Courses

Title: Functional Temporal Neuroimaging: Data Analyses & Hypotheses Testing (Code: BC2A)

Instructor: Profs. Arild Hestvik & S. Bapiraju

details coming soon...

NOTE: This session will run parallelly with Data Analytics & Visualization with R. Participants are encouraged to pick one of these two courses, and attend all sessions of that course.

.

Title: From Mind to Brain: Probing Phonological Theory in the Brain (Code: BC1)

Instructor: Prof. Tobias Scheer

The course is designed as a bridge between phonological theory and neurophysiological experiments: how can the latter provide evidence for or against specific properties of the former? There are countless publications investigating language, or sound, or phonology, with neurophysiological means. But there are only very few people in the world, and a small body of publications, which use neurophysiological evidence to speak to phonological theory. This is because you need to know both phonological theory and experimental work in neurophysiology in order to do so, but people on either side know very little, if anything, about the other. The typical paper based on neurophysiological evidence regarding sound is uninformed of what the issue tested, or the results, mean for phonological theory (usually: nothing). And 98% of phonologists have no idea how neurophysiological experimentation works, what could or could not be tested, what kind of evidence is out there, and anyway do not know how to do an experiment.

The goal of this course is to identify issues in phonological theory that have been or could be probed by neurophysiological evidence (see the overview by Monahan 2018). Experiment design here is crucial: you need to know what you want to test, and whether current neurophysiological instrumentation stands a chance to speak to the issue. The direction is thus Mind → Brain: the origin of the experiment is a property of phonological theory that you want to probe.

The breakdown of individual topics for the five days of the course could be as shown below. After a number of general points are made, we will look at specific cases of neurophysiological evidence speaking to issues in phonological theory. This supposes, in each case, an introduction to the background of the latter.

Days 1+2

1. Mind and Brain

Nobody believes that the brain does not exist: it's material, you can see, touch, smell it. But you cannot see, touch or smell the mind: the number "2", or "fear", or a noun phrase, or a phoneme, have no material existence. This is why materialists believe that they do not exist, or rather, that they reduce to their neural correlates. People who believe that the Mind does not exist are called reductionists (reductionism, eliminative materialism, identity theory: Churchland 1981, Smart 2007). What are the arguments for the Mind being distinct from the brain (in the sense of Descartes' mind-body dichotomy) and a valuable subject of scientific inquiry?

2. The integration challenge

For a long time, and still typically today, the Mind and the Brain were / are investigated separately: biologists and neurologists study the latter, while sociologists, psychologists, linguists etc. inquire on the former. Like in physics where people struggle to make sense of how the infinitely small relates to the infinitely big (Schrödinger's cat), the 21th century task in Cognitive Science is to understand how the Mind relates to the Brain: being a function of the brain, yet distinct from it. This is what philosopher José Luis Bermúdez calls the integration challenge (Bermúdez 2014: 85).

3. Modularity

Brief introduction: converging evidence from generative (inverted T) and psycho-linguistic (= behavioural) quarters.

4. Not the same thing

• language and linguistics is not the same thing

• linguistics and linguistic theory is not the same thing

• sound and phonology is not the same thing

• phonology and phonological theory is not the same thing

5. Granularity

Items of interest in phonological (or linguistic) theory may not be of the same size as items that current neurophysiological instrumentation can identify. If you try to probe items that are too small you will get no result and this can be interpreted as "your item does not exist", when that only means that the instrument is not powerful enough to see it. Poeppel & Embick (2005).

6. Experiment design

• goal: producing evidence for or against a property of some phonological (linguistic) theory.

• thus the starting point is an issue in phonological theory: Mind → Brain

• not the reverse Brain → Mind. This is a common way of doing EEG: you do not have any specific hypothesis, put electrodes on, have a task and then "see what neural activity we get". E.g. "what's the difference in neural activity when young and old people pronounce words?"

7. EEG monoculture

As we will see below, studies relevant for phonological theory are almost exclusively based on one specific brain response pattern (ERP = Event-Related Potential): MMN (Mismatch Negativity). MMN is perception, thus about all studies are in this modality. Production studies are rare in general for independent reasons (noise produced by jaw movement), and virtually absent among studies that speak to phonological theory.

8. Experimental evidence: neurophysiological vs. behavioural

• behavioural (psycho-linguistic studies): what happens between stimulus onset and behavioural response is a black box (just as Behaviourism)

• neurophysiological: you open the black box

• for a given issue, do neural and behavioural evidence converge?

• two case studies:

  • Sahin et al. (2009) (neuro) vs. Scheer & Mathy (2021) (behavioural)

  • how is the brain reacting on increased processing demand? Classical answer provided by psycho-linguistic studies: it takes more time. But by Sahin et al. (2009): the brain heats up.

Day 3

8. From Mind to Brain: topics studied or to be studied

8.1. The phoneme

Kazanina et al. (2006): the phoneme is real in the brain.

8.2. Brain localization of phonological features

• the locus (or loci) in the brain where a given feature sits is not very interesting for phonological theory (except that it provides evidence for the existence of features).

• but grouping of features is: phonological theory distinguishes between Place (e.g. labial vs. palatal), Manner (e.g. stop vs. fricative) and Laryngeal (e.g. voice vs. voiceless) features. They are distinct and independent, for example in a regular feature-geometric tree.

• evidence from brain localization supports this grouping into the three feature classes, which appear to be segregated in the human cortex. Mesgarani et al. (2014), Lawyer & Corina (2014), Arsenault & Buchsbaum (2015), Correia et al. (2015).

• granularity: phonologists agree that Place, Manner, Laryngeal are distinct, but not in which way exactly they are. Are they distinct within the same computational system (module), or do they represent three distinct computational systems (modules)? Scheer (2022).

Day 4

8.3. Underspecification

• introduction to underspecification in phonology

• evidence from MMN in favour of phonological contrast being underspecified: Lahiri et al. (2011, 2013).

• crucial here: the idea that MMN with varying standards probes phonemes, rather than a phonetic memory trace: studies based on the voice-voiceless contrast using VOT. Phillips et al. (2000), Hestvik & Durvasula (2016), Schluter et al. (2017), Monahan (2023).

• this is a debated issue: Rhodes et al. (2019), Han (2022).

• to be tested: predictions made by Laryngeal Realism (Iverson & Salmons 1995): languages with a two-way laryngeal contrast fall into two kinds: voice languages (voicing is an active prime, voicelessness is not phonological) and voicelessness languages (voicelessness is an active prime, voice is not phonological).

Day 5

8.4. How to identify phonological activity: Sahin et al. (2009)

• given the pool of alternations, which ones are phonological and which ones are not?

• other than phonological activity, an alternation may represent: i) allomorphy, ii) lexicalization, iii) phonetics, iv) analogy.

• take velar softening in English: is the alternation of [k] and [s] in electri[k] - electri[s]‑ity phonological in kind? Countless publications debate this question, without converging to a consensus.

• phonologists have no clue whether or not a given alternation is phonological in kind, i.e. involves phonological computation: they are unable to define the input to phonological analysis.

• For some time (in the 70s and 80s), they have tried to find ways to decide, but abandoned that because it didn't go anywhere. They now do whatever they want according to personal or theoretical inclination: some alternations are taken to qualify for phonology, others are not. In SPE (Chomsky & Halle 1968), about all alternations were thought to be phonological in kind. Since then an outsourcing movement placed alternations outside of phonology (reminiscent of syntactic Minimalism): Lexical Phonology, Government Phonology. In the standard incarnation of the latter theory (Kaye 1995), may be 10% of what SPE thought to be phonological is taken to be actual phonology.

• thus competing phonological theories to a large extent are not different because they expose different views on how phonology works, but rather because they do not try to account for the same phenomena. A theory that is designed to account for 10 alternations will be quite different from a theory that is supposed to deal with 100 alternations.

• imagine geologists who want to make a theory of stone but cannot distinguish stone from plastic. They analyze samples containing 10% plastic, or 30% plastic, or 60% plastic, and come up with competing theories. The competition then is due to the plastic, not to theorizing. Phonologists today are exactly in this position.

• Sahin et al. (2009) have set up an experimental protocol (for a different purpose: to show that the inverted T is real) that allows us to identify specifically phonological computation: in speech production, they have found a phonology-specific ERP at ~450 ms.

• if this protocol is controlled for, i.e. if their study can be replicated, it will be enough to run whatever candidate alternation through it: the presence or absence of the ~450 ERP will tell whether or not phonological computation is involved. That would be a revolution for phonological theory.

References & Relevant Literature

• Arsenault, Jessica S. & Bradley R. Buchsbaum 2015. Distributed Neural Representations of Phonological Features during Speech Perception. The Journal of Neuroscience 35: 634-642.

• Bermúdez, José Luis 2014. Cognitive Science. An Introduction to the Science of the Mind. Cambridge: CUP.

• Chomsky, Noam & Morris Halle 1968. The Sound Pattern of English. Cambridge, Mass.: MIT Press.

• Churchland, Paul M. 1981. Eliminative Materialism and the Propositional Attitudes. The Journal of Philosophy 78: 67-90.

• Correia, Joao M., Bernadette M.B. Jansma & Milene Bonte 2015. Decoding Articulatory Features from fMRI Responses in Dorsal Speech Regions. The Journal of Neuroscience 35: 15015-15025.

• Han, Chao 2022. The nature of phoneme representation in various-standard oddball paradigm. Ph.D. dissertation, University of Delaware.

• Hestvik, Arild & Karthik Durvasula 2016. Neurobiological evidence for voicing underspecification in English. Brain and Language 152: 28-43.

• Iverson, Gregory & Joseph Salmons 1995. Aspiration and laryngeal representation in Germanic. Phonology Yearbook 12: 369-396.

• Kaye, Jonathan 1995. Derivations and Interfaces. Frontiers of Phonology, edited by Jacques Durand & Francis Katamba, 289-332. London & New York: Longman. Also in SOAS Working Papers in Linguistics and Phonetics 3, 1993, 90-126. WEB.

• Kazanina, Nina, Colin Phillips & William Idsardi 2006. The influence of meaning on the perception of speech sounds. Proceedings of the National Academy of Sciences 103: 11381-11386.

• Lahiri, Aditi, Sonia A. Cornell & Carsten Eulitz 2011. "What you encode is not necessarily what you store”: Evidence for sparse feature representations from mismatch negativity. Brain Research 1394: 79-89.

• Lahiri, Aditi, Sonia A. Cornell & Carsten Eulitz 2013. Inequality across consonantal contrasts in speech perception: Evidence from Mismatch Negativity. Journal of Experimental Psychology: Human Perception and Performance 39: 757-772.

• Lawyer, Laurel & David Corina 2014. An investigation of place and voice features using fMRI-adaptation. Journal of Neurolinguistics 27: 18-30.

• Mesgarani, Nima, Connie Cheung, Keith Johnson & Edward F. Chang 2014. Phonetic Feature Encoding in Human Superior Temporal Gyrus. Science 343: 1006-1010.

• Monahan, Philip 2018. Phonological Knowledge and Speech Comprehension. Annual Review of Linguistics 4: 21-47.

• Monahan, Philip 2023. Neurophysiological grouping of natural classes and the role of phonological features. Paper presented at PhonolEEGy 2, Amherst, MA, 24 June.

• Phillips, Colin, Thomas Pellathy, Alec Marantz, Elron Yellin, Kenneth Wexler, David Poeppel, Martha McGinnis & Timothy Roberts 2000. Auditory cortex accesses phonological categories: an MEG mismatch study. Journal of Cognitive Neuroscience 12: 1038-1055.

• Poeppel, David & David Embick 2005. Defining the relation between linguistics and neuroscience. Twenty-first century psycholinguistics: Four cornerstones, edited by Anne Cutler, 103-118. Mahwah, NY: Erlbaum.

• Rhodes, R., C. Han & Arild Hestvik 2019. Phonological memory traces do not contain phonetic information. Attention, Perception, and Psychophysics 81: 897-911.

• Sahin, Ned T., Steven Pinker, Sidney S. Cash, Donald Schomer & Eric Halgren 2009. Sequential Processing of Lexical, Grammatical, and Phonological Information Within Broca's Area. Science 326: 445-449.

• Scheer, Tobias 2022. 3xPhonology. Canadian Journal of Linguistics 67: 444-499.

• Scheer, Tobias & Fabien Mathy 2021. Neglected factors bearing on reaction time in language production. Cognitive Science 45, article 13050.

• Schluter, Kevin, Stephen Politzer-Ahles, Meera Al Kaabi & Diogo Almeida 2017. Laryngeal features are phonetically abstract: mismatch negativity evidence from Arabic, English, and Russian. Frontiers in Psychology 8: article 746.

• Smart, John J. C. 2007. The Identity Theory of Mind. Stanford Encyclopedia of Philosophy, available online at http://seop.illc.uva.nl/archives/win2008/entries/mind-identity/.

Title: Data Analytics & Visualization Using R (Code: BC2B)

Instructor: Dr. Naveen Bhatraju

Data analysis is pivotal to research, enabling us to test hypotheses, uncover hidden patterns, and extract meaningful insights. Whether working with primary data (collected by us) or secondary data (leveraging public repositories), the most important goal of data analysis is to advance knowledge. As researchers, our responsibility is to ensure the reliability and reproducibility of our findings. This workshop will equip you with the essential skills to use R, a powerful and versatile programming language, for robust and reproducible data analysis. We will guide you through the process of exploring, cleaning, manipulating, analyzing, and visualizing data using R and RStudio. You will learn how to build reproducible pipelines, ensuring your research is transparent and trustworthy.

Target audience: Beginners with little or no prior experience in data analysis.

Learning objectives: By the end of this session participants would be able to:

•  Navigate the R environment and utilize RStudio to create basic data analysis workflows.

•  Import data from various formats (.txt, .csv, and .xls) into R.

•  Clean and manipulate the data effectively using ‘tidyverse’ r package.

•  Visualize the data using ‘ggplot2’ r package (part of tidyverse).

•  Create sharable and reproducible reports using R markdown.

NOTE: This session will run parallelly with Functional Temporal Neuroimaging: Data Analyses & Hypotheses Testing. Participants are encouraged to pick one of these two courses, and attend all sessions of that course.

Title: Language as a Cognitive System: Reflections on Nature & Mind (Code: DL1)

Instructor: Prof. Pritha Chandra, Linguistics & Cognitive Science, IIT Delhi

Natural language creates strings of infinite length, rearranges words within sentences, and copies features of one word onto another. These properties seem unique to human language, which raises concerns about the evolutionary possibility of language as one of the cognitive organs in the human mind. Research over the last few decades, since the early 1990s, has addressed this question trying to explain these seemingly unique properties of language in terms of ‘third factor principles’: laws that also shape other cognitive and natural systems. In this lecture, I will lead the participants through some of the major challenges that Language as a cognitive system throws at us and how Theoretical Linguists and Cognitive Scientists mitigate these concerns. Through this lecture, I advocate for inter-disciplinary research bringing insights from multiple disciplines, and for a meaningful interaction between the Natural Sciences, Cognitive Sciences and the Humanities. 

Title: Recursion & Natural Language (Code: DL2)

Instructor: Prof. Vaijayanthi Sarma, Linguistics & Cognitive Science, IIT Bombay

One of the defining theoretical ideas in Linguistics is recursion—the ability to embed structures within structures, enabling the infinite generative capacity of natural language. This lecture will provide a comprehensive introduction to recursion as instantiated in human languages, examining its formal properties, cognitive implications, and developmental trajectory.

We will explore key questions, including:

• Linguistic Domains of Recursion: Which levels of grammar (syntax, phonology, semantics) require recursion, and how does it manifest across different linguistic structures?

• Cross-System Comparisons: Which natural cognitive and communication systems exhibit recursion, and which do not? We will contrast recursion in human language with potential analogs in animal communication, mathematical formalisms, and artificial languages.

• The Humboldtian Insight: What is behind Wilhelm von Humboldt’s famous idea of the infinite use of finite means? How does recursion allow for the open-ended productivity and creativity of human language?

• Empirical and Grammatical Insights: How is recursion employed in syntactic and phonological descriptions of language? What evidence from linguistic theory and experimental studies supports its necessity in generative grammar?

• Acquisition and Cognitive Constraints: What is the developmental trajectory of recursive structures in language acquisition? What preferences do children exhibit cross-linguistically when processing and producing recursive utterances? Are some recursive structures more easily acquired than others, and why?

Through theoretical discussion and empirical insights, this lecture will offer a deeper understanding of recursion as a core property of human cognition, illustrating its central role in linguistic theory, cognitive science, and the broader study of the mind.

Title: The Rhythm of Natural Language (Code: DL3)

Instructor: Prof. Paroma Sanyal, Linguistics & Cognitive Science, IIT Delhi

Though speech appears to be a linear string of sound segments, phonological theory has often used hierarchical representational structures to explain the distributional patterns of sound segments in natural languages. The Strong or Weak prosodic positions on such structures explain why units of speech undergo augmentation or reduction processes in certain contexts. These prosodically motivated phonological processes need not be encoded in the lexical representation of natural languages and might not be apparent from the lexical data. Nevertheless, they constitute a critical part of the linguistic knowledge of native speakers allowing them to identify speakers from within and outside their group. In this lecture, we will discuss some of the methods of phonological data elicitation that are used to study this kind of linguistic knowledge. 

Title: Natural Language, Formal Language & Models of Computation (Code: DL4)

Instructor: Prof. T.V.H Prathamesh, Computer Science, Krea University

This session will explore the interface between formal languages, computational models, and natural language, highlighting how insights from formal systems contribute to our understanding of cognition. We will discuss key correspondences between different classes of formal languages and computational machines:

• Finite-state machines and their relationship to regular languages,

• Pushdown automata and their role in context-free languages,

• Turing machines and their connection to decidable languages.

While covering these topics in full mathematical detail is beyond the scope of a single session, we will focus on key intuitions, motivations, and fundamental theorems that shape our understanding of computational models. These formal insights are not only central to theoretical computer science but also inform linguistic theories of syntax and cognition.

The session will also touch on Large Language Models (LLMs), such as those based on Transformer architectures. We will examine their capabilities and limitations through the lens of formal computation and linguistic theory. Drawing on arguments from figures like Noam Chomsky and Emily Bender, we will discuss why LLMs, despite their impressive performance, do not provide a true model of human cognition. Specifically, we will consider:

• The absence of explicit structure and recursion in LLMs,

• Their reliance on statistical pattern recognition rather than rule-based computation,

• Why LLMs struggle with causality, meaning, and compositional generalization.

By bridging insights from formal language theory, computational models, and cognitive science, this session will offer a broad yet foundational perspective on the nature of language, computation, and intelligence.

Title: Foundations of Language in Use & "Meaning" (Code: DL5)

Instructor: Prof. Jooyoung Kim, Cognitive & Brain Sciences, IIT Gandhinagar

This lesson explores semantic and pragmatic aspects of language that reveal how humans conceptualize, integrate, reason, produce, and act upon meaning-making. At the conclusion of this lesson, learners will establish basic theoretical backgrounds in semantics and pragmatics relevant to current trends in cognitive and neurolinguistics.

The lesson begins with a comparison between truth-conditional and cognitive semantics in ways of interpreting linguistic cues such as lexical meaning (semantic representation) and counterfactuals (mental spaces). I also discuss their varied applicability to computational and neural research in linguistic meaning and use. Next, I will move on to pragmatics, to formalize pragmatic processing with basic conceptual and situational components such as (post) Gricean principles, common ground, and the question under discussion. Multimodal symbolic communication, like gestures, will be covered to complete the lesson, in addition to cases of semantic and pragmatic impairment.