Printer Friendly

From inflection to derivation: outcomes of Early Romani genitive in Piedmontese Sinti.

1. Introducing Piedmontese Sinti

The following pages will concern a group of derivatives in two Romani varieties, France and Piedmont Piedmontese Sinti. Romani is a group of Indo-Aryan varieties, spoken nowadays by many communities in Europe, Asia, America and Australia. With the word Sinti, we refer to a sub-group usually related to the so-called Northwestern branch of Romani varieties. Sinti varieties share strong German influence and a number of innovations, and emerged, in all likelihood, in German-speaking territory. Some Sinti communities migrated to other regions, as Northern Italy, Netherlands, Austria, Hungary, Bohemia, Slovakia, Russia, and Yugoslavia (see Matras 2002: 9). In this framework, what we call Piedmontese Sinti has its place. Piedmontese Sinti is in fact a label identifying two akin Sinti varieties that the nonvoluminous literature on the subject often considers as the same dialect (see Franzese 2002a: III, Calvet & Formoso 1987: 4), on the basis of an high proximity of the dialects and of the fact that Piedmontese Sinti is the name that the members of both communities use to designate their people and their language. Piedmont Piedmontese Sinti (from now on PPS), in decay, is spoken in Piedmont, in the periphery of big cities (Torino, Cuneo, Asti, Alessandria) and in some other little towns (see Franzese 2002c: 10). France Piedmontese Sinti (FPS), still used for everyday life, is spoken on the other side of the Alps, by a community settled in Southern France, in the area near Grasse (Region Provence-Alpes-Cote d'Azur). (1)

This work has mainly two sources: in the case of PPS, data come from the grammar and the dictionary composed by Sergio Franzese (2002a, 2002c), and, as far as FPS concerns, I used the lexicon and transcriptions collected in 1987 by George Calvet and Bernard Formoso (Calvet & Formoso 1987).

As said before, the few previous mentions of FPS and PPS point at them as the same variety. An initial but closer overview allows the identification of some differences, concerning phonological and morpho-phonological developments (2) and prosodie structure of the word. (3) Besides, the two varieties show some difference in the lexicon they borrowed from French and Piedmontese, which does not coincide. These differences raise the question about the terms of the relationship between the two varieties, although their high proximity has to be taken into account.

2. Genitive derivation in Piedmontese Sinti

In PPS and FPS we found a group of derivatives originating from Early Romani genitive in -ker-:

(1) FPS

tud'milk' [right arrow] tudeskero 'milkman', pl. tudeskere

manro, 'bread' [right arrow] manreskero, 'baker', pl. manreskere

jag, 'fire' [right arrow] jageskero 'lighter', pi.jageskere

lovo, 'coin' [right arrow] loveskero 'coin purse', pl. loveskere

nasalo 'ill/ill person' [right arrow] nasaleskero 'hospital', pl. nasaleskere

(2) PPS

eib 'language' or 'tongue' [right arrow] cibjakero 'lawyer', pl. cibjengere

moro 'bread' [right arrow] mareskero 'baker', pl. marengere

jag 'fire' [right arrow] jagakero 'matchstick', pi.jagengere

nasiben 'running' [right arrow] nasibaskeri 'car', pl. nasibangere

nasale 'ill people' [right arrow] nasalengero 'doctor', pl. nasalengere

The semantics of the derivational process leading to these derivatives consists in nominalizing a general relation between the referent and the source noun, that may be expressed by 'the one of x', where 'x' is the source noun. For example, tudeskero 'milkman' is 'the one of milk'.

This kind of lexical creation is common in many northwestern European Romani varieties (Matras 2002: 77). In both PPS and FPS it seems more productive for professional names, but we find many instances in which simply the general semantic relation previously assumed for the derivational process is expressed, see for instance PPS nasibaskeri 'car' from nasiben 'running', FPS nasaleskero 'hospital' from nasalo 'ill, ill person', FPS poreskero 'duvet' from por 'feather'.

These derivatives are the outcome of a change that occurred to ancient genitives and led an inflectional morpheme marking genitive case to acquire derivational meaning. Moreover, noteworthy is the fact that both FPS and PPS share this derivational process, but the outcome in terms of derivational paradigm is different.

3. Early Romani nominal inflection

Early Romani is the label used for that stage of Romani, reconstructed on the basis of structures largely spread in the attested varieties, which was presumably spoken by a somewhat uniform community in Greek-speaking territory, in a period that preceded the diaspora and dispersal of the community throughout Europe that led to nowadays dialectal differences (dated tentatively to the Byzantine period, from the thirteenth or fourteenth century onwards). (4) At that stage, Romani, as well as New Indo-Aryan languages, had innovated the inherited Indo-Aryan nominal system, and developed a rich nominal inflectional paradigm, that continues in many present day varieties. As reconstructed, at the Early Romani stage, noun inflected by gender, number, case and thematic status. (5) The system of nominal cases was composed, and still is in many varieties, by three layers. The lexical morpheme of the inflected noun was followed by an inflectional morpheme, belonging to the set of the so-called Layer I markers, that assigned the noun to a declensional class and specified the opposition between nominative and oblique case. (6) The Layer I marker distinguished also gender, number, and thematic status. It was followed by a Layer II morpheme, that was part of a closed set of agglutinative morphemes, signalizing the case and following the Layer I markers. Except genitive, these endings originated from postpositions and their status was a discussion topic until some years ago. (7) Some cases might be preceded by prepositions, labelled as Layer III, that form an open set, both of Indo-Aryan and borrowed origin; since our study focuses on a change that occurred to genitive, we will give attention mainly to Layer I and H

Let us consider the following examples, datives of chavo 'son' and chaj 'daughter' as they could have occurred in Early Romani:

(3) a. chav-es-ke


'to the son'

b. chav-en-ge


'to the sons'

c. cha-ja-ke


'to the daughter'

d. cha-jen-ge

daughter-OBL PL-DAT

In these examples, we have a lexical base followed by an inflective morpheme (Layer I, -es-, -a- or -(j)en-) marking nominative vs. oblique opposition, thematic status, gender and number, and it is followed by a second agglutinative morpheme (Layer II, -ke-, with the allomorph -ge- occurring after -n-) that marks the case.

4. Early Romani genitives

In nominal case system, genitive has a particular place, due to its different origin. Let us consider the following examples (in the varieties we will deal with, genitive Layer II markers are -ker-/-kr-/-k-), that show different genitives of raklo 'boy' and rakli 'girl'. The data are offered by Matras 2002, and they exemplify forms attested in Vlax dialects: (8)

(4) a. le rakl-es-k-i dej

ART.M.OBL boy.OBL-GEN-F.NOM mother

'the boy's mother'

b. la rakl-ja-k-i dej

ART.F.OBL girl. OBL-GEN-F NOM mother

'the girl's mother'

c. le rakl-es-k-e phrala

ART.M.OBL boy.OBL-GEN-PL brothers

'the boy's brothers'

d. le rakl-es-k-o dad

ART.M.OBL boy. OBL-GEN-M NOM father

'the boy's father'

e. le rakl-es-k-e dade(s)-sa


'with the boy's father' (Matras 2002: 90)

As examples show, genitive in Romani involve three morphemes: Layer I, marking the oblique case, inherent gender and number of the inflected noun, Layer II marking genitive case, and a third morpheme marking obligatory agreement in gender, number and case with its head.

This structure gives the genitive a morpho-syntactic status between noun and adjective. On the one hand, a noun inflected in the genitive form, due to its status of noun, is a prototypical controller (see Corbett 2006: 35). The genitives in (4), in fact, show inherent gender and number and they control the case of the article. On the other hand, genitives show obligatory agreement with their heads, and hence they have an adjectival behaviour (since adjectives are the prototypical target of agreement, see Corbett 2006: 40). Let us consider, for instance, (4)a, le rakleski dej, 'the boy's mother'. The head noun of the phrase is dej, that is a feminine noun inflected in nominative case. The article is in the masculine singular oblique form, le, and so it is not controlled by the head daj, but by the genitive rakl-es-k-i, that shows a masculine singular oblique Layer I morpheme (-es-). Besides the genitive rakl-es-k-i shows the marker--i', that agrees in gender, number and case (feminine, singular, nominative), with the head noun dej. So, as a noun, the genitive in the example controls the article, but, at the same time, as an adjective, it shows obligatory agreement with its head.

Romani genitive is also considered as an instance of sufflxaufnahme or double case marking: as you see in the examples, Vlax Romani shows a different agreement marker whether the head noun is nominative, e.g. (4)d le raklesko dad 'the boy's father', or oblique, e.g. (4)e le rakleske dadessa 'with the boy's father'.

5. Genitive in Piedmontese Sinti

As previously mentioned, the Early Romani nominal inflectional system continues in many varieties. Nonetheless, this is not the case for Piedmontese Sinti that has moved from a synthetic to an analytic nominal case marking. In fact, in both FPS and PPS, as well as in other Sinti dialects, noun inflects only for number and gender, and Layer III markers (namely prepositions) express entirely the syntactic function of case.

The FPS version of (3) and (4) follows below:

(5) ko cavo

of.ART.M.S boy

'to the son'

ka le cave

of ART.PL boy.PL

'to the sons'

(6) i daj do raklo

ART.F.S mother of.ART.M.S boy.M.S

'the boy's mother'

i daj di rakli

ART.F.S mother of.ART.F.S girl.F.S

'the girl's mother'

le phrala do raklo

ART.PL brother.PL of.ART.M.S boy.M.S

'the boy's brothers'

o ba do raklo

ART.M.S father of.ART.M.S boy.M.S

'the boy's dad'

kun o ba do raklo

with ART.M.S father of.ART.M.S boy.M.S

'with the boy's father'

As the examples show, Layer II markers are completely substituted by Layer III prepositions, and only nominative Early Romani Layer I morphemes are retained, assigning the noun to a declensional class and marking gender and number. In the Table 1, elaborated on the tables proposed by Elsik (2000) and adapted by Matras (2002: 83), the grey cells show synthetically which Early Romani Layer I markers have continued in PPS and FPS nouns (the line in bold is a Piedmontese Sinti innovation):

The only remainders of a passed nominal inflectional system are the group of genitive derivatives here under discussion and, for the case of PPS, some toponyms, as Milcmcite 'Milan', Ttirinate 'Turin', that are fixed unproductive forms coming from old locatives in -te (Franzese 2002a: 5).

6. Paradigm changes in FPS

As previously mentioned, the outcome of genitives in Piedmont Sinti is interesting for the diachronic process they may have undergone, that is an upgrading of an inflectional morpheme to a derivational one, and for the different paradigm reorganisation occurred in the two varieties.

At first, let us consider FPS. First of all, not every inflected forms of genitive are attested in FPS: the majority of derivatives show the ending -esker-, namely singular masculine Layer I marker -es- and genitive Layer II marker -ker-. (9) Note for example the derivative from the masculine noun tad 'milk':

(7) tad-es-ker-o

milk-OBL.M. S-GEN-M. S


This example could lead us to believe that genitive inflectional rules are always working in the same way as Early Romani, in that the morphemic analysis of the noun is totally transparent. Let us consider, instead, the following examples:

(8) boca 'wooden ball' [right arrow] boceskero 'bocce player'

jag, 'fire' [right arrow] jageskero 'lighter'

cib 'tongue' [right arrow] cibjaskero 'lawyer'

These examples, inasmuch they exceed from the rules governing genitive inflection, reveal that a change occurred, namely the desemanticiziation of Layer I morpheme, the reanalysis of morphemic structure and the reshaping of morpheme boundaries.

At first, we will discuss the derivatives boceskero 'bocce player' and jageskero 'lighter'. The former is derived from boca, a feminine noun that denote the wooden ball used to play bocce (an Italian variety of lawn bowling), and the latter from jag 'fire', feminine noun as well.

(9) boc-es-ker-o


'bocce player'




The analysis showed in (9) underline the same phenomenon: in an Early Romani perspective, both derivatives appears not well-formed, inasmuch that they present the masculine singular Layer I marker -es-, instead of feminine singular -a-, that we would expect, being the source nouns feminine (we should have, in fact, the forms bocakero and jagakero (10)). In other words, Layer I does not show any agreement with the inherent gender of the inflected noun, and this let us suppose that Layer I does not carry any gender marking anymore.

The last example from FPS is the term to indicate the lawyer, that is cibjaskero, from the feminine noun cib 'tongue'. We cannot determine whether the base is singular or plural: in the first case, the lawyer should be 'the one of the tongue', underlining the lawyers' ability of using the language to pursue their aims, while in the second case, namely lawyer as 'that of the tongues', the noun could express the lawyers' ability of using many languages (many ways to communicate). Even in this case, the derivative does not fit with the usual paradigm. In fact, if the base of the derivative was the singular, it should show -ja- Layer I marker, and then we should find *cibjakero, (11) while if formed from a plural, it should show -jen- Layer I marker, and then we should have the non-attested form *cibjengero. Meaningfully in this case, unlike other examples, we are not able to deal with the derivative by means of the genitive inflectional rules: if we strictly adopt the preceding segmentation criteria in analysing this derivative, we face with the impossibility to account for it.

(10) cib-jas-ker-o / cib-ja-s-ker-o

tongue- 7-GEN-M.S / tongue-OBL.F.S-?-GEN-M,S


With the first segmentation, cib-jas-ker-o, we would deal with a Layer I morpheme, -jas-, that is not attested among described Romani varieties; while if we consider -ja- as the regular feminine singular Layer I marker, we cannot explain the following segment -s- In both cases, the noun represents an exception to the regular nominal paradigm. To account for this derivative, there seems to be two possibilities: we may assume that FPS presented a regular derivative *cibjakero 'lawyer', built on a singular base, that, with the semantic bleaching of Layer I, could have undergone synchronic contamination with the other forms in -(e)skero (as previously mentioned, these forms are the majority in our source). This change would prove not only the desemanticization of Layer I, but an occurring bleaching of the morpheme boundary between Layer I and Layer II. A second way to account for the derivative is to assume it to be formed by the plural nominative base cibja, and the morphemes -(e)sker-o. This would lead, even more strongly, to a similar conclusion: the Layer I does not carry any case feature, since nominative base does not impose the selection of nominative Layer I morpheme, and there is no boundary between Layer I and II, but the same block is perceived as a single derivational morpheme. It is interesting to notice that in PPS we find cibjakero, 'lawyer', with a fully regular genitive inflection.

As far as the plural of the derivatives concerns, they show the same -esker- morpheme plus -e plural marker for nouns, es. tudeskere 'milkmen', boceskere 'bocce players', jageskere 'lighters', cibjaskere 'lawyers'. This suggests that the only gender/number information in our derivatives is given by the last marker, that passed from signalising the agreement with the head noun to marking the inherent gender and number of the derivative.

As these data show, genitives in FPS underwent a change. Layer I has gradually lost his features of marking gender and number of the source noun, and, through morphological levelling, the less marked masculine Layer I morpheme substituted the others (boceskero, jageskero). At the same time, the semantic change has regarded the case too, leading to the consequent bleaching of morpheme boundary between Layer I and II (cibjaskero). The obsolescence of case system accompanied by the permanence in the language of terms that once were clearly perceived as genitives may have led to a reanalysis of the sequence of Layer I and II in a single derivational morpheme: the morphemic sequence -es-ker-o 'Layerl-Layerll-gender and number of the head' is gradually reanalysed in -(e)sker-o 'derivational morpheme-inherent gender and number of the derivative'. We have thus no more a class of genitives, but a new productive derivational pattern, that can be expressed by the formalism [[[[X].sub.N].-(e)sker-].sub.N] 'the one of x'.

7. Genitive derivatives in PPS: a change in progress

The PPS derivatives mentioned in the previous paragraph should warn us about the noncoincidence between the two varieties.

In fact, a closer analysis of PPS data reveals a different and composite panorama. As previously mentioned, no nominal case system continues in both Piedmontese Sinti varieties and Layer I does not carry any case marking. Nonetheless, we do not find in PPS the same morphological levelling occurred in FPS with the spread of -es- morpheme, and the derivatives do inflect in agreement with the inherent gender and number of the source noun.

As the following examples show, we have derivatives with feminine base and feminine singular Layer I morpheme -a-/-ja- (11), derivatives with masculine base and masculine singular Layer I marker -es- (12) and derivatives with plural base and plural Layer I marker -en- (13):12

(11) jag' fire'[right arrow] jag-a-ker-o



cib 'tongue' [right arrow] cib-ja-ker-o

tongue-F.S-GEN-M. S


(12) moro 'bread' [right arrow] mar-es-ker-o



kast 'wood' [right arrow] kast-es-ker-o

wood-M.S-GEN-M. S


(13) nasale 'ill people' [right arrow] nasal-en-ger-o

ill-PL-GEN-M. S


Nevertheless, in PPS we trace instances of a tendency to morphological levelling towards the spread of--es- marker, although the process appears much weaker than in FPS. For example, the derivative kakavieskero 'smith', from the feminine noun kakavi 'pot', shows the same--(j)es-masculine Layer I marker and does not agree with inherent gender of the base.

(14) kakav-jes-ker-o

pot-M. S-GEN-MS


Noteworthy is what happens to plural forms of genitive derivatives in PPS. In fact, unlike FPS, in Piedmont variety plural Layer I marker is generalised for every plural form of the derivative, as the following examples show:
(15) jag-a-ker-o,         pl. jag-en-ger-e
     fire-F.S-GEN-M.S     fire-PL-GEN-PL

     cib-ja-ker-o,        pl. cib-jen-ger-e
     tongue-F.S-GEN-M.S   tongue-PL-GEN-PL

     mar-es-ker-o         pl. mar-en-ger-e
     bread-M.S-GEN-M.S    bread-PL-GEN-PL

     kast-es-ker-o        pl. kast-en-ger-e
     wood-M.S-GEN-M. S    wood-PL-GEN-PL

     nasal-en-ger-o,      pl. nasal-en-ger-e
     ill-M.S-GEN-M. S     ill-PL-GEN-PL

     kakav-ies-ker-o      pl. kakav-ien-ger-e
     pot-M.S-GEN-MS       pot-PL-GEN-PL

In these cases, gender and number of the source noun is a relevant morphological feature for the word formation process only in the singular forms, whereas the same features are not relevant in the plural forms. There, a morphological levelling, that is the selection of plural Layer I marker for every plural form, occurred. Since, unlike the case of singular forms, we cannot isolate the morpheme marking gender and number from the derivational one, we may suppose that the new morpheme -enger- has a status of portmanteau, carrying two kinds of information (number and derivation) in a single morpheme (namely, we cannot identify which morpheme does carry plural information and which is the derivational morpheme, or better, we may surely affirm that the controller that selects the number of the inflection is not the source noun anymore). Moreover, the extension of irregular masculine Layer I marker for those terms which are built on a feminine head allows us suppose that we are dealing with an ongoing change, in which the morphological levelling is spreading also to singular forms, where the semantics of Layer I is bleaching and may not carry gender information. An exceeding case like kakavieskero 'smith', pi. kakaviengere leads us to suppose that a new paradigm is spreading, in which we have a rule giving a derivative in -eskero for singular and -engere for plural.

Summing up, Table 2 and 3 show an overview of the paradigm changes occurred in PPS and FPS. Table 2 represents the Early Romani genitive construction: it shows the inflectional paradigm of the noun inflected in the genitive case (and then the complete set of Layer I + Layer II + agreement markers) accompanied by the head noun. Table 3 summarizes the outcome of this construction in both FPS and PPS. On the left, we find the FPS derivational paradigm. The sequence of two Early Romani morphemes (Layer I, Layer II), carrying case (Layer I and II), number and gender (only Layer II), has been reanalyzed and, through the extension of the less marked morpheme -es-, it is now a single derivational morpheme. The Early Romani agreement marker is now the exponent of gender and number of the derivative (thus it is not anymore an agreement marker). This new derivative, in a naturalness perspective, has a high degree of diagrammaticity (13): the new word is easily segmentable, the derivational morpheme carries only the derivational meaning, gender and number are marked only by the last inflectional morpheme; the meaning can be easily built by the sum of the three morphemes (it is quite transparent from a morpho-semantic point of view). On the right, PPS shows instead two different stages: a conservative paradigm that is closer to the Early Romani genitive inflection, and an innovative paradigm that is closer to FPS. We may assume an analogue reanalysis that has led the bi-morphemic sequence of Layer I and II to become derivational and the agreement marker to signal the gender and number of the derivative. The conservative paradigm shows a split between the singular and the plural forms of the derivative: like Early Romani, in the singular forms PPS did not lose the possibility to mark gender and number of the base in the derivational morpheme (as Layer I and II did), while we have a single derivational marker in the plural. The innovative paradigm is the result of a change towards an increased diagrammaticity. The gender of the source noun, being not relevant in the word formation process, is no more marked in the singular and we have thus a reduction of the set of singular derivational morphemes (only esker-continues, like FPS). Nonetheless, a split between singular and plural forms of the derivative remains (-esker- for singular and -enger- for plural) and therefore diagrammaticity is not complete: because of this split, derivational morphemes appear to be portmanteau, namely they do not mark only derivational meaning, but also number. The expression of the latter relies on both the inflectional and the derivational morpheme, and this redundancy results in the opaqueness of the boundary between the two morphemes and thus in the overall morpho-semantic transparency of the derivatives.

8. Genitive derivatives and degrammaticalization

The changes occurred both in PPS and FPS could add new material to the considerations on degrammaticalization, and, in particular, to those phenomena that Norde, in her volume on degrammaticalization (2009), has labelled as deinflectionalization.

According to Norde, "degrammaticalization is a composite change whereby a gram [i.e. grammatical morpheme] in a specific context gains in autonomy or substance on more than one linguistic level (semantics, morphology, syntax, or phonology)" (Norde 2009:120). In other terms, it is an upgrading, a movement in reverse, on the syntactic, morphophonological and semantic hierarchies used to describe the more frequent grammaticalization processes. Norde defines deinflectionalization as a "composite change whereby an inflectional affix in a specific linguistic context gains a new function, while shifting to a less bound morpheme type" (Norde2009: 152), involving in this definition changes from inflectional affix to clitic and to derivational affix. Leaving aside the controversial discussion whether or not is possible to determine a boundedness scale for morphemes, we will focus on that particular type of deinflectionalization consisting in the development of a derivational element from an inflectional one. This phenomenon could be considered as the reversal of the change derivational > inflectional affix that was already observed by Kurilowicz (1975). In the opinion of Kurylowicz, grammaticalization can be defined as

the increase of the range of a morpheme, advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status, e.g. from a derivative formant to an inflectional one. Numerous instances may be found for the evolution collective (derivative) > plural (inflectional), cf. the fate of the Slavic collective suffix -ja, -je (Russian plural druz-ja < drug "friend," Polish plural liscie < list 'leaf), the Persian plural in -ha, an old collective, the so-called broken plurals of Southern Semitic, originally collectives (Arabic plural kutub < kitab "book" etc.) (Kurilowicz 1975: 69). (14)

In Norde's volume, an instance of deinflectionalization involving a change of the type inflectional affix > derivational affix is the outcome of the Swedish old nominative -er, a nominal and adjectival suffix. With the obsolescence of nominal case, the suffix continues only for adjectives and for nominalised adjectives (es. en blinder 'a blind person'), a context that allows the development of derogatory meaning and the following reanalysis of the inflectional morpheme as a derivational suffix with derogatory meaning (e.g. slarv 'mess' [right arrow] en slarver 'a messy one') (see Norde 2009: 179-181).

Even though we do not have documents of intermediate stages of PPS and FPS, and hence our reconstruction has the status of hypothesis, nonetheless the presence of some elements leads us to consider that this could be the case for PPS and FPS.

As previously mentioned, genitive in Romani has a morpho-syntactic status between noun and adjective. The ambivalence of the morphological structure is reflected by the functional analysis. From this point of view, Koptjevskaja-Tamm (2000) analyses genitive in Romani, in particular nominais with genitive markers, which function as adnominal dependents (attributes, modifiers to other nouns) within larger NPs (see Koptjvskaja-Tamm 2000: 124). According to the author, these genitive adpositions can be used in Romani with a double function. In fact, we find 'anchoring' adpositions, in which the genitive aims to the identification of the extra-linguistic referent pointed at by the head, and 'non-anchoring' adpositions, in which genitive classifies or qualifies the noun and does not show referential function. For example, in Norwegian Lovari o parko le kraiesko 'the park of the king' (literally 'the park the king.GEN. AGR'), the genitive is used for the identification of the park we are dealing with in the extra-linguistic reality; in forosko gras 'town horse' (literally 'town.GEN.AGR horse') the genitive is used to classify the horse denoted by the head noun, namely to indicate the type of horse which the denoted horse belongs to.

Thus, from a semantic and functional point of view, the behavior of anchoring genitives is closer to prototypical noun, in that they are referential, while non-anchoring genitives behave like prototypical adjectives. In Koptjevskaja Tamm's words,

both [i.e. anchoring and non-anchoring genitive adposition] characterize entities via their relations to other entities,--but they differ as to whether the relation involves the referents of the head and the adnominal in a NP or not. Anchoring GAs [i.e. genitive adpositions] are closer to prototypical usages of nouns (and noun phrases) in that both involve reference. The non-referential characteristics of non-anchoring adnominals, on the contrary, make them potentially 'weaker' candidates for being treated as nouns and noun phrases (Koptjevskaja Tamm 2000: 145).

According to the author, non-anchoring genitives show a tendency to lose some nominal morpho-syntactic properties (15) and, above all, they "provide a powerful tool for qualifying entities by focusing on various aspects--material they are made up of, age, size, purpose, temporal and locational characteristics and so on" (Koptjevskaja Tamm 2000: 141), that is a word formation strategy well acknowledged in many Romani varieties.

This productivity of adjectival genitives may be the ground on which the changes that led to FPS and PPS genitive derivatives occurred. As we already mentioned, both PPS and FPS do not continue the Early Romani nominal case system and show the analytic expression of case. In a context of obsolescence of the case system, there has presumably been a moment in which the speakers used nominal cases and analytic constructions alternatively. At this stage, adjectival genitives, as a result of their different use, probably resisted more than other cases, and continued even when nominal case system was completely substituted by analytic constructions with Layer III.

The adjectives, that continued the genitive "relational" meaning, may be subsequently nominalized or formed by ellipsis, for example o gadzo tudeskero 'the man of the milk' > o tudeskero 'that of the milk' > 'the milkman'. (16) With the bleaching of synthetic genitive marking, these nominalizations have been reanalyzed and Layer II had to be considered as a derivational morpheme. Being Layer I weakened by obsolescence and being the inherent gender of the noun now marked by the final ex-agreement marker -o/-i, marking gender at the Layer I level became irrelevant, and thus we had morphological levelling (that interestingly has two different outcomes in the two varieties).

Even though evidence lies in reconstruction and it has to be handled with care, these derivatives may underline the role of obsolescence in degrammaticalization process, as Willis has recently suggested: "obsolescence acts as a catalyst in some cases. Where a particular morphosyntactic sub-system (e.g. case, person-number inflection etc.) is being lost, surviving instances of that sub-system are susceptible to reanalysis as members of some other category" (Willis, forthcoming).

At last, we will mention another aspect that may have some interest in the change here discussed. As said before, degrammaticalization consists in the movement of an element backwards on the grammaticalization hierarchies. Norde (2009) describes degrammaticalization by using backwards Lehmann's parameters, which are the criteria proposed by the latter in order to account for grammaticalization. Lehmann's parameters rise from the assumption that grammaticalization consists in the decrease of the autonomy of the sign, affecting specifically its weight, cohesion and variability. Weight is the property "which renders [the sign] distinct from the members of its class and endows it with prominence in the syntagm" (Lehmann 1995:122); variability is the "momentary mobility or shiftability [of a sign] with respect to other signs" (Lehmann 1995: 122); cohesion is the property which is responsible for the "contraction of the relations of the sign with the other signs" (Lehmann 1995: 122). Weight, variability and cohesion always have to be referred to paradigmatic and syntagmatic dimensions of the language, and thus, on the basis of these two levels, they "separate into two clearly distinct sets of criteria" (Lehmann 1995: 123). When a sign grammaticalizes, changing from a lexical to a grammatical or from a grammatical to a more grammatical status, it loses autonomy, its weight and variability decrease, while cohesion raises; when it degrammaticalizes, the opposite occurs.

Deinflectionalization, in particular, involves a change on the paradigmatic dimension of the sign. Developing derivational meaning from inflectional, a deinflectionalised item is expected to gain semantic substance (resemanticization), i.e. it undergoes a gain in the paradigmatic weight, or integrity, that is "its possession of a certain substance which allows it to maintain its identity, its distinctness from other signs" (Lehmann 1995: 126). As far as paradigmatic variability concerns, that is the "freedom with which the language user chooses a sign" (for example, whether a category has to be expressed or it can be omitted), the form loses obligatoriness, that is typical of inflection. Nevertheless, the crucial parameter identifying deinflectionalization is paradigmaticity (i.e. paradigmatic cohesion), "because what is most characteristic of these cases is that inflectional suffixes cease to be part of an inflectional paradigm (deparadigmaticization). Thus they develop into a less bound type of morpheme (severance), and they gain a new function or new meaning (resemanticization)" (Norde 2009: 231).

This parameters thus may help us to understand what is happening or has already happened in Piedmontese Sinti varieties. In changes like tudeskero 'of the milk' > tudeskero 'milkman' in fact we see a resemanticization: the "relational" meaning of genitive remains, but it has moved from syntactic to lexical ground. We have a higher paradigmatic variability, and so a deobligatoriflcation, in that the derivational process occurring in both FPS and PPS is a possible path to produce complex words, but definitely not as obligatory as an inflectional process.

Thus paradigmaticity, and specifically the deparadigmaticization of the elements involved in the change, can be considered the most relevant parameter for the identification of deinflectionalization and here lies the difference between FPS and PPS. In fact, while with FPS we can surely say that--esker- is not part anymore of an inflectional paradigm, in PPS the morphemes--esker-, -aker-, -enger- show, beside the new derivational value, some continuity of the old inflectional function with a stronger paradigmatic cohesion.


Calvet, Georges & Formoso, Bernard. 1987. Lexique tsigane: Dialecte sinto piemontais: Un dialecte tsigane parle dans le sud de la France. Paris: Publications Orientalistes de France.

Corbett, Greville G. 2006. Agreement. Cambrigde: Cambridge University Press.

Dressier, Wolfgang U. 1985. Morphonology: the dynamics of derivation. Ann Arbor: Karoma.

Elsik, Viktor. 2000. Romani nominal paradigms: Their structure, diversity, and development. In Elsik, V. & Matras, Y. (eds.), Grammatical relations in Romani: The noun phrase, 9-30. Amsterdam: John Benjamins.

Elsik, Viktor & Matras, Yaron. (eds.). 2000. Grammatical relations in Romani: The noun phrase. Amsterdam: John Benjamins.

Franzese, Sergio. 2002a. Grammatica di Sinto Piemontese (lingua romani [zingara]) (con collegamenti ipertestuali e supporto audio su CDROM). Edizioni " O Vurdon". ( GRAM V02 ns.pdf) (Accessed 2016-11-20)

Franzese, Sergio. 2002b. I Sinti Piemontesi. Origini--Stanziamenti--Tradizioni--La guerra--La lingua romani--Bibliografia e materia li. Edizioni "O Vurdon". ( SINTI PIEMONTAKERI ns.pdf) (Accessed 2016-11-20)

Franzese, Sergio. 2002c. Dizionario Italiano--Sinto Piemontese--Inglese--Francese (e registri inversi) (su CDROM). Edizioni "O Vurdon". (Available online version: http://progettoniglo. org/ita_sint .php and (Accessed 2016-11-20)

Friedman, Victor A. 1991. Case in Romani: old grammar in new affixes. Journal of the Gypsy Lore Society, Fifth Series 1 (2). 85-102.

Heine, Bernd & Claudi, Ulrike & Hunnemeyer, Friederike. 1991. Grammaticalization: a conceptual framework. Chicago: University of Chicago Press.

Hock, Hans Henrich. 1991. Principles of Historical Linguistics. Second revised and updated edition. New York: Mouton de Gruyter.

Koptjevskaja-Tamm, Maria. 2000. Romani genitives in cross-linguistic perspective. In Elsik. V. &

Matras, Y. (eds.), Grammatical relations in Romani: The noun phrase, 123-149. Amsterdam: John Benjamins.

Kurylowicz, Jerzy. 1975 [1965], Esquisses Linguistiques. Munich: Fink.

Lehmann, Christian. 1995. Thoughts on grammaticalization. Munchen--Newcastle: Lincom Europa.

Matras, Yaron. 1999. S/h alternation in Romani: An historical and functional interpretation. Grazer Linguistische Studien 51. 99-129.

Matras, Yaron. 2002. Romani. A linguistic introduction. Cambridge: Cambridge University Press.

Norde, Muriel. 2009. Degrammaticalization. Oxford: Oxford University Press.

Willis, Davis, (forthcoming). Degrammaticalisation. In Ledgeway, Adam & Roberts, Ian (eds.), Cambridge handbook of historical syntax. Cambridge: Cambridge University Press.

Giulia Meli, Universita degli Studi di Milano

Giulia Meli

Universita degli Stadi di Milano


(1) To be more specific, Calvet and Formoso, the authors of our source, had different informants. The data supplied by Calvet are less numerous (550 vs. 1500 lemmas) and they were acquired in 1956 by a 55-year-old non-active speaker, settled near Lyon since many years, married with a not-sinti woman, and fully integrated in French society. Formoso interviewed instead four informants, aged from 79 to 34 and part of the same family, settled near Grasse since 60s'. The informants, interviewed in 1982-1983, affirm to use Piedmontese Sinti for the endo-communitarian communication, and Franzese 2002b witnesses the stability of this situation up to more recent years.

(2) For instance, loss of aspiration in PPS voiceless stops, e.g. PPS tau, vs. FPS thav/tav 'thread', PPS pral vs. FPS phral 'brother', PPS ker vs. FPS ker/kher 'home'; different development of the so called s/h alternation, a morpho-phonetic pan-Romani change affecting sibilants that weaken in velar fricatives, and may have also an outcome in--j- or may result, in advanced stages, in a [empty set] (see Matras 1999), e.g. PPS som/jom vs. FPS jom 'I am', PPS dzassa vs. FPS dzaja 'you go'.

(3) Mainly oxytone, and then conservative, in PPS and paroxytone in FPS, e.g. PPS berali vs. FPS berali 'bee', PPS boliben vs. FPS boliben 'sky', PPS nasalo vs. FPS nasalo 'ill'.

(4) (see Matras 2002: 20)

(5) Even though the terms thematic and athematic come from Indo-Aryan linguistics (presence or not of a thematic vowel--an empty morpheme indicating the morphological class of the word--between the lexical and the inflectional morpheme), their use in the Romani domain is different. As Yaron Matras explains, "in Romani, 'thematic status' pertains to the split in the morphological treatment of pre-European vocabulary and European loans. [...] The thematic grammatical formants are mainly of Indo-Arian stock. By contrast, subsequent loans receive so-called 'athematic' morphology, largely borrowed from Greek as well as from later contact languages" (Matras 2002: 73).

(6) Actually, oblique endings are also used to mark accusative, and thus the opposition at the basis of Layer I is not between nominative and oblique, but between nominative and non-nominative case; nonetheless the literature uses the label "oblique" and we will maintain it to avoid confusion.

(7) The discussion had one of its last episodes with an intervention by Friedman (1991).

(8) Vlax is a branch of Romani dialects probably emerged in Romanian-speaking territory. Vlax dialects share "extensive Romanian influence on vocabulary, phonology, and loan morphology, as well as a series of internal innovations" (Matras 2002: 7). Vlax varieties have spread also out of Romanian-speaking territory, due to "many migration waves of Vlax speakers from the Romanian principalities, some of them at least connected with the abolition of serfdom in Romania, which lasted until the second half of the nineteenth century" (Matras 2002: 7).

(9) In the source we have about thirty derivatives. Just two noun show -enger-, namely plural Layer I marker in derivational morpheme, and one in -aker- with feminine singular Layer I marker, the rest is -(e)sker-.

(10) This last form is found also in PPS, where we have the form jagakero 'match'. To be precise, something similar is attested in FPS, too: the word jagakeri 'match', that is the only derivative showing a Layer I feminine marker in Calvet & Formoso 1987. Interestingly, in FPS the comparison of the derivatives jageskero 'lighter' and jagakeri, 'match' may suggest the relative chronology of the change. The regular derivative, jagakeri, refers to an object that had the same function of the lighter and had to be spread presumably before the diffusion of the latter. This derivative could thus date back to a period in which Layer I still marked the gender of the source noun. The second Jageskero, was coined instead to name an object that spread subsequently and in a period in which the change had already occurred.

(11) That is the noun we find in PPS for the same meaning.

(12) Since noun does not carry any synthetic case marking, and gender and number are the relevant features to our discussion, henceforth we will not mention oblique case in the glosses.

(13) (see Dressier 1985: 323-329)

(14) The same change is briefly mentioned by Heine et al. (1991: 213).4
Table 1: Continuation of Early Romani Layer I markers in Piedmontese

Class               Symbol   Example            Nom      Pl

  zero-masculines   MO-a     drab 'medicine"    --       -a
                               dzuv 'louse"
                    MO-A     cacipen 'truth"    --       -a
                    MO-O     ker 'home"         --       --
                               vast 'hand"
o-masculines        Mo       sero 'head"        -o       -e
i-masculines        Mi       --                 -i       -P
zero-feminines      FO-U     jag 'fire"         -a
                    FO-J     suv 'needle"       -ja
                    FO-O     piring 'pot'       --       --
i-feminines         Fi       bibi 'aunt"        -i       -N
Athematic           *Mo      kulpo 'knot"       -o       -i
u-masculines        *Mu      --                 -ll(s)   -i
i-masculines        *Mi      --                 -i(s)    -P
a-feminines         *Fa      felda 'campo"      -a       -i

Class               Symbol   Obl     Pl

  zero-masculines   MO-a     -es-    -en-

                    MO-A     -as-    -en-
                    MO-O     -es-    -en-

o-masculines        Mo       -es-    -en-
i-masculines        Mi       -jes-   -jen-
zero-feminines      FO-U     -a-     -en-
                    FO-J     -ja-    -jen-
i-feminines         Fi       -ja-    -jen-
Athematic           *Mo      -os-    -en-
u-masculines        *Mu      -lis-   -en-
i-masculines        *Mi      -es?-   -en-
a-feminines         *Fa      -a-     -en-

Table 2: Paradigm of Early Romani genitive constructions

Early Romani

               Layer I   Layer II

base.M-        -(j)es-   -ker-      Gender and case
base.F-        -(j)a-                 of the head
base.PL-       -(j)en-   -ger-

Early Romani   Early Romani

               Agreement marker

                                  S    PL

base.M-        Nom M              -o
base.F-        Nom F              -i   -e   + head noun
base.PL-       Obl                -e

Table 3: Paradigms of genitive derivatives in FPS and PPS

        derivational   inflectional
        morpheme       morpheme

                       S              PL

base-   -esker-        -o (/-i)       -e


Conservative paradigm

        derivational                  inflectional
        morpheme                      morpheme

        S              PL             S              PL

base.M- -esker-        -enger-        -o (/-i)       -e
base.F- -aker-

Innovative paradigm

        derivational                  inflectional
        morpheme                      morpheme

        S              PL             S              PL

base-   -esker-        -enger-        -o/(-i)        -e
COPYRIGHT 2016 The Slovak Association for the Study of English
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Meli, Giulia
Publication:SKASE Journal of Theoretical Linguistics
Article Type:Report
Geographic Code:4EXRO
Date:Sep 1, 2016
Previous Article:Variation and variants in the Dictionary of Multiword Expressions (focussing on complex nominals/noun compounds).
Next Article:Middle English names of merchants: etymology and aspects of usage.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters