scholarly journals Statistical language learning: computational, maturational, and linguistic constraints

2016 ◽  
Vol 8 (3) ◽  
pp. 447-461 ◽  
Author(s):  
ELISSA L. NEWPORT

abstractOur research on statistical language learning shows that infants, young children, and adults can compute, online and with remarkable speed, how consistently sounds co-occur, how frequently words occur in similar contexts, and the like, and can utilize these statistics to find candidate words in a speech stream, discover grammatical categories, and acquire simple syntactic structure in miniature languages. However, statistical learning is not merely learning the patterns presented in the input. When their input is inconsistent, children sharpen these statistics and produce a more systematic language than the one to which they are exposed. When input languages inconsistently violate tendencies that are widespread in human languages, learners shift these languages to be more aligned with language universals, and children do so much more than adults. These processes explain why children acquire language (and other patterns) more effectively than adults, and also may explain how systematic language structures emerge in communities where usages are varied and inconsistent. Most especially, they suggest that usage-based learning approaches must account for differences between adults and children in how usage properties are acquired, and must also account for substantial changes made by adult and child learners in how input usage properties are represented during learning.

2021 ◽  
Author(s):  
Sara Finley ◽  
Elissa Newport

While most morphemes in the world’s language involve continuous structure or concatenation (e.g., prefixes and suffixes), many languages show some form of non-adjacent, non-concatenative morphology. Non-concatenative morphology poses a challenge for statistical learning approaches to morpheme segmentation because the combinatorial possibilities greatly increase for non-adjacent dependencies. The present study explores the types of dependencies that human learners (school-aged children and adults) are able to extract from exposure to a miniature, artificial non-concatenative system. In Experiment 1, participants were exposed to 12 CCC ‘roots’ that fit into 72 CVCVC skeletons with a high variety of VV ‘residue’. Experiment 2 extended Experiment 1 to school-aged children (with adult controls). Experiment 3 replicated Experiment 1, but with ‘mixed’ consonant-vowel roots and residues. Across all three experiments, participants were able to recognize familiar items compared to novel items, but had limited ability to generalize the CCC roots to novel items, suggesting a limited ability to parse consonantal roots. Adults were better at generalizing to novel items compared to children.


ReCALL ◽  
2004 ◽  
Vol 16 (1) ◽  
pp. 145-157 ◽  
Author(s):  
CRISTINA ROS i SOLÉ ◽  
RAQUEL MARDOMINGO

This paper discusses a framework for designing online tasks that capitalizes on the possibilities that the Internet and the Web offer for language learning. To present such a framework, we draw from constructivist theories (Brooks and Brooks, 1993) and their application to educational technology (Newby, Stepich, Lehman and Russell, 1996; Jonassen, Mayes and McAleese, 1993); second language learning and learning autonomy (Benson and Voller, 1997); and distance education (Race, 1989; White, 1999). On the one hand our model balances the requirements of the need for control and learning autonomy by the independent language learner; and on the other, the possibilities that online task-based learning offer for new reading processes by taking into account new literacy models (Schetzer and Warschauer, 2000), and the effect that the new media have on students’ knowledge construction and understanding of texts. We explain how this model works in the design of reading tasks within the specific distance learning context of the Open University, UK. Trayectorias is a tool that consists of an open problem-solving Web-quest and provides students with ‘scaffolding’ that guides their navigation around the Web whilst modelling learning approaches and new learning paradigms triggered by the medium. We then discuss a small-scale trial with a cohort of students (n = 23). This trial had a double purpose: (a) to evaluate to what extent the writing task fulfilled the investigators’ intentions; and (b) to obtain some information about the students’ perceptions of the task.


2017 ◽  
Vol 26 (5) ◽  
pp. 451-457 ◽  
Author(s):  
Lucy C. Erickson ◽  
Rochelle S. Newman

The goal of this review is to provide a high-level, selected overview of the consequences of background noise on health, perception, cognition, and learning during early development, with a specific focus on how noise may impair speech comprehension and language learning (e.g., via masking). Although much of the existing literature has focused on adults, research shows that infants and young children are relatively disadvantaged at listening in noise. Consequently, a major goal is to consider how background noise may affect young children, who must learn and develop language in noisy environments despite being simultaneously less equipped to do so.


2020 ◽  
pp. 219-239
Author(s):  
Iris Berent

Woody Allen has famously said he is not afraid of dying; he just doesn’t want to be there when it happens. It’s no wonder his words struck a chord—“not being” is a scary proposition. Yet many Americans believe that their psyches will persist after the demise of their bodies. And it’s not only religious devotees who believe in the afterlife; young children say the same, and so do adults and children in other societies, including even those who are self-described “extinctivists.” Our afterlife beliefs, however, are remarkably inconsistent. On the one hand, we state that some aspects of our minds are immaterial, inasmuch as they survive our bodies. But on the other hand, we believe that some of these seemingly immaterial properties of the dead act like matter; for example, they are contagious, much like germs or excrement. Chapter 14 considers our views of what happens once we are no more. We will see that the collision between Dualism and Essentialism—the twin forces that stir up our misconceptions about our origins—are also responsible for these mistaken beliefs about our demise.


1952 ◽  
Vol 33 (2) ◽  
pp. 61-66 ◽  
Author(s):  
Lionel C. Lane

The one factor that is characteristic of these case illustrations is our acceptance of greater responsibility. It is quite clear that in our practice we have to begin to move out more firmly to certain clients. We accept the parent's right to make his own decision, but we question the wisdom of having him base it solely on his impulse to resist help and not include his wish to help his child. The community gives authority to the family agency. This is not the authority of the court or the protective agency, but the more subtle one that goes with professional skill and knowledge of the dynamics of human behavior. We are still, often, too hesitant about accepting a role that is essentially a preventive one. The limits imposed on us by the client's right to make his own decision about using help is a valid and essential part of the process. One could, however, pursue this idea to its ultimate end and maintain that a family has a right to pursue its own path to self-destruction. But when the lives of young children are affected, we cannot limit ourselves to such a passive role. To do so would mean that the problems of children are not our concern until the point of crisis and emergency has been reached, a development that we have encouraged by our passivity. There is a need for more concerted and aggressive efforts on our part so that we are more truly offering a preventive service to the community. A client's decision to use help will be influenced, we believe, by the degree of firmness and persistence with which we approach him. Furthermore, we need not fear that we shall force our help on the client too much, since this is a manifest impossibility anyway. A lack of conviction in offering service, as evidenced by our willingness to withdraw at the first rebuff, can only leave the parent more comfortable in continuing in an unhealthy situation. We need to be more aggressive in our approach, so that if we finally have to accept a client's refusal of our help, we are sure we have used, to the fullest extent, every available means to demonstrate our ability and desire to be of help to him.


2021 ◽  
Vol 10 ◽  
Author(s):  
Iris Broedelet ◽  
Paul Boersma ◽  
Judith Rispens

Since Saffran, Aslin and Newport (1996) showed that infants were sensitive to transitional probabilities between syllables after being exposed to a few minutes of fluent speech, there has been ample research on statistical learning. Word segmentation studies usually test learning by making use of “offline methods” such as forced-choice tasks. However, cognitive factors besides statistical learning possibly influence performance on those tasks. The goal of the present study was to improve a method for measuring word segmentation online. Click sounds were added to the speech stream, both between words and within words. Stronger expectations for the next syllable within words as opposed to between words were expected to result in slower detection of clicks within words, revealing sensitivity to word boundaries. Unexpectedly, we did not find evidence for learning in multiple groups of adults and child participants. We discuss possible methodological factors that could have influenced our results.


Author(s):  
Toni Cunillera ◽  
Estela Càmara ◽  
Matti Laine ◽  
Antoni Rodríguez-Fornells

Can even a handful of newly learned words help to find further word candidates in a novel spoken language? This study shows that the statistical segmentation of words from speech stream by adults is facilitated by the presence of known words in the stream. This facilitatory effect is immediate as the known words were acquired only minutes before the onset of the speech stream. Our results demonstrate an interplay between top-down lexical segmentation and bottom-up statistical learning, in line with infant research suggesting that integration of multiple cues facilitates early language learning. The ability to simultaneously benefit from both types of word segmentation cues appears to be present through adulthood and can thus contribute to second language learning.


2011 ◽  
Vol 14 (1) ◽  
pp. 39-57 ◽  
Author(s):  
Rebecca Treiman ◽  
Brett Kessler

Writing systems are usually studied in terms of the level of language that they represent, with little attention to the shapes that are used to do so. Those shapes are not random or accidental, however. They tend to be similar to one another within a script. Many of the Latin letters have a roughly vertical stem or hasta with an appendage or coda to the right. This arrangement is more common than one with the coda on the left of the hasta. We present data to show that young children are generally better at copying and writing from memory shapes such as <b> and <F>, which have the typical arrangement with the coda on the right, than those such as <d> and <J>, which do not. The results suggest that children start to learn about the statistics of the letter shapes before they know how or that these shapes represent language. Keywords: letter shapes; letters; statistical learning; Latin alphabet; reversal; left-right orientation; directionality; hasta-coda-structure


Author(s):  
Ana Franco ◽  
Julia Eberlen ◽  
Arnaud Destrebecqz ◽  
Axel Cleeremans ◽  
Julie Bertels

Abstract. The Rapid Serial Visual Presentation procedure is a method widely used in visual perception research. In this paper we propose an adaptation of this method which can be used with auditory material and enables assessment of statistical learning in speech segmentation. Adult participants were exposed to an artificial speech stream composed of statistically defined trisyllabic nonsense words. They were subsequently instructed to perform a detection task in a Rapid Serial Auditory Presentation (RSAP) stream in which they had to detect a syllable in a short speech stream. Results showed that reaction times varied as a function of the statistical predictability of the syllable: second and third syllables of each word were responded to faster than first syllables. This result suggests that the RSAP procedure provides a reliable and sensitive indirect measure of auditory statistical learning.


2019 ◽  
Vol 3 (2) ◽  
Author(s):  
Rick De Graaff

In this epilogue, I take a teaching practice and teacher education perspective on complexity in Instructed Second Language Acquisition. I take the stance that it is essential to understand if and how linguistic complexity relates to learning challenges, what the implications are for language pedagogy, and how this challenges the role of the teacher. Research shows that differences in task complexity may lead to differences in linguistic complexity in language learners’ speech or writing. Different tasks (e.g. descriptive vs narrative) and different modes (oral vs written) may lead to different types and levels of complexity in language use. On the one hand, this is a challenge for language assessment, as complexity in language performance may be affected by task characteristics. On the other hand, it is an opportunity for language teaching: using a diversity of tasks, modes and text types may evoke and stretch lexically and syntactically complex language use. I maintain that it is essential for teachers to understand that it is at least as important to aim for development in complexity as it is to aim for development in accuracy. Namely, that ‘errors’ in language learning are part of the deal: complex tasks lead to complex language use, including lexical and syntactical errors, but they are a necessary prerequisite for language development.


Sign in / Sign up

Export Citation Format

Share Document