Proceedings of the 8th Workshop on Cognitive Modeling And Computational Linguistics (CMCL 2018) 2018
DOI: 10.18653/v1/w18-0106
|View full text |Cite
|
Sign up to set email alerts
|

Modeling bilingual word associations as connected monolingual networks

Abstract: Word associations are a common tool in research on the mental lexicon. Studies report that bilinguals produce different word associations in their non-native language than monolinguals, and propose at least three mechanisms responsible for this difference: bilinguals may rely on their native associations (through translation), on collocational patterns, and on the phonological similarity between words. In this paper, we first test the differences between monolingual and bilingual responses, showing that these … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
5
0

Year Published

2019
2019
2024
2024

Publication Types

Select...
4
1

Relationship

0
5

Authors

Journals

citations
Cited by 5 publications
(5 citation statements)
references
References 32 publications
0
5
0
Order By: Relevance
“…It is also worth mentioning several recent evidence accumulation models allowing to represent the associations' corelatedness by varying the geometry of the response thresholds (e.g., correlated associations could be represented by closer points on a circular response threshold; Kvam, 2019;Kvam & Turner, 2021;Smith et al, 2020). Our method for estimating association spaces (which estimates the number and strength of possible associations but does not identify the actual words) also did not allow us to address other known contributors to associative strength, including orthographic and phonological similarity, as well as certain characteristics of the cue and the association, such as the general frequency of each possible association, independent of the cue-a particularly strong predictor in previous studies (Matusevych & Stevenson, 2019;Nelson et al, 2005).…”
Section: Limitations and Future Directionsmentioning
confidence: 98%
“…It is also worth mentioning several recent evidence accumulation models allowing to represent the associations' corelatedness by varying the geometry of the response thresholds (e.g., correlated associations could be represented by closer points on a circular response threshold; Kvam, 2019;Kvam & Turner, 2021;Smith et al, 2020). Our method for estimating association spaces (which estimates the number and strength of possible associations but does not identify the actual words) also did not allow us to address other known contributors to associative strength, including orthographic and phonological similarity, as well as certain characteristics of the cue and the association, such as the general frequency of each possible association, independent of the cue-a particularly strong predictor in previous studies (Matusevych & Stevenson, 2019;Nelson et al, 2005).…”
Section: Limitations and Future Directionsmentioning
confidence: 98%
“…Several computational models of human bilingualism exist, see Li (2013) for an overview. More relatedly to the current work that uses distributional approaches, aspects of the bilingual lexicon have been proposed for word associations (Matusevych et al, 2018). These associations are different in bilingual and monolingual speakers.…”
Section: Related Workmentioning
confidence: 99%
“…Additionally, we found that words with an older age of acquisition were less likely to have similar representations in M CW and M WC . This is again likely caused by the fact that words that are acquired later in life are less likely to be generated as responses in free association (Matusevych & Stevenson, 2018) and thus have fewer observations in our training data. We did not find an effect of word concreteness in our data.…”
Section: The Effects Of Fine Tuningmentioning
confidence: 99%
“…Researchers have made progress on (a) by using distributed semantics (DS) models (illustrated in Figure 1A), which exploit statistics of word use in large collections of text to derive semantic representations of words in the form of real-valued vectors (Howard et al, 2011;Jones & Mewhort, 2007;Landauer & Dumais, 1997;Mikolov et al, 2013;Pennington et al, 2014; for reviews, see Bhatia et al, 2019;Lenci, 2018;or Günther et al, 2019). These, and related models based on corpus statistics (e.g., Chaudhari et al, 2011;Ji et al, 2008;Matusevych & Stevenson, 2018;Peirsman & Geeraerts, 2009), can predict associations between pairs of words (e.g., stork and baby) by using the co-occurrence frequencies and the absolute frequencies of the words in language (Griffiths et al, 2007;Jones et al, 2018;Nematzadeh et al, 2017;Pereira et al, 2016). However, the combination of these representations with cognitive process models of memory-that is, requirement (b) from above-has been limited.…”
mentioning
confidence: 99%