be presented in the following  Table 1 above might not be representative of this problem, since they are (mostly or entirely) used in the singular only. Thus, with re-gard to the high number of occurrences for ollas dare etc. alongside ollam dare etc., we cannot cite the non-existence or extreme rarity of plural forms as a potential explanation for the non-existence or extreme rarity of case confusion in the plural forms as for the objective use of the accusative.
Incidences of objective phrases in the plural measured against those in the singular: olla ollam dare etc. 127 (36%) 228 (64%!) ollas dare etc. Table 2 What is more, we have found 3 examples for the case confusion, i.e. the use of the accusative for the ablative in the case of olla in the plural but only after prepositions as in CVM OLAS for cum ollis (LLDB-77798: acc. pro abl., CIL VI 25359, 7) and cum ollas duas for cum ollis duabus (CIL VI 28759, 3) or de olas for de ollis (CIL VI 29983, 2). This means that what we have stated about the case confusions after prepositions applies to olla, too. The evidence of the word olla, however, must not be overestimated because its attestation is very limited both geographically and chronologically, since nearly all occurrences come from the city of Rome and from the pre-Christian era, thus they only bear witness to the first three centuries AD and to the capital of the Roman Empire. 20 The remaining rare examples for the accusative-ablative confusion in the plural as for the direct object may have to be explained by another factor or by other factors. For example QVAE ANNOS AETATIS AGENS SEX ET DECE MESIB OCTO for quae annos aetatis agens sex et decem menses octo  dat./abl. pro acc., CIL VI 23629, 6), where agens is composed first with the accusative annos but then with the ablative mensibus, both the impact of case variation after vixit and metrical factors (metri causa solution) have to be taken into consideration. In another example in the plural, i.e. VERVS PROF EST () FVND TAVRIANIS DOVB| for Verus professus est () fundos Taurianos duos, (LLDB-61387: dat./abl. pro acc.), the case confusion 21 can be explained by textual linguistic reasons as well, 20 In EDCS there is only one occurrence labelled as Christian (CIL VI 28576 = ICUR VII 19465) but it dates to the 1st century AD (ICUR comments as: inscriptio… fortasse ne christiana quidem). However, in EDCS there is another item of olla in an inscription dated to 351-400, CIL VI 537: vivax sit spiritus ollis (in a hexameter). 21 Attested in the long text of the Tabula Alimentaria (CIL XI 1147, VI 85, Aemilia / Regio VIII, Veleia, 107 AD) also in the singular such as LLDB-61384: dat./abl. pro acc. / -um > O, VERVS PROF