Home |  English version |  Mappa |  Commenti |  Sondaggio |  Staff |  Contattaci Cerca nel sito  
Istituto di linguistica computazionale "Antonio Zampolli"

Torna all'elenco Contributi in atti di convegno anno 2018

Contributo in atti di convegno

Tipo: Abstract in atti di convegno

Titolo: Is inflectional irregularity dysfunctional to human processing?

Anno di pubblicazione: 2018

Formato: Elettronico

Autori: Marzi, C.; Ferro, M.; Pirrelli, V.

Affiliazioni autori: ILC-CNR; ILC-CNR; ILC-CNR;

Autori CNR:

  • MARCELLO FERRO
  • CLAUDIA MARZI
  • VITO PIRRELLI

Lingua: inglese

Abstract: Regularly inflected verb forms are classically associated with the formal transparency and predictability of their internal constituents [ 1 , 2 , 3 ]. Transparency ensures that full forms can be segmented uniquely into their internal constituents: as in walk-s/walk-ed. Predictability allows for a speaker to fill in an empty paradigm cell, using information from other known forms of the same lexical paradigm and its inflection macro-class. From this perspective, irregulars appear to be dysfunctional to the human processing system, as they make it hard to infer - say - bought from buy , or segment bought appropriately into its constituent parts. Likewise, an influential psycholinguistic tradition relegates irregulars to the lexical store, whereas regulars are segmented by rules into their simpler constituents [ 4 , 5 ]. Here, we offer a few reasons for questioning this view. First, transparency and predictability are not dichotomous notions. Secondly, their influence on processing is not unidirectional. Unpredictable stems in irregularly inflected forms of complex inflectional systems provide a lot of processing information, by dynamically constraining the number of possible alternative endings during serial processing. Thirdly, acquisition of word inflection does not consist in associating co-occurring cues and outcomes, but in discriminating between multiple cues that are constantly in competition for their predictive value for a given outcome. We present the results of a few computer simulations with Self-organising Recurrent Neural Networks (TSOMs, [ 8 , 9 ]) that learn how to inflect high-frequency verb paradigms in 6 languages: English, German, Italian, Modern Greek, Modern Standard Arabic and Spanish. After training, each TSOM was tested on a word recognition (serial recoding) and a word production (serial recall) task, and results were analysed with generalised regression models. Processing uncertainty is differently apportioned on regulars and irregulars, depending on the nature of the processing task. While irregulars are harder to produce when they are unknown because they typically have fewer neighbours than regulars have, they are readily accessed once they are acquired, for exactly the same reason. Our data are in line with psycholinguistic evidence [ 10 , 11 ] that lexical processing is paced by two types of uniqueness point: Marslen-Wilson's Uniqueness Point (UP), distinguishing unrelated onset-overlapping words [ 12 ], and the Complex Uniqueness Point (CUP), distinguishing paradigmatically-related words [ 11 ]. Late UPs are inhibitory and elicit prolonged reaction times in acoustic word recognition, explaining an early delay in word recognition of irregular stems. Similarly, late CUPs are inhibitory, and this accounts for a slowdown in the processing advantage of regulars, compared to irregulars, after UP. These structural factors interact in a variety of ways and concurrently affect human processing, to show that irregularly-inflected forms may in fact reflect communicative and processing constraints of the word processor. They provide strong evidence against a processing architecture that assumes compartmentalized, independent processing routes for some specific combinations of these factors (e.g. a rule-based route for a combination of transparency and predictability, and a memory-based route for all other combinations). In addition, they seem incompatible with Bayesian approaches to auditory word comprehension ignoring a word's internal structure [ 13 ]. We suggest that a different design of the human language processor, based on a computational architecture integrating memory and processing as two different dynamics of the same underlying mechanism, can shed light on the complexity of inflection, and vindicate the role of irregular inflection in the system.

Pagine da: 60

Pagine a: 60

Pagine totali: 1

Referee: Sė: Internazionale

Stato della pubblicazione: Published version

Parole chiave:

  • inflectional processing
  • temporal self organizing maps
  • letter prediction
  • morpheme boundary

URL: https://mentallexicon2018.ca/

Congresso nome: 11th International Conference on the Mental Lexicon

Congresso luogo: Edmonton, Alberta (Canada)

Congresso data: 25-28/09/2018

Congresso rilevanza: Internazionale

Congresso relazione: Contributo

Strutture CNR:

Allegati:

  • abstract (application/pdf)
  • poster (application/pdf)
 
Torna indietro Richiedi modifiche Invia per email Stampa
Home Il CNR  |  I servizi News |   Eventi | Istituti |  Focus