Bryn Mawr Classical Review

Reviewed by Wolfgang David Cirilo de Melo, Ghent University (wolfgang.demelo@gmail.com)

Version at BMCR home site

Just as tense expresses temporal deixis, demonstrative pronouns mark spatial deixis; and just as tense is a category that is notoriously difficult to describe because its usage is influenced by factors other than pure time reference, demonstrative pronouns are semantically complex because they can express more than physical distance. Within the Latin pronominal system, hic indicates that a referent is near the speaker or has some other connection with him, whereas iste is usually said to point to the addressee and ille is used for entities connected with neither; is is entirely endophoric and can only be employed for discourse-internal referents. Pronominal systems change easily: in Spanish, éste, from Latin iste, has become a first-person deictic; ése, from non-demonstrative ipse "himself," now functions as second-person deictic; aquél, the third-person deictic, continues a strengthened ecce ille; and endophoric is has been replaced by él, which goes back to Latin ille.

Traditionally, Oscan, Umbrian, and South Picene, the three languages which form the Sabellic branch of the Italic family, have been regarded as very similar to each other and have been examined through the lens of Latin. Such a procedure makes sense not only because these languages are in a close genetic and areal relationship with Latin, but also because they are attested through fairly small corpora, which makes independent examination difficult. But of course this way of looking at things brings certain risks with it; the peculiarities of individual languages can easily be overlooked. In his book, Dupraz promises to study the demonstrative systems of the Sabellic languages from a typological perspective, without imposing a Latinate frame on them, but not only do the three major branches of Sabellic have demonstrative systems nearly identical to each other in his final analysis, their shared system also bears a remarkable similarity to the way Dupraz believes Latin to function.

The structure of his book is straightforward. A brief introduction is followed by one chapter discussing theoretical issues of deixis. After this there are five chapters dedicated to the individual demonstrative pronouns, classified according to the stems which are traditionally reconstructed for Proto-Italic; this point is important here because especially Umbrian has undergone a number of sound changes which make it difficult to assign some attested forms to one proto-form or another. Chapter 7 is dedicated to a synchronic comparison between the Latin and Sabellic demonstrative systems. The final chapter is an attempt to reconstruct Proto-Italic deictic pronouns. A brief conclusion is followed by a thorough bibliography, an index locorum, and an index uerborum; the latter is particularly useful because Dupraz does not give any paradigm tables in the main body of the book, which means that accidental gaps in the attestations as well as systematic gaps due to suppletion within paradigms can only be gleaned from here.

In my review, I shall not follow the order of Dupraz's chapters. Chapter 6, which analyses obscure and isolated forms, discusses for instance Oscan and Umbrian esuf / esuf,¹ which corresponds to Latin ipse semantically, but has a different etymology. As none of the forms in this chapter have much impact on the overall theory advanced by Dupraz, I shall not dwell on them here.

Chapter 4 is uncontroversial. It deals with third-person deixis. Umbrian has the adverbial ulu / ulo "there," which continues the stem *ōl- also found in Latin ōlim "then." The Oscan pronominal forms continue a stem *ŏll- as in Early Latin ŏlli "to that man."²

Chapter 5 contains the longest discussion in the book. It deals with the endophoric pronouns, which are etymologically related and semantically equivalent to Latin is "he / this." Latin is exhibits considerable stem allomorphy: is continues a zero-grade *i-, ea a full-grade thematic stem *eio-, and eius a recharacterized genitive *esyo-s. The Sabellic situation is even more complex. The masculine nominative singular and the neuter nominative / accusative singular continue a zero-grade stem *i-, but this stem has been recharacterized by the original neuter nominative / accusative singular *id, yielding forms like Oscan izic "he" < *is-id-ke.³ The stem *eyo- is used for the remaining forms of the nominative and accusative, and *eyso- is employed for the other cases. The origin of the stem *eyso- is obscure; perhaps it arose by reanalysis of the old genitive plural *ey- sōm as *eys-ōm.

The really problematic parts of Dupraz's book are chapters 2 and 3. Dupraz notes that Oscan has a suppletive stem *eko- / *ekso- used for first-person deixis. The forms continuing *eko- are employed for the nominative and accusative, for instance ekak, the feminine accusative singular; the forms continuing *ekso- are used for the oblique cases, for example exac, the feminine ablative singular. Umbrian shows no reflexes of *eko-, but Dupraz believes that forms like the masculine ablative singular essu continue *ekso-. The Sabellic languages have a second suppletive stem for first-person deixis, *esto- / *esmo-. According to Dupraz, reflexes of *esto- are used for nominative and accusative, for instance the Umbrian masculine accusative singular estu, whereas those of *esmo- are employed in the other cases, as in the Umbrian masculine dative singular esmei. This suppletive paradigm is attested for Umbrian and South Picene, but not for Oscan. Dupraz believes that its absence in Oscan is due to chance.

The accidental absence of forms in a poorly attested language would not be surprising in itself, especially if motivated semantically. Latin republican inscriptions form a large corpus, yet iste is almost completely absent from them because it is used for second-person deixis, which is hardly necessary in inscriptions. However, Dupraz considers both *eko- / *ekso- and *esto- / *esmo- to be employed for first-person deixis, and that makes the absence of the latter in Oscan conspicuous. In chapter 7, where he compares usages in Sabellic and Latin, he admits that this distribution pattern is indeed "problematic" (p. 269). But his solution is unconvincing. He treats both hic and iste in Latin as proximal deictics, claiming that the difference between them is that iste is "dialogical," indicating the speaker's concern, while hic is non-dialogical. This unorthodox view is not even borne out by the examples from Cato's prayers he cites in support of his theory (agr. 132. 1-2 and 134. 2-3, discussed on p. 272-5). This rather strange distinction is then said to exist in Sabellic as well, where *eko- / *ekso- is said to correspond to hic, whereas *esto- / *esmo- is said to correspond to iste. However, the question still remains why Oscan should use the non-dialogical forms in epitaphs, while South Picene should use the dialogical ones. Dupraz argues (p. 282-3) that there are political reasons for this: South Picene epitaphs are aristocratic and their content is a matter of political concern, hence the use of dialogical pronouns; by contrast, in Oscan the aristocratic structures are said to have been replaced by republican, collegial ones, which means that epitaphs are no longer a matter of political concern and therefore non-dialogical. To my mind, this argument contains far too much special pleading to be sound. Dupraz has followed the traditional reconstructions, but one can only make sense of them if one accepts untraditional, very odd theories of deixis. It would have been better to start from the synchronic meanings and distribution patterns of the forms and base the reconstruction on these.

There is an alternative way of dealing with these forms, proposed by Penney (2002),⁴ but explicitly rejected by Dupraz (p. 111-5). Penney starts from the semantically convincing assumption that both *eko- / *ekso- and *esto- / *esmo- are used for first-person deixis without any differences in meaning. He argues that *eko- / *ekso- is restricted to Oscan, which does not have *esto- / *esmo-, whereas Umbrian and South Picene have *esto- / *esmo-, but not *eko- / *ekso-;⁵ the question whether Sabellic also had second-person deixis is left open by Penney, who points out that our corpus is too restricted in length and variety for us to be able to know the entire system. One of Dupraz's objections to this proposal is that it means that the deictic systems in the Sabellic languages would differ from each other; this point of criticism is not convincing, as for instance Oscan also has a perfect in –t(t)- without an Umbrian equivalent, whereas Umbrian has a perfect in –nç- without an Oscan equivalent. More important is the question how Penney can handle the Umbrian forms which Dupraz believes to go back to *ekso-. Due to Umbrian assimilations, a cluster *-ks- cannot show up as such in forms like the masculine ablative singular essu, and–ss- can also go back to other consonant clusters. Penney assumes that the paradigm had a three-way split and that the allomorphs were *es-to-, *es-t-so- > *esso-, and *es-t-so-mo- > *esmo-; a proto-form *ekso- did not exist in Umbrian and South Picene. The distribution of the allomorphs is as follows: *esto- in the nominative and accusative (but *esso- in the feminine nominative singular), *esso- in the genitive and ablative, and *esmo- in the dative and locative. The feminine nominative singular could either be an innovation based on one of the oblique stems or it could reflect the Indo-European distribution of *to- and *so-, as evidenced in their Greek reflexes. Dupraz objects to this degree of allomorphy, but reconstructs a five-way split in the paradigm of *i- for Proto-Italic (p. 298) and notes that typologically, polymorphism is common among demonstratives (p. 298-9, n. 49). For all these reasons, Penney's reconstruction seems preferable to me.

As I cannot agree with Dupraz's synchronic analysis entirely, I cannot fully accept his Proto-Italic reconstructions either. He reconstructs a stem *ekso- for Proto-Sabellic, but argues that *eko- is an Oscan innovation. Yet it is typologically unusual for suppletive paradigms to be created in this way; normally they come about because two pre-existing paradigms merge, and where there are really new suppletive forms, they tend to be longer, not shorter, than the forms they are based on. As for *esmo-, Dupraz believes that *i-, which originally had first-person deixis, had an allomorph *esm- in Proto-Italic, which acquired a theme- vowel and developed an independent paradigm with first-person deixis; *esto- was created as a suppletive form. Again this does not tally with how suppletion normally arises. For those who accept the standard theories concerning Latin demonstrative pronouns (and thus not for Dupraz), the differences in deixis between Latin iste and Sabellic *esto- require a historical explanation. A change from first-person deixis to second-person deixis within the history of Latin is certainly conceivable; the opposite change within the history of Umbrian and South Picene is equally possible, as the Spanish parallel at the beginning of my review shows.

In this review I have focused on the less convincing aspects of Dupraz's book. Despite my criticism, it is a work that deserves to be read: Dupraz lists and discusses all demonstrative forms in the Sabellic languages, and for most stems his analyses are undoubtedly correct. Moreover, the exposition is lucid and a joy to read. The book has been well produced. Infelicities in the English and typographical errors are almost completely absent.

1.   Roman is used to transliterate native alphabets, while italics stand for the Latin script or reconstructed forms.
2.   Classical Latin has ille with i- rather than o- by analogy to is.
3.   The particle *-ke, also found in Latin hi-c or huius-ce, can be attached to endophoric pronouns in Sabellic, unlike in Latin, where is is never combined with –ke.
4.   Penney, J. H. W. (2002), "Notes on some Sabellic demonstratives," Oxford University Working Papers in Linguistics, Philology and Phonetics, 131-42.
5.   The position of South Picene within Italic is unclear. Penney's proposal does not entail that South Picene is closer to Umbrian than to Oscan; it could be a fully independent branch within Sabellic.