CAOSS and transcendence: Modeling role-dependent constituent meanings in compounds

Günther, Fritz; Marelli, Marco

doi:10.1007/s11525-021-09386-6

Cited by 10 publications

(16 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, the model has been employed to detect automatic meaning-combination effects during the processing of novel compounds , and to demonstrate that semantic transparency effects during the processing of familiar compounds should also be attributed to such active meaningcomposition processes . It even allows us to empirically investigate and account for phenomena that have been notoriously hard to capture using other approaches, such as the systematic semantic shift experienced by some words when they are used as compound constituents (Günther & Marelli, 2021), or a language-invariant mechanism of compounding by being able to partially predict compound meanings in one language from a CAOSS model trained on another language (Günther & Marelli, 2018). Taken together, these studies also show that compositional distributional models have been evaluated using a wider range of methods and data than just by comparing human judgments to distributional similarities, which is the only class of phenomena that Lake and Murphy (2021) picked out when reviewing the current empirical status of these models.…”

Section: Novel Combinationsmentioning

confidence: 99%

Patterns in CAOSS: Distributed representations predict variation in relational interpretations for familiar and novel compound words

Günther

Marelli

2022

Cognitive Psychology

Self Cite

View full text Add to dashboard Cite

Section: Novel Combinationsmentioning

confidence: 99%

Patterns in CAOSS: Distributed representations predict variation in relational interpretations for familiar and novel compound words

Günther

Marelli

2022

Cognitive Psychology

Self Cite

View full text Add to dashboard Cite

“…The morphological transcendence hypothesis (Libben, 2014), the updating mechanism proposed by Günther and Marelli (2021), and the notion of semantic diversity (e.g., Hoffman et al, 2011), potentially explain the priming asymmetry between Experiments 2 ( seabird – ocean ) and 4 ( ocean – seabird ). In Experiment 2, where the compound is the prime, the semantic information of the constituent (SEA+) may be narrower, or shifted away from the associate, than the semantic information of its whole-word counterpart (SEA).…”

Section: Discussionmentioning

confidence: 99%

Constituent processing in compound and pseudocompound words.

Melvie¹,

Taikh²,

Gagné³

et al. 2023

Canadian Journal of Experimental Psychology / Revue canadienne

View full text Add to dashboard Cite

Theories of multimorphemic word recognition generally posit that constituent representations are involved in accessing the whole multimorphemic word. Gagné et al. (2018) found that pseudoconstituents and constituents become available when processing pseudocompound and compound masked primes (e.g., sea is activated in season and seabird). Across four experiments, we examine whether readers access the semantic information of such pseudoconstituents and constituents. Experiments 1 and 2 show that masked pseudocompound and compound primes do not influence lexical decision responses to semantic associates of their pseudoconstituents or constituents (e.g., seabird and season do not influence processing of ocean, an associate of sea). Experiments 3 and 4 show that an associate of the first constituent does not influence processing of the pseudocompound but does facilitate processing of the compound (e.g., ocean facilitates processing of seabird but not of season). While compounds have been found to be sensitive to the activation of their constituents via semantic priming (e.g., El-Bialy et al., 2013;Sandra, 1990), our findings suggest that primarily morphological, rather than semantic, activation of the constituents occurs in a masked priming paradigm. Public Significance StatementMorphological information makes an important contribution to reading words, so much so that not only are constituent morphemes (e.g., car in carport) activated, even pseudomorphemes (e.g., car in carpet) are activated. However, this study finds that although both morphemes and pseudomorphemes are activated, the semantic information associated with the morphemes and pseudomorphemes plays a relatively small role in reading such words.

show abstract

“…Of course, this proliferation is further increased when one considers other meaning families of board and their roles within compound words (e.g., whiteboard and blackboard; chessboard and checkerboard; leaderboard, etc.). Günther and Marelli (2021) present an implemented computational system to represent role-dependent constituent phenomena and to test the predictions of morphological transcendence. The behavior of the system supports many of the predictions of transcendence and also reveals new system-based phenomena.…”

Section: Lexical Action Creates Morphological Transcendence: Compound...mentioning

confidence: 99%

“…Günther and Marelli ( 2021 ) present an implemented computational system to represent role-dependent constituent phenomena and to test the predictions of morphological transcendence. The behavior of the system supports many of the predictions of transcendence and also reveals new system-based phenomena.…”

Section: Lexical Action Creates Morphological Transcendence: Compound...mentioning

confidence: 99%

“…The behavior of the system supports many of the predictions of transcendence and also reveals new system-based phenomena. Moreover, the Günther and Marelli ( 2021 ) approach accords with a key underlying assumption of morphological transcendence, namely that the propensity for the proliferation of lexical elements is evidence that the lexical system is not organized and does not function in order to maximize efficiency. Rather, it seems to be designed to maximize opportunity for lexical activation and lexical activity.…”

Section: Lexical Action Creates Morphological Transcendence: Compound...mentioning

confidence: 99%

See 1 more Smart Citation

From Lexicon to Flexicon: The Principles of Morphological Transcendence and Lexical Superstates in the Characterization of Words in the Mind

Libben

2022

Front. Artif. Intell.

View full text Add to dashboard Cite

The field of mental lexicon research has benefitted greatly from the founding metaphor of a dictionary in the mind. That metaphor, however, had its origins in a perspective in which the lexicon was seen as a static repository of representations with fixed structural properties. This paper presents a contrasting view. It is a view that highlights that words are activities that we perform, rather than simply representations that we have. It is proposed that lexical representations are best seen as hierarchies of action within a highly interconnected and dynamic system. The paper presents two principles of lexical organization: morphological transcendence and lexical superstates. The former principle claims that through the activities of language comprehension and production, lexical forms can develop variant forms. Thus, the form key may develop into forms such as key- (e.g., keyboard) and -key, (e.g., turnkey). The paper also discusses how transcendence leads to lexical superstates, which do not have a fixed morphological structure. As part of a lexical superstate, alternative morphological structures exist as potential realizations. Which one is actually realized will depend on the specific circumstances of a lexical action. An account is presented in which the effects of semantic transparency are treated in terms of transcendence and superstate interactions. It is claimed that this approach, which highlights the dynamic and flexible nature of the mental lexicon, has implications for how we approach the modeling of language and cognition in general.

show abstract

CAOSS and transcendence: Modeling role-dependent constituent meanings in compounds

Cited by 10 publications

References 50 publications

Patterns in CAOSS: Distributed representations predict variation in relational interpretations for familiar and novel compound words

Patterns in CAOSS: Distributed representations predict variation in relational interpretations for familiar and novel compound words

Constituent processing in compound and pseudocompound words.

From Lexicon to Flexicon: The Principles of Morphological Transcendence and Lexical Superstates in the Characterization of Words in the Mind

Contact Info

Product

Resources

About