This study examines some lexical and morphosyntactic variation found among the Swahili varieties in Zanzibar, Tanzania. It has been reported that there are three varieties in the island, and the new data collected in the villages inside the island suggest that there is a need for closer investigation and finer categorization of the Zanzibar varieties. Furthermore, there has been little discussion about the relationship between the Standard variety and other varieties or variation in the use of these varieties. In this paper, we report on the use of these Zanzibar varieties, namely, the town variety, Kiunguja-Mjini, the Northern varieties Kichaani, Kikibeni, Kitumbatu-Gomani, Kinungwi and Kimatemwe, and the Southern varieties Kijambiani, Kipaje and Kimakunduchi at the level of the lexicon. In addition, we examine the varieties of the Northern province – Kichaani, Kikibeni, Kitumbatu-Gomani, Kinungwi, and those of the Southern province – Kijambiani, Kipaje at the level of the grammar. In this paper, we concentrate on tense/aspect, the copula sentence, relative clause, and imperative. Among the data on these languages, we investigate, in particular, the variation among these varieties on the one hand, and the variation between these varieties and Standard Swahili on the other.

No description available.

No description available.

This study incorporates insights from Usage Based Grammar (UBG) into variationist research on morphosyntactic variation in Spanish. Specifically, this dissertation investigates the impact on pronoun use of lexical frequency, or the number of times a finite verb appears in a large data set based on spontaneous speech from 221 speakers in two locales (New York City, NY and Boston, MA), as well as a series of context-based frequency metrics in a Variationist study of Spanish Subject Personal Pronoun (SPP) variation (e.g. Yo creo vs. creo ‘I think’). This investigation elucidates the nature of frequency effects (both lexical and contextual) on pronoun use and on the other linguistic factors that have been shown to impact pronoun use. Through this investigation, this dissertation is able to draw conclusions on the nature of linguistic variation and make inferences surrounding the mental representations underlying sociolinguistic patterns. In the past, frequency has been investigated in subject pronoun production as it pertains to the rate of the finite verb, with researchers counting the instances of each verb's occurrence within a corpus. This approach has produced mixed results. One study has shown that frequency modulates or amplifies the effects of other linguistic predictors, providing evidence that suggests lexical frequency does not directly impact pronoun use in a uniform or monotonic way (Erker & Guy, 2012). A few studies have replicated some version of these modulating effects, though they have not found as consistent amplification effects across linguistic constraints. Other studies have found contradictory frequency effects, showing only a main effect of frequency (high frequency corresponding to high pronoun use in some studies and low pronoun use in others) with no amplification effects, or no frequency effects at all. Further, Usage Based Grammar frameworks, which are often referenced in studies exploring lexical frequency, posit that speakers are not only sensitive to the rate of use of linguistic forms, but also the detailed contexts in which these forms appear. Such “rich memories”, as they are referred to in UBG, are said to constitute the mental representations of these forms. The mixed results in the literature, together with the UBG notion of rich memories, motivate the current study, which investigates the relationship between contextual frequency and pronoun use, since contextual frequency metrics (as opposed to overall frequency) might shed more light on frequency effects in morphosyntactic variation. The contextual frequency metrics analyzed in the current dissertation consist of the frequencies at which finite verbs appear in four combinations of the factor values of two variables, referred to as Switch Reference (i.e. whether the previous verb has a different referent or the same referent as the target site of variation) and Preceding Pronoun (i.e. whether the immediately preceding site of pronominal variation has a pronoun present or absent). The four combinations on which contextual frequency metrics are based are therefore: (1) ‘Different Referent/Preceding Pronoun Present’, (2) ‘Different Referent/Preceding Pronoun Absent’, (3) ‘Same Referent/Preceding Pronoun Present’, or (4) ‘Same Referent/Preceding Pronoun Absent’. Analysis of 88,001 tokens of pronominal presence or absence generally replicate the modulating effects of overall verb frequency observed by Erker & Guy (2012), i.e. the effects of several linguistic factors are amplified for frequent verb forms. Moreover, the analysis of contextual frequency reveals that verb forms must reach a certain overall frequency threshold in order for contextual properties to impact pronoun use. This finding aligns with the UBG prediction that the most frequent context in which a verb appears will dominate the overall pronominal tendencies of the verb, as long as that verb is sufficiently frequent in discourse. Overall, this study concludes that the linguistic variation observed in language use aligns with the usage-based approach that contextual frequency effects accumulate in the mental representations that underlie sociolinguistic patterns.

El presente trabajo abarca el estudio de la variación en las construcciones adverbiales locativas. Esta variación se manifiesta en dos niveles: (1) la alternancia entre la construcción estándar ADVERBIO + preposición de + PRONOMBRE PERSONAL (por ejemplo, delante de mí), y su variante normativamente no recomendada ADVERBIO + PRONOMBRE POSESIVO (por ejemplo, delante mío); y (2) la alternancia entre la construcción posesiva con el sufijo -o (por ejemplo, delante mío), y la construcción posesiva con el sufijo -a (por ejemplo, delante mía), variante normativamente aún más disuadida. El objetivo fue estudiar estas dos variables lingüísticas y sus posibles predictores sociolingüísticos. Con esta finalidad, se han analizado datos provenientes de Twitter. La recolección de datos se llevó a cabo en febrero del 2017. Se recogieron 2357 observaciones de construcciones adverbiales locativas de 21 países de habla hispana (incluyendo los Estados Unidos), de las cuales 1006 eran de la variante posesiva. En cuanto a la variación entre construcciones preposicionales y posesivas, se clasificaron los datos según cuatro variables independientes: el género (o sexo) del tuitero, la procedencia geográfica del tuit, el tipo de adverbio precedente y el número gramatical del referente. Con relación al uso del sufijo -a del posesivo, las observaciones fueron clasificadas según dos variables adicionales: la vocal final del adverbio y el género del referente gramatical. Se ha construido un modelo estadístico de regresión logística que mide el efecto de las variables independientes investigadas sobre el uso de (1) la construcción con el posesivo y (2) la construcción con el posesivo del sufijo -a respectivamente. Referente a (1), la alternancia en los complementos, el análisis revela que la zona dialectal del tuitero, el tipo del adverbio precedente y el número gramatical son los factores estadísticamente significativos para predecir el uso de la construcción posesiva. En lo que concierne a (2), la alternancia en los sufijos del posesivo, los datos apuntan a que tanto la zona dialectal del tuitero como el género del referente gramatical condicionan el uso del posesivo con la desinencia -a. / The present thesis studies the morphosyntactic variation in adverbial locative phrases in Spanish. This variation occurs at two levels: (1) the alternation between the standard construction ADVERB + the preposition de + PERSONAL PRONOUN (as in delante de mí) and the normatively nonstandard construction ADVERB + POSSESSIVE PRONOUN (as in delante mío); (2) the alternation between the possessive construction with the suffix -o (as in delante mío), and the possessive construction with the suffix -a (as in delante mía), the last variant being normatively highly discouraged. The goal of this study was to investigate these two linguistic variables and their possible sociolinguistic predictors. The data was obtained by API searches of Twitter data in February of 2017 and yielded material from 21 Spanish-speaking countries (including the US). 2357 tweets containing the constructions of interest were gathered, of which 1006 were possessive constructions. Concerning the alternation between constructions with prepositions and possessives, the data was coded according to four independent variables: the gender (or sex) of the Twitter user, the location of the user, the type of adverb used in the phrase, and the grammatical number of the referent. As to the use of the suffix -a with possessives, two additional variables were coded for, the final vowel of the preceding adverb and grammatical gender of the referent. A statistical model was constructed in order to measure the effect of the independent variables on the use of (1) the possessive construction, and (2) the possessive construction with the suffix -a, respectively. The results of the first analysis indicate that the geographical region of the Twitter user, as well as the preceding adverb and the grammatical number of the referent, were statistically significant in predicting the use of the possessive construction. Furthermore, with regards to the variation in the suffix of the possessive pronoun, the analysis supports the view that the geographical region of the user and the grammatical gender of the referent are predictors that condition the use of the possessive pronoun with the suffix -a.