1Mauro Tosco, one of the most prolific authors in the field of Cushitic linguistics, has recently published a grammar of a little-known variety of the Dullay-cluster (East Cushitic) in southwestern Ethiopia. His grammar of Gawwada [ISO 639‑3: gwd/glottocode: gaww1239] adds to an impressively long list of publications on many Cushitic languages, which includes several grammatical descriptions, namely of the Karre and Tunni varieties of Somali (Tosco 1989 and 1997), the critically endangered language Dahalo (Tosco 1991) and Dhaasanac (Tosco 2001). Following the grammatical sketch by Amborn et al. (1980), focusing on Gollango and Harso-Dobase, and Savà’s grammar of Ts’amakko (2005), Tosco’s Gawwada grammar is the third and most detailed description of a Dullay variety. It is published, both on paper and electronically (in pdf), as the eighth volume of the series Cushitic and Omotic Studies, which is edited by Tosco himself. The present review is based on the paper version.
2With 14 chapters of very different lengths, the grammar has a fairly flat hierarchical structure. The chapters concentrate either on a subdomain of the grammar (e.g. phonology, clause linkage) or on a word class (e.g. pronouns, adjectives). With the exception of §6 on space, movement and deixis, and part of §12 on information structure, the chapters generally adopt a semasiological approach. Across the book, ample cross-references are made between complementary sections. The appendix (p. 340‑387) includes a selection of transcribed, glossed and translated proverbs and riddles, five folktales and a testimonial. These texts and other spontaneous and elicited data were collected during Tosco’s fieldwork in the years 2000‑2010. Tosco worked predominately from Arba Minch, a regional town outside the Gawwada area, but undertook frequent trips to the Gawwada villages about 130 km to the southwest. The vast majority of examples presented in the grammar come from recorded folktales and conversations. The recordings and annotations of three of the appended texts plus another four texts from which examples are cited are accessible, following registration, in the CorpOrAn corpus.
3The introduction (§1) provides information on the classification of the language, language contact issues, the fieldwork setting, and earlier studies on Gawwada and the Dullay dialect cluster, and closes with a summary of salient grammatical features of the language. Importantly, for a reader who has lost track of the ever-changing glossonyms in Ethiopia, it clarifies how the terms “Gawwada”, “Dullay” and “ʕale (‘Ale)” relate to each other. The culturally and politically motivated term “ʕale” refers to Dullay-speaking agriculturalists in the highlands and excludes the Ts’amay pastoralist lowlanders (who are speakers of Ts’amakko). “ʕale” has also recently been gaining ground in the linguistic literature, even though, as Tosco points out, highland Gawwada is more closely related to Ts’amakko in the lowlands than to other highland Dullay varieties.
4The phonology chapter (§2) presents the phoneme inventory, phonological processes and phonotactic rules. Noteworthy members of the consonant inventory are three implosives, three ejectives and a voiced pharyngeal fricative. Gawwada seems to be the only East Cushitic language lacking a phonemic voice distinction for both pulmonic oral plosives and fricatives, whereas in neighboring languages voicing contrasts are only absent in plosives. The typical Cushitic vowel inventory, i.e. five vowels with phonemic length contrast, are presented as ten phonemes, whereas the 22 members of the consonant inventory are not presented as 44, despite all of them also making a phonemic distinction between short (simplex) and long (geminate) realizations. The section on the accent system consists of a mere four sentences (p. 70‑71) and mentions accentual minimal pairs in the verbal system. As later elaborated on (§7, p. 210), the negative imperfective and the affirmative perfective are exclusively distinguished by accent, e.g. ʕuk‑tí (drink-ipfv.neg.2sg) ‘you don’t drink’ vs. ʕuk‑ti (drink-pfv.2sg) ‘you drank’. Throughout the grammar, only elements on which an accentual opposition can arise are marked by an acute accent, i.e. certain suffixes of the negative and the sequential paradigm and the third person subject clitic (called “individualizer”). The prosodic realization of words transcribed without acute accents is not commented on, even though, as Tosco later mentions in passing (p. 75), “words may bear a lexical accent”.
5The brief third chapter defines the units of the morphological analysis (word, stem, affix, clitic), delineates the word classes and points out the productive nature of different reduplication processes. Tosco discards the term “root” for underived stems on the grounds that the distinction between inflection and derivation is “dubious” (p. 75) — only to take a fairly clear stand on what is inflection and derivation in later chapters and to resort to “basic stem” in order to avoid the term “root”.
6The chapter on nominal morphology (§4) is concerned with gender, number and case marking on nouns and specific nominal subclasses. Like closely related languages, Gawwada has three genders (masculine, feminine and plural) and three numbers (preternumeral, singulative and plurative). While gender is an inflectional category and as such is reflected in elements agreeing with nouns, number is rather derivational. Unfortunately, as in many descriptions of Cushitic languages, no strict terminological distinction is made between morphological number markers and semantic number values: a preternumeral is understood to be a morphologically unmarked and semantically number-neutral form (p. 82: “no implication about number”), a singulative as a morphologically marked singular-reference form and a plurative as a morphologically marked plural-reference form (p. 87). However, these one-to-one matches between form and meaning are not always corroborated by the data. As pointed out by Tosco himself, for instance, preternumerals can be “collectives” (i.e. have plural reference), e.g. ɗil‑o (m) ‘Amharas’ (p. 83), t’oonaq‑o (m) ‘bees’ (p. 84), kor‑o (m) ‘people’ (p. 86), and singulatives do not need to have singular reference, e.g. Ɂinn‑akk‑o (sing-m) ‘flies’ (p. 85; more on frozen singulatives on p. 96f.). Gawwada is a language with nominative-accusative alignment but does not mark case on nominal subjects and objects. Tosco considers Gawwada to have only one true case marker, the so-called “associative”, which marks static locations or goals as well as nominal possessors. The vocative case is not very productive.
7Chapter 5 presents a variety of pronouns. Personal pronouns fall into independent, negative (or “emphatic” on p. 123), enclitic subject, enclitic oblique, independent associative and possessive pronouns. The latter are either directly suffixed to the possessee, or they are nominalized and used in apposition to the possessor (so literally “people ours” for ‘our people’). Oblique and possessive pronouns do not only distinguish gender in the third but also in the second person singular. Furthermore, Gawwada has a dedicated vocative and a reciprocal/reflexive pronoun. The “pronominal head” is a gender-sensitive nominalizer. I find the pronoun chapter to be one of the most difficult to follow, as it does not become clear which pronouns constitute a paradigm and how the pronominal distinctions map (or do not map) onto the marking patterns on nominal constituents. Understanding is also hampered, for instance, by the choice of the term “oblique” — not further explained by Tosco — for forms that pronominalize unmarked objects of transitive verbs and by numerous instances of erroneously used “possessee” for “possessor” (p. 87, 137, 260f. and Table 5.9). The “individualizer” and the impersonal clitic could have been treated in §5.4 on the subject pronoun series. According to Tosco, the “individualizer” has two realizations. On the one hand, as proclitic to the verb, it is a third person subject pronoun Ɂí= (see indv in (3) below). On the other hand, it is a suffix ‑í, e.g. on the demonstrative =s‑í (see (3) below) and the contrastive =kk‑í. I am unable to challenge this analysis, but I am missing arguments that Ɂí= and ‑í are in fact distributional variants of the same abstract morpheme. Gawwada’s associative pronouns are said to “cover the same syntactic roles of the corresponding caseform [sic] of nouns” (p. 134), but elsewhere (p. 109‑111) associative-marked nouns are shown to express locative complements/adjuncts as well as nominal possessors, whereas the associative pronouns in the examples of §5.5 refer to locative constituents and beneficiaries, but not pronominal possessors (p. 135). Most interrogative pronouns are missing from §5 but discussed in §12.10.3.
- 1 See the use of =s-í in (1) and =sa in (2) below.
8Chapter 6 on the expression of space, movement and deixis first introduces the deictic form of nouns, which is usually found in front of demonstratives (see (3) below), and then claims that Gawwada has only a distance-neutral demonstrative =sa — which is unusual, if not unique, for languages in East Africa. While I am ready to subscribe to this account of the demonstrative system, the argument is entirely dependent on the analysis of the demonstrative =s‑í, which can contrast with =sa and then locate an entity “closer to the speaker” (p. 151), as being bimorphemic, namely made up of =sa plus the “individualizer” ‑í (see above). I would have liked to read more on why =sí could not simply be considered to be monomorphemic and representing the proximal demonstrative.1 After a discussion of the functions of the associative case and the adessive and applicative enclitic adpositions for the expression of space and movement, the chapter concentrates on positional nouns, including relational and absolute (cardinal) positional nouns. Most remarkable from a genetic and areal perspective is the absolute frame of reference (see Levinson 2003) that Gawwada speakers employ to locate objects or to express directions: the positional noun ‘uphill’ corresponds approximately to northeast, ‘downhill’ to southwest and ‘straight/across’ to a location on an axis orthogonal to the first two. The speaker’s perspective is introduced into this absolute frame of reference by the centrifugal (away) and centripetal (towards) suffixes on the positional nouns. The centrality of the absolute frame of reference is also reflected in other triplets of nouns and verbs for areas in and movement towards a cardinal direction.
9Chapter 7 on verbal morphology covers inflection and derivation. Tosco distinguishes two inflectional classes with distinct suffix sets in three paradigms. Most verbal forms index the subject: 1sg, 2sg, 3m, 3f, 1pl, 2pl, 3pl. Surprisingly, the maximum of seven subject indexes are distinguished only in the sequential paradigm used in dependent clauses; declarative paradigms (used in independent clauses) reduce them to five (with syncretic forms for 1sg/3m and 2sg/3f) or, if negative, to four (with one form for all singular subjects). The affirmative imperative has a singular-plural distinction. Gawwada is an aspect-prominent language (imperfective vs. perfective), a future suffix is the only tense marker, and polarity is marked inflectionally. Furthermore, verbs are marked for habitual “mood” and sequential, imperative and infinitive “modality”. As shown, Tosco adopts a fairly idiosyncratic use of the labels “mood” and “modality” (p. 193), and what I would have considered modality — the expression of necessity, possibility, desire — is not addressed in the grammar. The presentation of derivational morphology jumps back and forth between word class-maintaining and word class-changing processes. Derivation is either suffixal, e.g. the common Cushitic causative, middle and passive extensions, or reduplicative, as the punctual (partial, rightward reduplication), the pluractional (partial, leftward) and the non-productive frequentative (full or partial, leftward). Depending on the semantic class of a verb, the punctual expresses, to put it simply, a single instance or a low intensity of an action, e.g. yiɁ‑a ‘eat’ > yiɁ~iɁ ‘take a bite’, while the pluractional expresses the opposite, i.e. multiple instances or a higher intensity, e.g. ʕaɗɗ‑ ‘stick into’ > ʕa~ʕaɗɗ‑ ‘repeatedly stick into’.
10The following short chapters deal with adjectives (§8), numerals and quantifiers (§9), and ideophones (§10), and represent the last word class-centered chapters of the grammar. Adjectives are gender and number-agreeing modifiers. They are said not to be used as heads of phrases (p. 236), but this is contradicted by examples such as ex. 11 on p. 240. As predicates, adjectives take verbal inflectional morphology. In comparative constructions, the standard of comparison is marked by the “oblique” (≈ object form), if pronominal, and the “applicative-centripetal” (≈ dative), if nominal — which shows that Gawwada does not follow the Source Schema that is otherwise widely applied in comparative constructions in the Horn of Africa (Zelealem & Heine 2003). Ideophones constitute a word class of about 70 morphologically invariant lexemes, which are syntactically integrated with the light verbs pay‑ ‘say’ and pay‑as‑ ‘make say’. Tosco considers the ideophone to be the syntactic object of the light verb, which I find little convincing. Ideophones are said to be used on their own as predicates, but the only illustration for this use is a riddle, a genre known for its brevity and ellipses (see §3.1 in Treis Forthcoming for the exceptional use of bare ideophones in Kambaata riddles). The grammar does not make any mention of interjections; however, one animal-directed interjection might be hidden in the ideophone list (p. 253): keet ‘an order given to an animal in order to make it go straight’.
11Chapter 11 explores the syntax of the noun phrase. Whereas Gawwada is head-final in the clause, all modifiers, including relative clauses, follow the head in the noun phrase. Noun phrases that are neither subjects nor objects are marked by the associative case (see §4), the “applicative” (more on this later), or by one of a set of enclitic postpositions: adessive, terminative, instrumental and causal (§11.4). Gawwada has neither a dedicated relative verb form nor a relative marker (see (3) below). Relative clauses whose head noun is omitted are nominalized with the “pronominal head” (see §5). The description of conditional clauses, which seems misplaced in §11, shows that an element pay‑n‑i lit. ‘it will say’ is used between the otherwise morphologically unmarked conditional clause and the main clause. This ‘say’-based conjunction does not have any known parallel outside of Dullay.
12Chapter 12 on the syntax and the information structure of the simple clause first broaches the topics of word order, alignment and verbless clauses. The central section (§12.5) deals with the order of elements in the verbal complex, which is “a fixed array of elements ended by the verb and made up of different pronominal and adpositional clitics and containing all — or at least most — grammatical information” (p. 283). The section is dense, and it takes time and several rereads to understand Table 12.1 (the erroneous use of “applicative pronouns” for “associative pronouns”, three times on p. 284, does not facilitate understanding). The section gives an idea how the maximally three preverbal and maximally one postverbal enclitic elements are ordered and under which syntactic and/or pragmatic conditions they occur inside the verbal complex. However, I must admit that even after having read it several times, I am often still at a loss to understand seemingly simple examples, as Tosco tends to comment on them only minimally. No doubt it would have been helpful if the verbal enclitics of selected examples, their referents (in the clause or context) and their functions had been explained in more detail. §12 continues with a description of the conditions under which phrases (and not “clauses”, as stated on p. 290) occur post-verbally. Another detailed section (§12.7) discusses, based on many examples from natural data, how topic and focus are expressed in the absence of dedicated markers, namely through word order, the presence or absence of subject enclitics and verbal agreement, noun incorporation and the multifunctional contrast marker =kka. From an areal perspective, it is important to underline that Gawwada makes little use of clefting as a focusing device. The last sections discuss the grammar of questions.
13Chapter 13 draws together information on the expression of negation. Typologically most significant is the absence of dedicated negative particles, if we leave aside the prohibitive negator and a negative third person subject clitic. Negation is instead primarily expressed through dedicated negative paradigms. Tosco also shows that the noun Ɂol‑o ‘thing’ is being grammaticalized into an imperative negator. In addition, the chapter treats emphatic negation and the negation of nominal and adjectival predicates as well as existential clauses.
14Finally, Chapter 14 is a revised version of Tosco (2008) on clause linkage and first introduces the phrase and clause coordinator =pa ‘and’ (see (2)‑(3) below). In §14.3, the reader learns that Gawwada, unlike many other Cushitic languages (Treis & Vanhove Forthcoming), does not have converbs, i.e. dependent non-final verb forms. Instead, the language uses a sequential verb form in a non-first clause, i.e. a clause following the main clause. The aspect, tense and mood of the main verb has scope over the sequential verb. A bare sequential clause expresses “a logical consequence” (p. 331) of the event in the main clause, or possibly simply a subsequent event, as the examples show. Postpositions found on noun phrases (§11) are also used to mark the semantic relations between linked clauses, e.g. the “applicative-centrifugal” signals a simultaneity or anteriority relation (see (3) below), the “applicative-centripetal” a purpose relation and the instrumental a concomitance relation. Adverbial clauses to which these postpositions encliticize precede the main clause. The grammatical description ends with a discussion of the coordinator =ye ‘and’, which is a dedicated conjunction to link an imperative with a declarative clause (or vice versa), and the disjunctor =m ‘or’. Information on reason clauses is found in §14.4, on conditional clauses in §11.6. Complement clauses are nowhere mentioned in the grammar — is this an oversight or does Gawwada not have any? The appended texts amply illustrate direct speech reports; the description of these constructions, however, does not extend beyond a brief mention of how pronominal addressees are expressed, in the unlikely context of the space, movement and deixis chapter (§6.6).
15As has probably already become clear, the grammar under review is a tough read. Undeniably, the language presents many analytical challenges, but I believe that the difficulties in understanding are only partly attributable to Gawwada’s complexities. They also seem to be the result of nontransparent terminology, unclear wording, missing arguments, insufficient pointers to what is actually shown in the examples and, last but not least, numerous errors.
16The idiosyncratic terminology is a major obstacle. Many of Tosco’s terminological choices are, at best, unfamiliar or, worse, misleading. The term “preternumeral” for the number-neutral form of the noun (usually “transnumeral” or “general number” in the literature) is just unfamiliar, and the reader will probably find it acceptable after its choice is justified on p. 82. Other terms that I find more unfortunate are, for instance, bare “participant” for speech act participant, “declension” for downdrift, “associative” for the locative/genitive case, “jussive” for a directive verb form exclusive to the first person plural (i.e. a hortative), “oblique” for the object pronoun, “specifier” for a partially reduplicated form of positional nouns, “superlative” for constructions expressing a high degree (‘very’) and “object relative clause” for a relative clause that modifies an object noun — rather than for a relative clause in which the head noun functions as object.
- 2 Glosses not found in the Leipzig Glossing Rules: assoc associative, deict deictic, ideoph id (...)
17To me, the most confusing grammatical label is “applicative” for a clitic =n that is not used on its own and is mainly found in the following two combinations: =n‑a (applicative-centrifugal) and =n‑u (applicative-centripetal). Again, the terminological choice is not justified by the author, and given the bewildering functional range, I am unable to reconcile Tosco’s use of the term “applicative” with the typological literature (e.g. Polinsky 2013; Zúñiga & Creissels Forthcoming). The elements =n‑a and =n‑u are either NP enclitics, as in (1)‑(2), proclitics in the verbal complex, as in the second occurrence in (3) or, in clause-linking function, enclitics to the verb, see the first occurrence in (3).2
- 3 My corrections and additions are marked by […] in the gloss line and by [ …] in the f (...)
| (1) |
Ɂan~anu |
piy‑e=s‑í=n‑a |
ʕak‑í |
|
idp.[1sg]3~neg |
land‑f=dem‑indv |
exist‑ipfv.neg.1sg |
|
‘I do not live in this country’ (p. 110) |
| (2) |
torr‑í=sa |
haarr‑atte=pa=n[‑]a |
|
news‑f\deict=dem |
donkey‑[assoc.f]=link=appl‑out |
|
Ɂoraap‑att‑atte |
mala |
|
hyena‑sing‑assoc.f |
how |
|
‘How [> what] is that story about the Donkey and the Hyena?’ (p. 146) |
| (3) |
Ɂi=hapap |
pay‑ay=n‑a |
ʕanɗ‑e=s‑í |
|
indv=ideoph |
say‑ipfv.3f=appl‑out |
water‑pl=dem‑indv |
|
ʕuk‑ti |
karʕ‑itt‑o |
|
|
drink‑pfv.3f |
stomach‑sing‑m |
|
|
Ɂi=n‑a=hoq~q‑i=pa |
ɓooy‑i=pa |
|
indv=appl‑out=be_full~punct‑pfv.3m[=link] |
explode‑pfv.3m=link |
|
‘He [> it] said hapap (flapped its wings), but the water she [> it] had drunk and that filled its stomach made it explode, and …’ (p. 156) |
18In order to better understand what characterizes the Gawwada applicative, one would first need information about the differences between the base construction and the applicative construction, most importantly, about how the applied phrase is encoded in the base construction (if it can be expressed at all). Some elaboration on how the Gawwada applicative does or does not map onto comparative concepts of the applicative that are proposed in the literature would have been very welcome.
19Among other interlinearization errors, the person of pronouns and indexes are often wrongly glossed, e.g. sbj.2 for sbj1 (ex. 3, p. 276), obl.1sg.f for obl.2sg.f (Table 12.1), pfv.3m for pfv.1sg (ex. 38, p. 287). It is disconcerting to see the following inconsistent grammatical glosses, some of which reflect conventions from Tosco’s earlier publications that were not all replaced during revisions: imp ~ gen (impersonal ~ general), punct ~ sem (punctual ~ semelfactive), ade ~ diff ~ sit (adessive ~ diffusive ~ situative), pluract ~ iter (pluractional ~ iterative), appl ~ mov (applicative ~ movement), indv ~ spec (individualizer ~ specific), sg ~ s (singular), pl ~ p (plural), ipfv ~ impf (imperfective), pfv ~ pf (perfective) and ipv ~ impv ~ imp (imperative) — the first gloss of each pair or triplet is what is usually used and registered in the list of abbreviations (p. 16‑18); the second or third is a variant gloss that is sometimes used and is absent from the abbreviations list. Wording errors of the type “words end in [> begin with] one and only one consonant” (p. 77), “The Adessive may also be followed [> may follow]” (p. 264), “the Pronominal head acts as subject [> predicate]” (p. 282) are not rare. The numerous typos such as “analiyzed” (p. 208) and “instanxces” (p. 228) could have easily been detected by a spell checker. A reader cannot help but ask: Has an earlier unproofed manuscript possibly been printed by mistake?
20Tosco occasionally compares findings in Gawwada with those in related languages and then too hastily extrapolates to East Cushitic, when a phenomenon is actually only attested in Lowland East Cushitic. To the best of my knowledge, languages of the Highland branch have neither a dedicated linker between imperative and declarative clauses (p. 325), nor an l‑based root for ‘have’ (p. 115), nor, most importantly, the typical verbal complex in which pronominal elements and adpositional clitics are concentrated before the inflected verb (p. 283).
21The author is to be lauded for the database of natural speech on which this work is based, for highlighting typological and genetic rarities in the grammar of Gawwada and for providing a comprehensive account covering all grammatical domains from phonology to information structure, which does not find many parallels in Cushitic grammaticography. However, I strongly believe that peer review and proofreading would have helped to improve the quality and clarity of the description, strengthen the arguments and make it more accessible both to a Cushitist and to a typological readership.