The development of constructions from the right edge: a multinomial regression analysis of clitic left and right dislocation in child French

2020 ◽  
pp. 1-25
Author(s):  
Morgane JOURDAIN ◽  
Karen LAHOUSSE
Keyword(s):  
Regression Analysis ◽  
General Trend ◽  
Child Language ◽  
Multinomial Regression ◽  
Corpus Study ◽  
Learning Mechanism ◽  
Left And Right ◽  
The Right ◽  
Syntactic Properties ◽  
Multinomial Regression Analysis

Abstract The aim of the present research is to investigate the development of left and right dislocation in child French through a corpus study of three children until age 2;7 from the corpus of Lyon (Demuth & Tremblay, 2008). We extracted a total of 704 dislocations and analysed their syntactic properties. We show that (i) right dislocations are more frequent than left dislocations and (ii) left dislocations are significantly more complete than right dislocations (fewer omissions of verbs or pronouns). We compare these results to the hypothesis of Freudenthal, Pine, Jones & Gobet (2015, 2016) according to which some properties of child language can be explained by a learning mechanism from the right edge of the sentences from the input. We will show that this hypothesis can explain the general trend found in our data, but it is not sufficient to account for the entire development of dislocation in French.

scholarly journals Longitudinal stent elongation or shortening after deployment in the coronary arteries: which is dominant?

2021 ◽  
Vol 73 (1) ◽  
Author(s):  
Magdy Algowhary ◽  
Mohammed Aboel-Kassem F. Abdelmegid
Keyword(s):  
Regression Analysis ◽  
Intravascular Ultrasound ◽  
Coronary Arteries ◽  
Coronary Stents ◽  
Multinomial Regression ◽  
Stent Deployment ◽  
Stent Length ◽  
Multinomial Regression Analysis

Abstract Background Stent manufacturers always record stent shortening data while they do not record stent elongation data. The aim of this study is to identify both stent shortening and elongation occurring after deployment in the coronary arteries and know their percentage. Results The length of coronary stents was measured by intravascular ultrasound (IVUS) by (1) edge-to-edge (E-E) length, measured from the appearance of the first distal strut to the last proximal strut, and (2) area-to-area (A-A) length, measured from the first distal struts seen at more than one IVUS quadrant to the last proximal struts seen at more than one IVUS quadrant. Stent shortening was defined as both E-E and A-A lengths were shorter than the manufacturer box-stated length (shortened group). Stent elongation was defined as both E-E and A-A lengths were longer than the manufacturer box-stated length (elongated group), otherwise unchanged group. Consecutive 102 stents deployed in ischemic patients were included. Stent elongation was detected in 67.6% (69 stents), and shortening was detected in 15.7% (16 stents), while unchanged stents were detected in 16.7% (17 stents). Although the 3 groups had similar box-stated length and predicted foreshortened length, they had significantly different measurements by IVUS, p<0.001 for each comparison. Differences from box-stated length were 1.9±1.4mm, −1.4±0.4mm, and 0.4±0.3mm, respectively, p<0.001. The elongated group had significantly longer differences from the corresponding box-stated and predicted foreshortened lengths, while the shortened group had significantly shorter differences from the corresponding box-stated length and similar foreshortened length. By multinomial regression analysis, the plaque-media area and stent deployment pressure were the independent predictors of the stent length groups, p=0.015 and p=0.026, respectively. Conclusions Change in stent length is not only shortening—as mentioned in the manufacturer documents—but also stent elongation. Stent elongation is dominant, and the most important predictors of longitudinal stent changes are plaque-media area and stent deployment pressure.

scholarly journals On the givenness of OV word order: a (re)examination of OV/VO variation in Old English

2018 ◽  
Vol 24 (1) ◽  
pp. 1-22
Author(s):  
TARA STRUIK ◽  
ANS VAN KEMENADE
Keyword(s):  
Word Order ◽  
Information Structure ◽  
Mixed Model ◽  
Linear Mixed Model ◽  
Multinomial Regression ◽  
Corpus Study ◽  
History Of ◽  
The Given ◽  
New Hypothesis ◽  
Multinomial Regression Analysis

OV/VO variation in the history of English has been a long-debated issue. Where earlier approaches were concerned with the grammatical status of the variation (see van Kemenade 1987; Pintzuk 1999 and many others), the debate has shifted more recently to explaining the variation from a pragmatic perspective (see Bech 2001; Taylor & Pintzuk 2012a), focusing on the given-before-new hypothesis (Gundel 1988) and its consequences for OV/VO. While the work by Taylor & Pintzuk (2012a) focuses specifically on the newness of VO orders, the present study is particularly concerned with the givenness of OV word order. It is hypothesized that OV orders are the result of leftward movement from VO orders, triggered by givenness. A corpus study on a database of subclauses with two verbs and a direct object, collected from the YCOE (Taylor et al.2003) corpus, and subsequent multinomial regression analysis within a generalized linear mixed model shows that OV word order is reserved for given objects, while VO objects are much more mixed in terms of information structure. We argue that these results are more in line with an analysis which derives all occurring word orders from a VO base than an analysis which proposes the opposite.

Sign in / Sign up

Export Citation Format

Share Document