ReviewToward a neural theory of language: Old issues and new perspectives
Highlights
► I explore the intersections between research in linguistics and cognitive neuroscience. ► The research in this field results fragmented with no unit of analysis and consensus on the object of study. ► I investigate the main issues, questions, and concerns that prevent the integrated study of language and brain. ► I discuss a feasible way for linguistics to pursue a theory susceptible of neuroscientific testability. ► I propose an interdisciplinary program to achieve a theory capable of predictions on the neurobiological basis of language.
Introduction
Let me begin with two quotations:
Linguistic structure is no less real then speech, and no less amenable to study. Linguistics signs, although essentially psychological, are not abstractions. The associations, ratified by collective agreement, which go to make up the language are realities localized in the brain.
The distinctive features would be more than a universal schema for classifying phonemes in all their diversity across languages; the features would be ‘real’ in the sense of being universal neural mechanisms for producing and for perceiving sounds of speech.
The most relevant point of these statements is the opposition between the terms “psychological”, “abstractions” and “schema” and the terms “real”, “brain”, and “universal neural mechanisms”. I would like to draw attention to this dichotomy, which at first glance, as we will see below, merely describes a terminology-based conflict. The essential point to note for now is that this opposition is based on the idea that language can be investigated in terms of the physiological properties of the brain (i.e., in correlation with the brain mechanisms that form the basis for language processing). This perspective is, of course, endorsed by contemporary scholars in the field of cognitive neuroscience. Crucially, the scholars who stated these phrases are, respectively, the linguists Ferdinand de Saussure (1986 [1916], p. 15) and Roman Jakobson (Jakobson & Waugh, 1979, p. 123).
As a linguist, it is interesting to note that Saussure’s and Jakobson’s works were in some way stimulated by contributions from neuroscience during their lives. Moreover, following the view of Baudouin de Courtenay (1895), who suggested that a study of language pathology may contribute to the understanding of language structure, Jakobson was the first linguist to apply linguistic theory to aphasia research (Jakobson, 1941, Jakobson, 1956, Jakobson, 1964, Jakobson, 1970). The early interest of leading linguists on the neural correlates of language seems symptomatic of the developments that, between the end of the 20th and the first decade of the 21st century, have led linguistics to the promising field of the cognitive neuroscience of language (CNL). Although the entry of linguistics into the field of cognitive neuroscience has been of extreme importance for the potential development of an integrated discipline, this progress has raised some crucial issues and controversies due to a problematic dialogue resulting from the different scientific traditions involved.
The purpose of this article is to explore the intersections between research in linguistics and cognitive neuroscience. After focusing on the main issues, questions, and concerns that limit the joined study of language and the brain, I will propose a possible way to pursue a linguistic theory that is susceptible to neuroscientific testing. Contrary to the brain-centric approach, a possible program of integration will be proposed that enables the testing of assumptions and predictions based on real-time neural constrains that characterize the functional-anatomic organization of language in the brain.
To better clarify the problems raised by this article, we need to briefly outline the field of research and discuss the role of linguistics and other disciplines in cognitive neuroscience. The CNL is a part of the wide multidisciplinary field of cognitive science, which emerged through a long period of gestation between the late 1950s and the 1980s by renewing and further developing very old issues that culminated in the scientific revolution of the 17th century. The cognitive science involves different disciplines, including psychology, computer science, neuroscience, linguistics, philosophy, and anthropology, that converged at a particular stage in their history to study what human cognition is, what it does, and how it works. Indeed, most of the preparatory developments that led to this convergence previously occurred in the computer sciences, psychology, and neuroscience. In the last phases of this gestation, with the birth of generative grammar, Noam Chomsky transformed linguistics by placing it coherently within the field of cognitive science (cf. Bechtel & Graham, 1998). The turning point in this transformation was the rejection of the behaviorist’s account of language (Chomsky, 1959) and the emphasis placed on the assumption that the faculty of language is embedded within the broader architecture of the mind–brain. Chomsky’s incursion into psychology generated a number of controversies that are still ongoing (and we will see that such disputes remain within the relationship between linguistics and cognitive neuroscience). However, the generative grammar framework has had a broad practical impact for the maturation of cognitive science because it initially offered a model that was coherent with a variety disciplines, such as psychology and artificial intelligence, that attempted to answer the question of how mental representation and computation work together.
While disciplines such as psychology and linguistics contributed to the birth of cognitive science only after undergoing internal revolution and artificial intelligence had already been created, 19th century neuroscience first developed the idea that the brain was not merely an organ where mental processes occurred, but a system of integrated components performing different and specific mental functions. The past and future challenges of neuroscience are to divide the brain into its functional components and (a more difficult task) to determine how they work together as a system. Information on the distinct functions performed by the brain could be used to corroborate or guide the development of psychological models of cognitive processes (Bechtel & Graham, 1998, pp. 24–33).
Notwithstanding this advantage, neuroscience was excluded from the constellation of disciplines converging onto cognitive science for a period of time. First, there were pragmatic reasons for this exclusion because the questions asked and the tools used in most neuroscience research were remote from the inquiries being conducted in cognitive psychology. Additionally, the success of artificial intelligence minimized the relevance of neuroscience. The belief was that the relationship between psychology and neuroscience could be a relationship similar to the one that exists between hardware and software. The computer raised the exciting possibility that the mind could be understood almost entirely independently of the brain; if the operations of the mind are akin to the execution of a program, then almost all of the relevant aspects of the mind could be captured by that program, independent of whatever was running it, be it transistors or neurons. The reasons for neglecting the brain were challenged beginning in the 1980s due to the neuroscientist’s interest in the study of cognitive systems, such as vision and memory. Finally, the emergence of several sophisticated neural recording techniques (e.g., PET, fMRI, and MEG) and updated applications of the older EEG (ERPs) and non-invasive techniques to stimulate the cortex (TMS) resulted in the communication and collaboration between neuroscientists and cognitive scientists.
With respect to the CNL, starting with the Broca–Wernicke–Lichtheim classical model (reviewed and modified by Geschwind, 1967), the first decade of the 21st century has seen psychologists, neuroscientists, and linguists make an impressive range of efforts to test, evaluate and investigate the neural correlates of language and to develop new, functional-anatomic models of language and speech (Ben Shalom and Poeppel, 2008, Poeppel and Hickok, 2004). The ‘holy grail’ of language research is thought to be in the brain, and the research undertaken will allow us, for the first time, to solve all of the conundrums that have conflicted scientists for centuries.
By reassembling the pieces of the puzzle, one can legitimately think that in the current state of research, a common program exists by which psychology, neuroscience and linguistics are unified by the aim to jointly investigate the nature of the knowledge of language, the acquisition processes of language, the biological mechanisms that permit the elaboration and the use of such knowledge and the neural computations involved in these processes. In short, it is possible that a sort of neurobiological research program exists that attempts to elaborate a unified model able to make predictions on the innate ability of language and its use in the world from an interdisciplinary perspective. This program could be summarized as follows:
- (a)
A theory of language capable of explaining how listeners convert sounds into meanings by perception and how speakers convert meaning into sounds by production must be consistent with a range of neurophysiological and neuroimaging techniques so that the theory may be verified through empirical data.
Unfortunately, a cohesive correlation between language theory and neuroscientific methods is extremely complex. Currently, this exciting interdisciplinary perspective suffers from unresolved epistemological and methodological questions, and a coherent research program has not yet been formulated. Scholars from the different disciplines involved in the study of CNL separate sounds, words and sentences. Others “[…] investigate not linguistic units, but activities, such as speaking, listening, reading and writing. It is quite difficult to home in on a unit of analysis on which consensus (one that would hopefully reflect understanding) exists” (Grodzinsky, 2003, p. 551). As stated by Grodzinsky, the study of language is unique compared to the study of the visual system. Debates within the field of vision research exist; however, there are no debates concerning the basic unit of analysis. In contrast, within the study of the neural basis of language, little is agreed upon.
Nevertheless, in contrast to vision and other cognitive abilities, we have to consider that language is characterized by both perception processes and the intricate interrelation between perception and production processes. Language is unique in that it combines a set of finite meaningless sound units in a potentially infinite manner, forming increasingly complex units that are given meaning and establishing complex structural relations among these units (this generative nature of language is shared with music and mathematics: see Zatorre et al., 2002, Patel, 2003, Dehaene et al., 1999, Gelman and Butterworth, 2005 for a discussion). Thus, the continuous, varying acoustic waveforms involved in the speech signal can be converted into discrete neural representations through a series of appropriate computations. Crucially, this process is correlated with different types of memory instantiated in the brain. Importantly, following its emergence, language has become a social object, i.e., language occurs pragmatically within a social context, where speakers and listeners implement sets of conversational strategies to contribute to the success of communication (also including gestures, facial expressions, and the direction of an individual’s gaze to amplify or better convey the message involved in the acoustic signal).
However, these almost intuitive assumptions can only partially justify the difficulties in solving the epistemological and methodological issues present within the field of CNL. To the special status of language, one has to add the above-mentioned fact that when different disciplines are integrated to enhance cognitive science, they already have existing models for analyzing cognitive processes involved in language. These toolboxes contain different (and in some cases incompatible) instruments, and the enthusiasm for the enterprise has not been accompanied by an effort to integrate assumptions, models and techniques within a coherent framework. Therefore, the results produced in this area of research seems, in some ways, confused and sterilized.
Section snippets
Two different scientific models
Although Chomsky’s perspective has facilitated and stimulated the birth of cognitive science, this occurred at the cost of the neuroscientific perspective. From its birth, the generative approach to language (specifically the notion of Universal Grammar) theoretically separated the study of linguistic representations from the study of the neurophysiological process producing them. This separation was based on the idea, developed by the scientists of the 17th century (in particular by
Traditional levels of analysis and neural properties of language
First, we must note that during the development of linguistics, the levels of analysis proposed—phonetics, phonology, morphology, lexicon, syntax, semantics and pragmatics—appear to be abstract notions that are not immediately suitable to describe the biology of human language. In fact, their status as autonomous, explanatory elements in mental and cerebral processing is no longer clear. These terms may not reflect the neural properties of language (cf. Lancker Sidtis, 2006, pp. 279–280).
The seduction of neural data and the brain-centric approach
Recently, Weisberg, Keil, Goodstein, Rawson, and Gray (2008) studied whether irrelevant neuroscience information in an explanation of a psychological phenomenon may interfere with an individual’s ability to critically consider the underlying logic of this explanation. Surprisingly, the authors have elegantly demonstrated that naïve adults who are not experts in neuroscience (among which students in a neuroscience course were included) judge that explanations with logically irrelevant
Examining the linguistic brain at work: reallocating the study of linguistics representation and computation within the brain
The necessity of proceeding in this way is more explicit if we reexamine recent neurocognitive data with the aim to provide a more integrated view of the functional brain anatomy of language and suggest hypotheses about functional anatomy and the computations associated with particular brain regions. To accomplish this task, I will concentrate on the crossroads of classical notions of language processing that are strongly established and empirically well supported within linguistic theory:
An integrated model to study language and brain
The functional approach sketched above can be seen as the converse of the brain-centric reductionist approach: In fact localization does not provides per se the explanation of the observed phenomena, because the underlying higher cognitive functions depend on how neural processes are organized. Cognitive functions necessitate and likely emerge from the formation of distributed computational operations in a common format, as a type of meta-representations. Meta-representations have become
Conclusions
In this article, I proposed a possible way to integrate linguistics and cognitive neuroscience, arguing that both disciplines have a great deal to offer to each other. Despite extensive knowledge acquired regarding the physical structures of the brain in relation to language, this knowledge by itself has so far not been able to provide a comprehensive picture of how brain functioning performs the mental processes we use in producing and in understanding language and speech (instead, it has just
Acknowledgments
I am very grateful to Andrea Calabrese for his continuing suggestions during the different phases of the manuscript that permitted me to improve the work. I would also like to thank Sandra Miglietta and Luigina Garrapa for valuable comments on earlier versions of this manuscript. Finally, deep gratitude goes to an anonymous reviewer for his detailed comments and stimulating suggestions of which the article surely benefited. This work was supported by Italian Ministry of Education, University
References (205)
- et al.
Brain responses reveal hardwired detection of native-language rule violations
Neuroscience Letters
(2008) - et al.
Conduction aphasia and the arcuate fasciculus: a reexamination of the Wernicke–Geschwind model
Brain & Language
(1999) From grasp to language: embodied concepts and the challenge of abstraction
Journal of Physiology – Paris
(2008)- et al.
Frontal cortex and the discovery of abstract action rules
Neuron
(2010) - et al.
Neural circuits of hierarchical visuo-spatial sequence processing
Brain Research
(2009) - et al.
Neuropragmatics: brain and communication
Brain & Language
(2000) - et al.
Cortical coordination dynamic and cognition
Trends in Cognitive Sciences
(2001) The role of inferior frontal cortex in phonological processing
Cognitive Science
(2001)- et al.
What you encode is not necessarily what you store: evidence for sparse feature representations from mismatch negativity
Brain Research
(2011) - et al.
Neural systems behind word and concept retrieval
Cognition
(2004)
The motor somatotopy of speech perception
Current Biology
The contribution of the frontal lobe to the perception of speech
Journal of Neurolinguistics
Volumetric vs. surface-based alignment for localization of auditory cortex activation
NeuroImage
Mapping syntax using imaging: problems and prospects for the study of neurolinguistic computation
Sub-centimeter language organization in the human temporal lobe
Brain & Language
Towards a neural basis of auditory sentence processing
Trends in Cognitive Science
Pathways to language: fiber tracts in the human brain
Trends in Cognitive Sciences
Number and language: how are they related?
Trends in Cognitive Sciences
The varieties of naming errors
Cortex
Cortical interactions underlying the production of speech sounds
Journal of Communication Disorders
On Broca, brain, and binding: a new framework
Trends in Cognitive Sciences
The functional neuroanatomy of language
Physics of Life Reviews
Response to Wilson: what does motor cortex contribute to speech perception?
Trends in Cognitive Sciences
Towards a functional neuroanatomy of speech perception
Trends in Cognitive Sciences
Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language
Cognition
Neuropharmacology and linguistic neuroplasticity
Brain & Language
The spatial and temporal signatures of word production components
Cognition
An information theoretical approach to prefrontal executive function
Trends in Cognitive Sciences
Lexicality drives audio-motor transformations in Broca’s area
Brain & Language
Core syntax: A minimalist approach
From monkey-like action recognition to human language: an evolutionary framework for neurolinguistics
Behavioral and Brain Sciences
Toward a model of grammaticality judgments
Journal of Linguistics
Neuropragmatics: a foreword
Italian Journal of Linguistics
On the inseparability of grammar and the lexicon: evidence from acquisition, aphasia, and real-time processing
Language and Cognitive Processes
Versuch einer Theorie phonetischer Alternationen. Ein Kapitel aus der Psychophonetik
A companion to cognitive science
Functional anatomic model of language: assembling the pieces
The Neuroscientist
Analysis by synthesis: a (re-)emerging program of research for language and vision
Biolinguistics
The perception of voice onset time: an fMRI investigation of phonetic category structure
Journal of Cognitive Neuroscience
Functional MRI of language: new approaches to understanding the cortical organization of semantic processing
Annual Review Neuroscience
The ontology of phonology (revised)
Markedness and economy in a derivational model of phonology
Auditory representations and phonological illusions: A linguist's perspective on the neuropsychological bases of speech perception
Journal of Neurolinguistics
Categorical speech representation in human superior temporal gyrus
Nature Neuroscience
Review of skinner’s verbal behavior
Language
Aspects of the theory of syntax
Knowledge of language. Its nature, origin and use
Language and problems of knowledge
The minimalist program
New horizons in the study of language and mind
Cited by (18)
Electrophysiological evidence of phonemotopic representations of vowels in the primary and secondary auditory cortex
2019, CortexCitation Excerpt :We want to address this issue within the neurobiology of language perspective. In doing this, we need to coherently link linguistic primitives together with neurophysiological primitives traditionally assumed to be at the core of the computation and representation of speech sounds (Embick & Poeppel, 2015; Grimaldi, 2012). According to linguistic theory (Halle, 2002; Stevens, 2002), the most relevant representational linguistic primitives are not phonemes, but rather smaller units: i.e., distinctive features.
From Neurons to Language and Speech: An Overview
2023, NeuromethodsNeurocomputational Properties of Speech Sound Perception and Production
2023, NeuromethodsTonotopic mapping of vowel space in the auditory cortex
2023, Language Related ResearchA Critical Perspective on the (Neuro)biological Foundations of Language and Linguistic Cognition
2022, Integrative Psychological and Behavioral Science