Search Results

You are looking at 1 - 10 of 48 items :

  • "conditional inference tree" x
Clear All

purpose of the GSSG. Further explanation will be provided in Section 2 . Quantitatively, we provide typological data obtained from a phylogenetically and geographically weighted dataset of 400 languages worldwide, and investigate the actual distribution of sortal classifiers and morphosyntactic plural markers in languages of the world via the statistical test of conditional inference tree ( Breiman 2001 ). As a disclaimer, the main goal of this study is the validation of the GSSG, not necessarily the universalist view that attempts to account for it. The observations

structures are functionally constrained or conditioned. To this end, we focus on the variable use of two different types of gerunds (i. e., deverbal nominalizations formed with the -ing suffix) that were competing over the same (grammatical) contexts in the seventeenth century, using two types of multifactorial models that have recently started to gain traction in variationist studies: random forests, and conditional inference trees ( Tagliamonte and Baayen 2012 ; other applications in, e. g., Robinson 2011 ; Kerz and Wiechmann 2013 ; Claes 2016 ; Szmrecsanyi et al

, Abbas Ali and Seyyed Ehsan Golparvar. 2016. The Sequencing of Adverbial Clauses of Time in Academic English: Random Forest Modelling. Journal of Language Modelling 4(2), 225-244. Rezaee, Abbas Ali and Seyyed Ehsan Golparvar. 2017. Conditional Inference Tree Modelling of Competing Motivators of the Positioning of Concessive Clauses: The Case of a Non-native Corpus. Journal of Quantitative Linguistics 24(2-3), 89-106. Saif, Shahrzad. 2006. Aiming for Positive Washback: A Case Study of International Teaching Assistants. Language Testing 23 (1). 1-34. Schoonen, Rob. et al

. These more idiosyncratic functional and formal properties are often insufficiently addressed in learner grammars. The article demonstrates, on the basis of two case studies, how insights and methods from Construction Grammar can help to improve the presentation of this topic. More specifically, it elaborates on the key determinants of L2 construction learning (involving frequency, proto- typicality and form-function mapping, among others) and illustrates what statis- tical techniques such as collostructional analysis and conditional inference trees can reveal

) home. I will focus on American English, where this vari- ation seems to be more common, as one can conclude from language users’ intui- tions and experts’ comments.2 In this study, I will test some of the factors that are mentioned in these discussions, such as figurative vs. literal meaning and the semantics of arrival (see Section 2.2). The data, which will be described in Section 2.1, come from the spoken component of the Corpus of Contemporary American English (COCA) (Davies 2008–). I use conditional inference trees (Hothorn et al. 2006) to test the impact

, which suggests a largely exonormative orienta- tion with some first signs of (potential) innovations. Cross-varietal analyses using conditional inference trees and random forests reveal that the length of since- and while-clauses is most heavily influenced by the semantics of a clause. The position of since-clauses, however, is most dependent on variety status, whereas the posi- tion of while-clauses is most dependent on clause meaning. 1 Introduction Adverbial subordination has been studied intensively with regard to the syntac- tic and semantic properties of

individual speakers and the NPs’ head nouns were included as random intercepts, I show that speakers pluralize presentational haber in 47% of the cases and that the variation is conditioned by three general cognitive con- straints (markedness of coding, structural priming, and statistical pre-emption). Using a conditional inference tree, I show that the former two cognitive con- straints work in tandem to promote the pluralized construction for the encoding of conceptualizations that statistical preemption tends to reserve to the singular construction. The results

Anwendung kommt, auf dem aus der einfachen Kollexemanalyse gewonnenen Maß der Kollostruktionsstärke (G-Werte), das sich auf die bedingte Wahrscheinlichkeit stützt: Die Häufigkeiten der V INF -Lexeme im jeweiligen Korpusteil sowie die Gesamtfrequenz aller V INF -Lexeme des entsprechenden Korpusteils werden dabei in Betracht gezogen. 2.4 Random Forests von Conditional Inference Trees Um die funktionalen Veränderungen im Sinne der semantischen und syntaktischen Gebrauchsmerkmale der verstehen -Konstruktion während des 20. Jahrhunderts zu erfassen, werden drei statistische

.1% linguistics, cultural studies 1.2% 3.3% 2.5% 46.6% 25.6% 20.8% 29.3% 70.7% 4 Methodology 4.1 Conditional inference trees and forests Decision trees are a top-down binary data-splitting approach, which are popular because of their easy interpretability and low bias. Decision trees can also handle missing values using so-called surrogate splits. If a variable is missing for a specific observation, another predictor variable is used, such that this split is similar to the best split ( Twala 2009 ). Bootstrap aggregation (bagging) reduces the high variance of a single decision

motivated learning behaviour, using correlation and regression procedures ( Sections 4 and 5 ). However, we here propose novel ways of analyzing the data in this way: On the one hand, we use alternative methods of regression by means of random forests and conditional inference trees, which allow us to estimate the relative importance of a large number of variables for a comparatively small number of observations. On the other hand, we complement the traditional ‘scale-based’ approach, in which summarized scales of questionnaire items act as predictors, by an ‘item