Development and validation of a language learning model for behavioral and functional-imaging studies

https://doi.org/10.1016/S0165-0270(01)00525-8Get rights and content

Abstract

To obtain a more precise understanding of the mechanisms involved in language acquisition in healthy subjects and the re-acquisition of language following brain damage, we developed a word learning model which can be used in behavioral and functional-imaging studies. We generated a set of spoken pseudowords, which were normalized with respect to duration and loudness, and (based on a rating study with 40 students) selected 50 pseudowords, which yielded few associations with existing words and were of neutral emotional valence. The selected pseudowords were paired with object drawings in a pseudo-randomized manner and each subject received a different combination of pairings. These auditory–visual pairs were used to train ten subjects solely on the basis of different frequencies of ‘correct’ and ‘incorrect’ pairings. This procedure resembles the acquisition of the first words during childhood. During a five-session training protocol, with one session per day, there was a linear increase of the learning curves, which then reached a plateau. Learning remained stable at 1 and 4 weeks retest intervals. Furthermore, subjects correctly translated the pseudowords into their native language, demonstrating the acquisition of word meanings. These findings indicate that the described associative learning program meets the requirements for a robust and ecologically meaningful model to study the neural mechanisms of language learning and plasticity.

Introduction

Brain regions associated with language acquisition have only been poorly explored, despite the existence of techniques which allow imaging and manipulation of language associated brain regions. Non-invasive imaging of the neural basis of language can be performed using positron emission tomography (PET) or blood-flow sensitive techniques like functional magnetic resonance imaging (fMRI), or functional transcranial Doppler sonography (fTCD). Neurophysiological imaging methods include electroencephalography (EEG) or magnetoencephalography (MEG). Additionally, the recent advent of transcranial magnetic stimulation (TMS) allows inhibition and even facilitation of language functions in a regionally localized manner (Mottaghy et al., 1999). Most imaging studies have been performed during execution of various language tasks; but few researchers have studied language acquisition using functional imaging (Weiller et al., 1995, Musso et al., 1999, Thulborn et al., 1999, Warburton et al., 1999).

The mechanisms of language acquisition or re-acquisition traditionally have been studied using behavioral approaches—e.g. in the context of aphasia rehabilitation (Belin et al., 1996, Huber et al., 1997, Aftonomos et al., 1999, Pulvermüller et al., 2001). However, investigations involving aphasia patients have been hampered by the high inter-individual variability with respect to medical status, severity and type of language deficits, and the extent of language re-acquisition. Therefore, to our knowledge, no language acquisition training programs have yet been developed which allow variations of the protocol to be monitored using functional imaging.

There are several reasons for a paucity of language (re-) acquisition studies. Animal models are not available in this field. Primary language acquisition is difficult to probe because toddlers do not cooperate consistently during functional imaging. Second language grammar acquisition has recently been studied using functional imaging, but designs are plagued with the difficulty to control prior exposure to the ‘new’ language (Musso, 2001).

Our goal was to develop a model of language learning which: (1) allows to monitor changes in the pattern of brain activation with increasing lexical proficiency using neuroimaging techniques; (2) can also be used to study the effects of pharmacological and TMS interventions on language acquisition; and (3) has relevance to understanding the mechanisms of language reacquisition following stroke with aphasia.

For reasons of simplicity, clinical relevance, and pragmatics, we focused on the semantic component at this stage of our language acquisition model. We designed an implicit learning procedure. Implicit paradigms, i.e. learning occurs without awareness, generally lead to more homogenous performance strategies across subjects and are thus better suited for functional imaging studies than explicit (learning with awareness) paradigms. Additionally, an implicit learning procedure can be used in rehabilitation training programs in cognitively impaired patients because of lower attentional demands. The procedure also closely resembles current views of how children pick first word meanings, that is on the statistical co-occurrence of percept and sound, without explicit training or feedback (Kuhl, 2000, Kuhl, 2001). Even though implicit learning strategies and the general ability to remember words are not specific to language acquisition, both cognitive functions certainly are critical components in child language acquisition and in second language learning. This view was also expressed by Baddeley (1998) who summarizes evidence from developmental and functionals imaging studies that the capacity to memorize nonwords is associated with children's general level of language development and that the phonological store resides within Wernicke's area. Furthermore, recent evidence suggests that some of the processes involved in the acquisition of new words are different from those involved in the acquisition of facts (Waxman and Booth, 2000) because knowledge of a new word, but not of a new fact, is extended to other members of the same category. Further support for domain-specific operations in word acquisition stems from observations of severely amnesic patients who are capable of acquiring new vocabulary words despite their profound memory encoding impairment for other verbal and nonverbal materials (Van der Linden et al., 2001). Thirdly, exposure to isolated words may facilitate word acquisition during childhood, as demonstrated by the recent finding that the frequency of hearing a word in isolation is a better predictor of learning this particular word than the total frequency of exposure to that word (Brent and Siskind, 2001). All three lines of evidence demonstrate that acquisition of a new lexicon shares some of the principles of acquisition of other mnemonic materials, but there is also evidence for unique mechanisms involved in language acquisition.

Section snippets

Stage I

To avoid phenomena like onomatopoeia (the sound of a word is related to its meaning) and the above stated difficulty to control for prior exposure of an existing language, a set of pseudowords was used. The procedure of stimulus generation and selection is depicted in Fig. 1. Pseudowords consisting of four letters were generated with freeware (‘Pseudo’) available in the internet (http://www.nici.kun.nl/∼heuven/tools/index.html). In a first step, 40 native German speakers (14 male; mean age,

Results

The post-training query revealed that none of the subjects were aware of the frequency scheme underlying correct and incorrect pairings. The majority of the subjects felt that their initial classifications (correct–incorrect) subsequently determined which pairings were correct and which were incorrect.

For a percentage of correct responses, a significant interaction of the two factors training session and block emerged (linear/quadratic trends: both F(1,9)>10.40, P≤0.01). To explain the

Discussion

This study demonstrates that subjects learn the meaning of words they are repeatedly exposed to without receiving explicit feedback, i.e. without knowing immediately after each exposure if the pairing between the seen object and the heard word was correct or not. Rather, word learning was accomplished through a simple statistical association between words and objects and without conscious awareness of the scheme underlying correct and incorrect pairings. The resulting learning curves were

Acknowledgements

Research supported by the Innovative Medizinische Forschung of the University of Münster IMF: KN-42 99 21, the Deutsche Forschungsgemeinschaft DFG: Kn 285/ 6-1, the Nachwuchsgruppe des Landes NRW (Nachwuchsgruppe Knecht, 2000 NWG). We are grateful to Sandra Kamping, Andreas Jansen, Julia Bernotat, Bianca Dräger, and Henning Stubbe for help in conducting the study.

References (27)

  • L.B. Aftonomos et al.

    Improving outcomes for persons with aphasia in advanced community-based treatment programs

    Stroke

    (1999)
  • P. Belin et al.

    Recovery from nonfluent aphasia after melodic intonation therapy: a PET study

    Neurology

    (1996)
  • S.D. Goldinger

    Words and voices: episodic traces in spoken word identification and recognition memory

    J. Exp. Psychol. Learn. Mem. Cogn.

    (1996)
  • Cited by (63)

    • Cardiovascular exercise, learning, memory, and cytokines: Results of a ten-week randomized controlled training study in young adults

      2023, Biological Psychology
      Citation Excerpt :

      Participants were asked not to change their habitual physical activity level throughout the study phase, including the period until follow-up assessment. The vocabulary learning task followed the experimental paradigm introduced by Breitenstein and Knecht (2002), which has been shown to induce activity in the hippocampus (Breitenstein et al., 2005). Participants heard pseudowords and saw black-and-white pictures on a computer screen.

    • Quantitative review finds no evidence of cognitive effects in healthy populations from single-session transcranial direct current stimulation (tDCS)

      2015, Brain Stimulation
      Citation Excerpt :

      Individuals are presented with paired images and pseudo-words or words from an unfamiliar language and must learn the pairing of the two. The accuracy with which one can respond to correctly- or incorrectly-joined pairs is thought to be a measure of linguistic learning [9]. Individuals are presented with a series of images (either simple line drawings or photos) and asked to name them as quickly as possible.

    View all citing articles on Scopus
    View full text