scholarly journals Do We Need Neural Models to Explain Human Judgments of Acceptability?

2020 ◽  
Author(s):  
Wang Jing ◽  
M. Alex Kelly ◽  
David Reitter
Keyword(s):  
Word Order ◽  
Native Speakers ◽  
Native Speaker ◽  
Language Models ◽  
Neural Models ◽  
Word Similarity ◽  
Simple Language ◽  
N Gram ◽  
Acceptability Judgments ◽  
Processing Language

Native speakers can judge whether a sentence is an acceptable instance of their language. Acceptability provides a means of evaluating whether computational language models are processing language in a human-like manner. We test the ability of language models, simple language features, and word embeddings to predict native speakers’ judgments of acceptability on English essays written by non-native speakers. We find that much sentence acceptability variance can be captured by a combination of misspellings, word order, and word similarity (r = 0.494). While predictive neural models fit acceptability judgments well (r = 0.527), we find that a 4-gram model is just as good (r = 0.528). Thanks to incorporating misspellings, our 4-gram model surpasses both the previous unsupervised state-of-the art (r = 0.472), and the average native speaker (r = 0.46), demonstrating that acceptability is well captured by n-gram statistics and simple language features.

Lexical and discourse factors in the second language acquisition of Spanish word order

2003 ◽  
Vol 19 (4) ◽  
pp. 273-304 ◽  
Author(s):  
Tammy Jandrey Hertel
Keyword(s):  
Second Language Acquisition ◽  
Word Order ◽  
Native Speakers ◽  
Native Speaker ◽  
Control Group ◽  
Order Effects ◽  
Advanced Learners ◽  
Spanish Word ◽  
Native Speakers Of English ◽  
Beginning Learners

This study investigates the acquisition of Spanish word order by native speakers of English. Specifically, it considers the development of sensitivity to the distinct interpretations of subject-verb (SV) vs. verb-subject (VS) order, as determined by lexical verb class (unaccusative and unergative verbs) and discourse structure.Participants included a native speaker control group and learners at four proficiency levels. Results from a contextualized production task indicate that beginning learners transferred the SV order of English for all structures. Intermediate learners showed a gradual increase in the production of lexically and discourse-determined inversion, although their data was also characterized by indeterminacy and variability. The advanced learners demonstrated a sensitivity to the word order effects of unaccusativity and discourse factors, but also tended to overgeneralize inversion to unergative verbs in a neutral discourse context.

Discourse Configurationality and the Noun Phrase in Eastern Armenian

2020 ◽  
Vol 50 (2) ◽  
pp. 137-166
Author(s):  
Katherine Hodgson
Keyword(s):  
Noun Phrase ◽  
Word Order ◽  
Native Speakers ◽  
Native Speaker ◽  
Syntactic Structure ◽  
Original Proposal ◽  
Use Of Data ◽  
Word Order Variation ◽  
Order Variation ◽  
Surface Patterns

Abstract The subject of this paper is discourse-related word order variation associated with nominals in Eastern Armenian. Hale (1983) proposes that there are some languages which are non-configurational, i.e. lacking hierarchical syntactic structure. The properties that he proposed to be characteristic of non-configurational languages are a) free word order b) extensive use of null anaphora and c) discontinuous constituents. Armenian possesses all of these characteristics. However, since Hale’s original proposal, it has been pointed out that many apparently non-configurational languages do in fact have hierarchical syntactic structure, but that the surface patterns are determined primarily by discourse properties rather than by grammatical relations. These languages have been termed ‘discourse configurational’, defined by É. Kiss (1995) as follows: a language is discourse configurational if (discourse-) semantic functions topic (what sentence is ‘about’) and/or focus (identification) are associated with particular structural positions. It has been argued that this is indeed the case for the clause in Eastern Armenian (see e.g. Comrie (1984), Megerdoomian (2011) and Tamrazian (1994)). Making use of data from approximately 10,000 words of transcribed spontaneous speech by native speakers of Eastern Armenian discussion with native speaker consultants, and the Eastern Armenian National Corpus (www.eanc.net), I argue that the noun phrase exhibits similar discourse configurational properties to those found in the clause, and that these are responsible for word order variation within it. The interaction between noun-phrase-internal discourse-related movement and analogous discourse-related movement operations within the clause is responsible for the appearance of apparently discontinuous noun phrases. Thus the existence of Hale’s ‘non-configurational’ properties in EA does not justify the proposal that this language lacks hierarchical syntactic structure.

scholarly journals Old Grammars New (?) Scope: Adjective Placement in Native and Non-Native Spanish

Languages ◽  
2021 ◽  
Vol 6 (1) ◽  
pp. 22
Author(s):  
Erin Pettibone ◽  
Ana Teresa Pérez-Leroux ◽  
Gabrielle Klassen
Keyword(s):  
Word Order ◽  
Native Speakers ◽  
Native Speaker ◽  
First Language ◽  
Mirror Image ◽  
Variable Order ◽  
Show Evidence ◽  
Preference Task ◽  
Study Participants ◽  
Nominal Position

Prior studies have examined the association between modifying adjective placement and interpretation in second language (L2) Spanish. These studies show evidence of convergence with native speaker’s intuitions, which is interpreted as restructuring of the underlying grammar. Two issues deserve further study: (i) there are debates on the nature of native speaker’s interpretations; (ii) previous results could be explained by a combination of explicit instruction and access to the first language (L1). The present study re-examines native and non-native intuitions on the interpretation of variable order adjectives in pre-nominal and post-nominal positions, and extends the domain of inquiry by asking if L2 learners have intuitions about the order of two-adjective sequences, which appear in mirror image order in English and Spanish (faded blue pants vs. pantalones azules desteñidos). Two-adjective sequences are rare in the input, not typically taught explicitly, and have a different word order that cannot be [partially] derived from the L1 subgrammar. Two groups of non-native speakers (n = 50) and native speaker controls (n = 15) participated in the study. Participants completed a preference task, testing the interaction between word order and restrictive/non-restrictive interpretation, and an acceptability judgement task, testing ordering intuitions for two-adjective sequences. Results of the preference task show that the majority of speakers, both native and non-native, prefer variable adjectives in a post-nominal position independent of interpretation. Results of the acceptability judgement task indicate that both native and non-native speakers prefer mirror image order. We conclude that these results support underlying grammar reanalysis in L2 speakers and indicate that the semantic distribution of variable adjectives is not fully complementary; rather, the post-nominal position is unmarked, and generally preferred by both native and non-native speakers.

The Involvement of Native Speakers in Teaching and Learning Japanese Language at UiTM

2020 ◽  
Vol 4 (3) ◽  
pp. 14
Author(s):  
Choong Pow Yean ◽  
Sarinah Bt Sharif ◽  
Normah Bt Ahmad
Keyword(s):  
Language Learning ◽  
Language Learners ◽  
Language Proficiency ◽  
Teaching And Learning ◽  
Native Speakers ◽  
Native Speaker ◽  
Japanese Culture ◽  
Japanese Language ◽  
Foreign Language Learners ◽  
Lexical Density

The Nihongo Partner Program or “Japanese Language Partner” is a program that sends native speakers to support the teaching and learning of Japanese overseas. The program is fully sponsored by The Japan Foundation. The aim of this program is to create an environment that motivates the students to learn Japanese. This study is based on a survey of the Nihongo Partner Program conducted on students and language lecturers at UiTM, Shah Alam. This study aims to investigate if there is a necessity for native speakers to be involved in the teaching and learning of Japanese among foreign language learners. Analysis of the results showed that both students and lecturers are in dire need of the Nihongo Partner Program to navigate the learning of the Japanese language through a variety of language learning activities. The involvement of native speaker increases students’ confidence and motivation to converse in Japanese. The program also provides opportunities for students to increase their Japanese language proficiency and lexical density. In addition, with the opportunity to interact with the native speakers, students and lecturers will have a better understanding of Japanese culture as they are able to observe and ask the native speakers. Involvement of native speakers is essential in teaching and learning of Japanese in UiTM.

The role of cross-linguistic stress pattern frequency and word similarity on the acquisition of English stress pattern by native speakers of Brazilian Portuguese

2012 ◽  
Vol 12 ◽  
Author(s):  
Amanda Post Silveira
Keyword(s):  
Second Language ◽  
Native Speakers ◽  
Mother Tongue ◽  
American English ◽  
Brazilian Portuguese ◽  
Stress Pattern ◽  
Target Language ◽  
English As Second Language ◽  
Word Similarity ◽  
L1 Transfer

This is a preliminary study in which we investigate the acquisition of English as second language (L2[1]) word stress by native speakers of Brazilian Portuguese (BP, L1[2]). In this paper, we show results of a multiple choice forced choice perception test in which native speakers of American English and native speakers of Dutch judged the production of English words bearing pre-final stress that were both cognates and non-cognates with BP words. The tokens were produced by native speakers of American English and by Brazilians that speak English as a second language. The results have shown that American and Dutch listeners were consistent in their judgments on native and non-native stress productions and both speakers' groups produced variation in stress in relation to the canonical pattern. However, the variability found in American English points to the prosodic patterns of English and the variability found in Brazilian English points to the stress patterns of Portuguese. It occurs especially in words whose forms activate neighboring similar words in the L1. Transfer from the L1 appears both at segmental and prosodic levels in BP English. [1] L2 stands for second language, foreign language, target language. [2] L1 stands for first language, mother tongue, source language.

Learning N-Gram Language Models from Uncertain Data

2016 ◽  
Author(s):  
Vitaly Kuznetsov ◽  
Hank Liao ◽  
Mehryar Mohri ◽  
Michael Riley ◽  
Brian Roark
Keyword(s):  
Uncertain Data ◽  
Language Models ◽  
N Gram

Rescore in a Flash: Compact, Cache Efficient Hashing Data Structures for n-Gram Language Models

2020 ◽  
Author(s):  
Grant P. Strimel ◽  
Ariya Rastrow ◽  
Gautam Tiwari ◽  
Adrien Piérard ◽  
Jon Webb
Keyword(s):  
Data Structures ◽  
Language Models ◽  
Cache Efficient ◽  
N Gram

Towards corpus and model: Hierarchical structured-attention-based features for Indonesian named entity recognition

2021 ◽  
pp. 1-12
Author(s):  
Yingwen Fu ◽  
Nankai Lin ◽  
Xiaotian Lin ◽  
Shengyi Jiang
Keyword(s):  
Language Processing ◽  
State Of The Art ◽  
Named Entity Recognition ◽  
Entity Recognition ◽  
Language Models ◽  
Neural Models ◽  
Performance Models ◽  
Named Entity ◽  
High Resource ◽  
Benchmark Datasets

Named entity recognition (NER) is fundamental to natural language processing (NLP). Most state-of-the-art researches on NER are based on pre-trained language models (PLMs) or classic neural models. However, these researches are mainly oriented to high-resource languages such as English. While for Indonesian, related resources (both in dataset and technology) are not yet well-developed. Besides, affix is an important word composition for Indonesian language, indicating the essentiality of character and token features for token-wise Indonesian NLP tasks. However, features extracted by currently top-performance models are insufficient. Aiming at Indonesian NER task, in this paper, we build an Indonesian NER dataset (IDNER) comprising over 50 thousand sentences (over 670 thousand tokens) to alleviate the shortage of labeled resources in Indonesian. Furthermore, we construct a hierarchical structured-attention-based model (HSA) for Indonesian NER to extract sequence features from different perspectives. Specifically, we use an enhanced convolutional structure as well as an enhanced attention structure to extract deeper features from characters and tokens. Experimental results show that HSA establishes competitive performance on IDNER and three benchmark datasets.

Euro-English: A Swedish perspective

English Today ◽  
2003 ◽  
Vol 19 (2) ◽  
pp. 35-41 ◽  
Author(s):  
Marko Modiano
Keyword(s):  
European Union ◽  
Native Speakers ◽  
Native Speaker ◽  
The Other ◽  
The European Union ◽  
Lingua Franca ◽  
European Continent

This survey considers the emergence of English as a language shared across the European Union in particular and the European continent at large, and together with its distinctive ‘lingua franca’ dimension among the mainland European nations. It considers in particular the situation of ‘non-native speakers’ who regularly use the language as well as the concept of a ‘Euro-English’ in general and the Swedish, ‘Swenglish’ and English relationship on the other. It concludes by considering the liberation of non-native users from ‘the beginning of native-speaker norms’.

Sign in / Sign up

Export Citation Format

Share Document