The paper explores parallelisms between Bantu (specifically Otjiherero) and Romance (through Latin and Spanish) with respect to left and right peripheries, and subject and object clitics. The analysis is formulated in Dynamic Syntax (DS, Cann et al. 2005) and centrally involves notions of structural underspecification. Through providing detailed analyses of different word order possibilities in the Bantu and Romance languages discussed, we show how DS concepts of structural growth over initially underspecified tree relations, such as the building of linked structures and unfixed nodes, provide a uniform basis for analyses of word order variation across the two language groups. We then extend our analysis to include Bantu subject/object markers, which we analyze by employing the same formal tools as used in the analysis of Romance (object) clitics, namely unfixed nodes which have to be construed within a tightly locally restricted domain. Empirical support for our analysis comes from restrictions on the presence of object markers in passive and locative inversion constructions in Otjiherero, which we show to follow from independent constraints of the availability of unfixed nodes within a given domain. The analyses of Bantu and Romance presented show that despite differences in surface morphology between the two language groups, both exhibit a striking parallel with respect to the way lexical information and general structure building principles of DS interact. The difference between Romance clitic systems and the agglutinative morphology of Bantu subject and object makers is thus seen to be comparatively superficial, while the DS analysis brings out the strong structural parallelism between the two language groups.
In this paper, Italian and Bantu clitic clusters are analyzed and compared. I claim that in both language families, pronouns check case in a low clitic position, and in so doing they reverse the order of arguments. In Italian, clitics move to a high clitic position where they check person and number features. Different types of Italian clusters are individuated: they differ syntactically, phonologically, and morphologically. As in Kayne’s (1994) proposal, clusters can be formed either on one single functional head or on adjacent heads; the former type can appear both in enclisis and proclisis, the latter only in proclisis. Clitic pronouns can end with an epenthetic vowel or a class marker/ inflectional morpheme; the former (consonantal clitics) can appear in clitic clusters in any position, the latter (morphologically complex clitics) cannot be the first element in clitic clusters dominated by one single head. Finally, some clusters are inserted as lexical units, others are two independent words: only the former display the linking vowel [e]. The intricate interplay of these (partially) independent properties explains a number of restrictions on clusters found in Italian (and other Romance languages). The hierarchy of person and number features in the high clitic position is also discussed, which explains other restrictions on clusters.
Romance pronominal clitics and Bantu object markers vary in gender and number, replace arguments, and surface to the left of the verbal root in declarative clauses. Both types of morphemes are regularly analyzed as affixes on the verb. It is argued that both have syntactic properties that justify treating them differently from lexical affixes. The argument is first made for French unstressed pronominal objects like me, le, en, drawing from their distribution in contemporary French as well as from their historical behavior. It is shown that they are syntactic objects, and that their morphophonological affix-like properties should be treated independently of their syntactic behavior. Then evidence is presented pointing to the fact that Bantu object markers are also syntactic objects of some sort. The last section asks whether one can arrive at a unified treatment of both elements. The conclusion is that by focusing on the morphophonological properties of these pronominal elements, one runs the risk of overlooking their syntactic properties.
Romance and Bantu languages show a range of contrasting morphological and syntactic properties, ranging from WH extraction strategies and the existence of V2 to directionality of affixation and degree of fusion and agglutination in verbal inflectional morphology. Using Isizulu and Chishona (Bantu) and standard French and Italian (Romance) as examples, this paper correlates these contrasts in terms of a common right-branching Split-INFL structure, and contrasting preferences for movement (STAY, in Optimality Theoretic terms). It is proposed that the same preference which motivates WH extraction strategies in the four languages also motivates the contrast between the relatively prefixal and agglutinative verbal inflectional morphology of Isizulu and Chishona and the suffixal and more fusional verbal inflectional morphology of French and Italian. The conclusions reinforce intuitions that languages belonging to the Romance and Bantu language families are typologically similar.
The paper considers several aspects of Bantu and Romance DPs, including: (i) Noun Class and grammatical gender; (ii) apparently derivational properties of the two; (iii) ordering among nouns and their modifiers; and (iv) concord in DPs. Several conclusions are argued for. Firstly, Bantu Class is a gender system like that of Romance, with gender-specific Spell-Out of number features. Secondly, despite some surface evidence to the contrary, gender/Class is an uninterpretable feature, without derivational functions. Thirdly, DPs of the two languages share a common architecture; and fourth, in both families concord is the result of the Agree relation. Thus many properties of nouns and DPs are common to both language groups, as the hypothesis of UG leads us to expect.
The paper reconsiders the evidence for the interpretable status of φ-features within DP in Romance languages, focusing on Italian. The conclusions are that, given a reasonably local interpretation of what counts as interpretable, gender is not an interpretable feature, and number (or plur, in the sense of Heycock & Zamparelli 2005) is interpretable, but is not sufficient to drive the process of verbal and predicative agreement. The notion of “default”, particularly with respect to the Person feature, is discussed and various alternatives are explored.
This paper claims that feature sharing should be analyzed as the result of at least two different processes, which are named here Agreement and Concord, and inquires how these two processes are manifested inside nominal expressions (NEs). Agreement is the transfer of the Person features of the possessor (the “subject” of the NE) onto some functional head (parallel to subject Agreement in the clause) with the effect that (Genitive) Case is assigned. On the contrary, Concord is the transfer of Number, Word Class, and Case specifications from a functional head onto a modifier, which is first-merged as a Specifier of that functional head. The claim is that, quite differently from Agreement, Concord arises from the merger of a modifier, underspecified for uninterpretable features, in the specifier of a functional head, carrying a copy of those features. In other words, Concord is directly enhanced by the Spec-Head configuration; it does not involve merger of a probe which targets a goal and, as a consequence, never triggers (overt or covert) movement. This proposal can dispense with a number of otherwise unmotivated movements and can derive the different properties of these two kinds of feature sharing phenomena. The argument is supported by observing macro-parallelisms across Bantu and Romance languages, in particular Swahili and Xhosa on the one hand and Romanian and Italian on the other hand.
In this paper, I propose a unified syntactic analysis of Luganda and Italian simple nouns. I argue that Italian and Luganda nouns are formed in the syntax via merge and move operations. More specifically, I show that in both languages all nouns are formed via the merger of the nominalizer head [n] with a nominal stem [LP] yielding the nominal structure [nP [n [LP]]] and that syntactic movement is necessary in the noun formation process of Italian nouns to derive the correct morpheme order.
In order to prove that the structure [nP [n [LP]]] is representative for both languages, I demonstrate that the nominalizer head [n] corresponds to both the Italian gender feature and the Luganda class feature and that, therefore, gender and class are the same feature. The data analysis in sections (2) and (3) of this paper supports the claim that gender and class are the same feature because of their identical inflectional and derivational functions. At the inflectional level, gender and class trigger VP and DP agreement and at the derivational level gender and class function as n-marked heads whose merger with an XP yields a noun.
Based on interface evidence, this paper shows that the interpretation of discourse categories relies on syntactic conditions. Specifically, the existence of a systematic connection between formal properties and discourse functions shows that Topics are licensed in dedicated positions in the C-domain, which correspond to specific realizations at the PF-interface. Different types of Topics thus correspond to specific tonal events and show A’-properties which discard a movement approach. Finally, the interface properties of postverbal subjects are examined, showing that their position does not depend on the argument structure of the verb and is not necessarily different from the position of (referential) preverbal subjects. Though the present analysis is mainly concerned with Italian, a comparison with Bantu data is also proposed in order to show similarities, discuss (apparent) discrepancies and highlight implications for future research.
Word order variation in Romance and Bantu has been related to information structure portrayed in the different discourse functions of the sentential elements involved. Based on the distribution of new information focus in Romance and Bantu, this paper argues that discourse functions need not be directly encoded in syntax. The position defended here is that syntax generates all possible structures which are filtered out at the interface with the phonological component. The prosodic phrasing of these structures is what indicates focused constituents occurring in positions of prominence. The paramount significance of prosody for the determination of focus is particularly illustrated in those cases both in Romance and Bantu where, for syntactic reasons, change in word order is restricted but prosodic effects still accompany focus. The proposed interface approach to focus accounts for the variation in focus strategies and the intimate relation of focus to prosody which is attested in Romance and Bantu. Crucially, the evidence from Romance and Bantu is complementary. The Romance data provide the necessary syntactic evidence for not positing a designated focus position in the syntactic hierarchy for focus, whereas the Bantu data show that prosodic effects may emerge in varying ways, providing evidence for not linking a single syntactic position to a given prosodic effect. The paper in this respect highlights the similarities and differences of the role of prosody in indicating focus in stress versus tone languages.
Both Bantu and Romance languages use a V(erb) S(ubject) construction to express thetic (“out-of-the-blue”) sentences. Two types of languages can be distinguished within these language families, with respect to the verbal agreement in a thetic VS sentence: in type 1 the verb has default agreement, whereas in type 2 the verb agrees with the postverbal subject. In the Bantu languages these two types also display a difference in the use of conjoint and disjoint verb forms. Collins (2004), Carstens (2005), and Baker (2008) have previously analyzed such agreement and word order phenomena. These accounts, attributing the differences between types 1 and 2 to parameter settings of the Agree system, do not offer a satisfactory explanation. This paper proposes that the difference is due to the status of the agreement of the verb, which is pronominal in type 1 languages and purely grammatical in type 2. Arguments for this analysis are found in Case, Binding Theory and information structure. The focus in this paper is on the Romance languages French and Italian, and on the Bantu languages Sesotho and Makhuwa.
The paper explores parallelisms between Bantu (specifically Otjiherero) and Romance (through Latin and Spanish) with respect to left and right peripheries, and subject and object clitics. The analysis is formulated in Dynamic Syntax (DS, Cann et al. 2005) and centrally involves notions of structural underspecification. Through providing detailed analyses of different word order possibilities in the Bantu and Romance languages discussed, we show how DS concepts of structural growth over initially underspecified tree relations, such as the building of linked structures and unfixed nodes, provide a uniform basis for analyses of word order variation across the two language groups. We then extend our analysis to include Bantu subject/object markers, which we analyze by employing the same formal tools as used in the analysis of Romance (object) clitics, namely unfixed nodes which have to be construed within a tightly locally restricted domain. Empirical support for our analysis comes from restrictions on the presence of object markers in passive and locative inversion constructions in Otjiherero, which we show to follow from independent constraints of the availability of unfixed nodes within a given domain. The analyses of Bantu and Romance presented show that despite differences in surface morphology between the two language groups, both exhibit a striking parallel with respect to the way lexical information and general structure building principles of DS interact. The difference between Romance clitic systems and the agglutinative morphology of Bantu subject and object makers is thus seen to be comparatively superficial, while the DS analysis brings out the strong structural parallelism between the two language groups.
In this paper, Italian and Bantu clitic clusters are analyzed and compared. I claim that in both language families, pronouns check case in a low clitic position, and in so doing they reverse the order of arguments. In Italian, clitics move to a high clitic position where they check person and number features. Different types of Italian clusters are individuated: they differ syntactically, phonologically, and morphologically. As in Kayne’s (1994) proposal, clusters can be formed either on one single functional head or on adjacent heads; the former type can appear both in enclisis and proclisis, the latter only in proclisis. Clitic pronouns can end with an epenthetic vowel or a class marker/ inflectional morpheme; the former (consonantal clitics) can appear in clitic clusters in any position, the latter (morphologically complex clitics) cannot be the first element in clitic clusters dominated by one single head. Finally, some clusters are inserted as lexical units, others are two independent words: only the former display the linking vowel [e]. The intricate interplay of these (partially) independent properties explains a number of restrictions on clusters found in Italian (and other Romance languages). The hierarchy of person and number features in the high clitic position is also discussed, which explains other restrictions on clusters.
Romance pronominal clitics and Bantu object markers vary in gender and number, replace arguments, and surface to the left of the verbal root in declarative clauses. Both types of morphemes are regularly analyzed as affixes on the verb. It is argued that both have syntactic properties that justify treating them differently from lexical affixes. The argument is first made for French unstressed pronominal objects like me, le, en, drawing from their distribution in contemporary French as well as from their historical behavior. It is shown that they are syntactic objects, and that their morphophonological affix-like properties should be treated independently of their syntactic behavior. Then evidence is presented pointing to the fact that Bantu object markers are also syntactic objects of some sort. The last section asks whether one can arrive at a unified treatment of both elements. The conclusion is that by focusing on the morphophonological properties of these pronominal elements, one runs the risk of overlooking their syntactic properties.
Romance and Bantu languages show a range of contrasting morphological and syntactic properties, ranging from WH extraction strategies and the existence of V2 to directionality of affixation and degree of fusion and agglutination in verbal inflectional morphology. Using Isizulu and Chishona (Bantu) and standard French and Italian (Romance) as examples, this paper correlates these contrasts in terms of a common right-branching Split-INFL structure, and contrasting preferences for movement (STAY, in Optimality Theoretic terms). It is proposed that the same preference which motivates WH extraction strategies in the four languages also motivates the contrast between the relatively prefixal and agglutinative verbal inflectional morphology of Isizulu and Chishona and the suffixal and more fusional verbal inflectional morphology of French and Italian. The conclusions reinforce intuitions that languages belonging to the Romance and Bantu language families are typologically similar.
The paper considers several aspects of Bantu and Romance DPs, including: (i) Noun Class and grammatical gender; (ii) apparently derivational properties of the two; (iii) ordering among nouns and their modifiers; and (iv) concord in DPs. Several conclusions are argued for. Firstly, Bantu Class is a gender system like that of Romance, with gender-specific Spell-Out of number features. Secondly, despite some surface evidence to the contrary, gender/Class is an uninterpretable feature, without derivational functions. Thirdly, DPs of the two languages share a common architecture; and fourth, in both families concord is the result of the Agree relation. Thus many properties of nouns and DPs are common to both language groups, as the hypothesis of UG leads us to expect.
The paper reconsiders the evidence for the interpretable status of φ-features within DP in Romance languages, focusing on Italian. The conclusions are that, given a reasonably local interpretation of what counts as interpretable, gender is not an interpretable feature, and number (or plur, in the sense of Heycock & Zamparelli 2005) is interpretable, but is not sufficient to drive the process of verbal and predicative agreement. The notion of “default”, particularly with respect to the Person feature, is discussed and various alternatives are explored.
This paper claims that feature sharing should be analyzed as the result of at least two different processes, which are named here Agreement and Concord, and inquires how these two processes are manifested inside nominal expressions (NEs). Agreement is the transfer of the Person features of the possessor (the “subject” of the NE) onto some functional head (parallel to subject Agreement in the clause) with the effect that (Genitive) Case is assigned. On the contrary, Concord is the transfer of Number, Word Class, and Case specifications from a functional head onto a modifier, which is first-merged as a Specifier of that functional head. The claim is that, quite differently from Agreement, Concord arises from the merger of a modifier, underspecified for uninterpretable features, in the specifier of a functional head, carrying a copy of those features. In other words, Concord is directly enhanced by the Spec-Head configuration; it does not involve merger of a probe which targets a goal and, as a consequence, never triggers (overt or covert) movement. This proposal can dispense with a number of otherwise unmotivated movements and can derive the different properties of these two kinds of feature sharing phenomena. The argument is supported by observing macro-parallelisms across Bantu and Romance languages, in particular Swahili and Xhosa on the one hand and Romanian and Italian on the other hand.
In this paper, I propose a unified syntactic analysis of Luganda and Italian simple nouns. I argue that Italian and Luganda nouns are formed in the syntax via merge and move operations. More specifically, I show that in both languages all nouns are formed via the merger of the nominalizer head [n] with a nominal stem [LP] yielding the nominal structure [nP [n [LP]]] and that syntactic movement is necessary in the noun formation process of Italian nouns to derive the correct morpheme order.
In order to prove that the structure [nP [n [LP]]] is representative for both languages, I demonstrate that the nominalizer head [n] corresponds to both the Italian gender feature and the Luganda class feature and that, therefore, gender and class are the same feature. The data analysis in sections (2) and (3) of this paper supports the claim that gender and class are the same feature because of their identical inflectional and derivational functions. At the inflectional level, gender and class trigger VP and DP agreement and at the derivational level gender and class function as n-marked heads whose merger with an XP yields a noun.
Based on interface evidence, this paper shows that the interpretation of discourse categories relies on syntactic conditions. Specifically, the existence of a systematic connection between formal properties and discourse functions shows that Topics are licensed in dedicated positions in the C-domain, which correspond to specific realizations at the PF-interface. Different types of Topics thus correspond to specific tonal events and show A’-properties which discard a movement approach. Finally, the interface properties of postverbal subjects are examined, showing that their position does not depend on the argument structure of the verb and is not necessarily different from the position of (referential) preverbal subjects. Though the present analysis is mainly concerned with Italian, a comparison with Bantu data is also proposed in order to show similarities, discuss (apparent) discrepancies and highlight implications for future research.
Word order variation in Romance and Bantu has been related to information structure portrayed in the different discourse functions of the sentential elements involved. Based on the distribution of new information focus in Romance and Bantu, this paper argues that discourse functions need not be directly encoded in syntax. The position defended here is that syntax generates all possible structures which are filtered out at the interface with the phonological component. The prosodic phrasing of these structures is what indicates focused constituents occurring in positions of prominence. The paramount significance of prosody for the determination of focus is particularly illustrated in those cases both in Romance and Bantu where, for syntactic reasons, change in word order is restricted but prosodic effects still accompany focus. The proposed interface approach to focus accounts for the variation in focus strategies and the intimate relation of focus to prosody which is attested in Romance and Bantu. Crucially, the evidence from Romance and Bantu is complementary. The Romance data provide the necessary syntactic evidence for not positing a designated focus position in the syntactic hierarchy for focus, whereas the Bantu data show that prosodic effects may emerge in varying ways, providing evidence for not linking a single syntactic position to a given prosodic effect. The paper in this respect highlights the similarities and differences of the role of prosody in indicating focus in stress versus tone languages.
Both Bantu and Romance languages use a V(erb) S(ubject) construction to express thetic (“out-of-the-blue”) sentences. Two types of languages can be distinguished within these language families, with respect to the verbal agreement in a thetic VS sentence: in type 1 the verb has default agreement, whereas in type 2 the verb agrees with the postverbal subject. In the Bantu languages these two types also display a difference in the use of conjoint and disjoint verb forms. Collins (2004), Carstens (2005), and Baker (2008) have previously analyzed such agreement and word order phenomena. These accounts, attributing the differences between types 1 and 2 to parameter settings of the Agree system, do not offer a satisfactory explanation. This paper proposes that the difference is due to the status of the agreement of the verb, which is pronominal in type 1 languages and purely grammatical in type 2. Arguments for this analysis are found in Case, Binding Theory and information structure. The focus in this paper is on the Romance languages French and Italian, and on the Bantu languages Sesotho and Makhuwa.