Thank you, Elisa, Amod, and Ethan, for giving me the opportunity to share my thoughts on your fantastic blog! I’m a researcher in the field of South Asian Studies working with Vedic and Sanskrit texts. Since writing my dissertation on the Gāyatrī-Mantra, my research has focused on mantras. In 2022, I was involved in the organization of the workshop “Mantras: Sound, Materiality, and the Body” (see the report). At that workshop, my colleague Georgi Krastev made some remarks on the problems that arise when “mantra” is conceptualized as a category with discrete boundaries. These remarks have had a great impact on my thinking about mantras.

Over the last few years, I have developed an approach to the subject that has helped me deal with a number of questions that have haunted mantra studies for a long time:

What are mantras? Are mantras language? How do they work?

These questions (seemingly) require long, complex, and nuanced answers, and the academic literature one should engage with and refer to in order to answer them is vast. In this blog post, I won’t even attempt to provide such answers. What I offer here are just my own thoughts and ideas, not elaborate scholarly theories. But perhaps, even in this unrefined form, they can help others who are grappling with the fundamental questions of mantra studies – not by providing answers to these questions, but by offering an alternative approach to them.

What are mantras?

Defining the concept “mantra” is notoriously difficult – or rather, famously impossible. Jan Gonda (“The Indian Mantra,” Oriens 1963: 246) confidently stated that

How could one identify such a “definite semantic kernel”? In fact, throughout history, the word “mantra” (and its cognates, derivatives, and translations) has been used to denote such a wide variety of phenomena that finding the lowest common denominator is hard. One might think of the stereotypical mantra as a brief, fixed Sanskrit formula that is attributed a special religious or magical power. However, among mantras, there is considerable variation in length, modifiability, and language. Not everything that is called a mantra is brief, fixed, and immutable; many mantras are composed in languages other than Sanskrit, and many practitioners would deny that the mantras they use have anything to do with religion or magic.

Moreover, the Vedic/Sanskrit term mántra itself has variants and translations in many languages, e.g., Tamil mantiram, English “mantra” (historically also “muntru” and “mantram”), Tibetan sngags, Chinese 真言 zhēn yán, and so forth. It is obvious that these terms cannot be synonymous in the sense that all of them have identical denotations and connotations.

There exists no systematic and comprehensive study of what exactly terms such as mántra, mantiram, or “mantra” have encompassed (and what not) in various cultures and traditions in different historical periods, or how they have related to similar terms (vidyā, dhāraṇī, maṅgala, etc.). In the absence of such data, defining “mantra” as one lexeme, let alone as one category with discrete boundaries, is an unattainable goal. While most will agree that a typical mantra is somehow connected to speech and sound and that it is supposed to achieve a certain effect, this can hardly be called “a definite semantic kernel.”

Hard-pressed as they are to provide answers, scholars and educators often define “mantra” on the basis of whatever kinds of mantras they happen to know, and in doing so emphasize aspects and associations that happen to be prevalent in the language they are writing in. In English, for example, “mantra” is strongly associated with repetition, fixedness, religion, and spirituality. For many practitioners and scholars, repetitivity is a key feature of mantras (cf. my forthcoming article “New Directions in Mantra Studies: Exploring the Emergence of Mantra Repetition,” ARGOS *2025).

In Early Vedic, in contrast, mántra denoted the product of the poets – usually entire hymns, and not just brief formulas. These hymns were certainly recited repeatedly in the process of memorization, but not in performance (moreover, when composing hymns, the Ṛgvedic poets even avoided repeating material from other hymns verbatim). Thus, the two concepts “mantra” and mántra are just as different from each other as, for example, “poem” and “spell.”

How should we deal with such contradictions?

Let us imagine an ideal scenario, one in which we can approach reality without any preconceived notions and labels, more or less like an infant with the cognitive abilities of an adult. Approaching reality as something entirely new, one could group the elements on the basis of how similar they are to each other, ignoring what they are traditionally called or what they are usually associated with. Only after this process is completed, one could think about drawing a circle around some of them, thus delimiting categories of things with more or less fixed boundaries. One could then think about labeling and using them for heuristic purposes, while always remaining aware that their boundaries are completely arbitrary. In this way, one could arrive at a model of reality that is as truthful, useful, and adaptable as possible.

It is of course completely unrealistic to pursue such an approach.

Our experience of reality is always shaped by preconceptions, and updating categories or getting rid of useless ones often fails because of our own stubbornness and inertia as well as that of our fellow human beings. All we can do is try our best.

As the example above already illustrates, the idea that mantra is a category that can be used in any scholarly discourse without defining it properly is dubious. Arguably, there exists no one “semantic kernel” – thus no one category “mantra” – that we can somehow discover or reconstruct by studying mantras from various times and traditions.

For this reason, we should drop the idea of such a category, thus starting from scratch (insofar as this is possible). Instead of trying to find out what the essence of mantras is, it might be more advantageous to identify recurrent aspects and properties and gather entities characterized by them in order to be able to make meaningful comparisons. In doing so, it makes sense to start with entities that have been labeled or treated as a “mantra” (or mántra, sngags, etc.) – if only for heuristic reasons. Looking at such entities, there are indeed a number of such aspects and properties that recur so often that they deserve closer study. As I will show below, in the case of entities labeled “mantras” (etc.), this can be their association with language and the principle according to which they are supposed to work.

If we do not step back far enough from what seems to be one incomprehensible and multifaceted phenomenon, we run the risk of misinterpreting reality as well as misunderstanding each other. Especially in the case of “mantras,” we should always assume that the notion we are dealing with differs from person to person and from source to source. In many conversations with knowledgeable colleagues I have been confronted with the idea that the interlocutor and I (and potentially even others) know what we are talking about, and that all that remains to be discussed are facets and aspects of one and the same phenomenon. In fact, there are several phenomena, some of which happen to share the same label. As fascinating as it may be, the idea of a single category “mantra” should be dropped as soon as we start speaking about mantras.

Are mantras language?

The answer to this question depends on how mantras are defined. For the sake of argument, let us define language as a system whose primary function is to convey information and to make those who have the power of interpreting language think, feel, or do something specific (i.e., they are so-called speech acts). The primary form of language in the case of humans is verbal, that is, it is characterized by sequences of orally produced sounds. For language to work, there needs to be at least one being capable of producing and/or interpreting it in a manner that sufficiently conforms to its grammatical rules. The typical case is when two or more people converse in the same language with the goal of conveying information, but language is also used when individuals speak to themselves or read or write texts – all of which can be in their native language or another language, correctly or with mistakes. In my view, the crucial aspect in all these cases is that language is meant to mediate between intelligent agents.

As I have argued, the notion that all “mantras” are a linguistic category is untenable, because they do not form one category at all. Most entities called “mantra” are indeed associated with language: most of them clearly have a verbal form, and many convey information and/or perform an action. However, not all of them seem to work in this way, which is why some doubt has been cast on their status as language proper. In particular, their efficacy is often not believed to depend on their being understood, even if their content is perfectly comprehensible and perhaps even appropriate in the respective communication situation. Paradoxically, mantras are generally even supposed to be superior to normal language. How can this be explained?

How do mantras work?

When looking at religious texts (in the broadest sense) from all over the world, one recurring aspect stands out, namely what I propose to call the “superlinguistic principle.” This principle consists in the idea or belief that a text – that is, a fixed verbal sequence – may achieve a certain effect through a mechanism that differs in some way from, or surpasses, that of normal speech. While other manifestations of language such as prayers or incantations are intended as a means of communication and presuppose that there is a conscious being who receives the message (e.g., a deity), certain texts are supposed to become effective regardless of whether they are understood by anyone. They are “superlinguistic” in that they do not mediate between intelligent agents, as normal language does.

Exactly how the superlinguistic principle works varies from case to case. For example, the user of a text may believe that the language it is composed in has a special power, that there is a paranormal or magical connection between the words of the formula and the objects they refer to, or that the sounds of the formula create vibrations that affect the mind or soul of the reciter (or, to give another example, the molecules in an object). We can easily imagine cases where a persuasive person or text is able to make someone believe in such a potency by providing a plausible theory.

However, I would argue that belief in the superlinguistic principle in many cases precedes explanations or can be entertained without them. It is possible to believe in the superlinguistic power of a certain text without knowing how it works; such belief is not necessarily dependent on intricate theories. Nor does it necessarily depend on the form, function, or content of the text. It is, in a way, a paratextual element.

This does not mean that textual content and superlinguistic potency never interact. In fact, the content of a text may even strengthen the belief of its user in its efficacy, and the belief in its efficacy may inspire a user to think about its content. The origin, language, form, and recitation style can be supporting factors, too; in South Asia, for example, a Sanskrit formula that has a certain structure (e.g., om + the name of a deity) is much more likely to be considered superlinguistic. But even in an environment where there are ideas about what a superlinguistic text should look like, superlinguistic power may also ascribed to texts that don’t conform to these ideas (we may, for example, imagine a traditional Hindu priest who believes in the power of a curse spoken in Arabic).

I can think of various motivations for developing a belief in the superlinguistic principle. One of them is the desire to avoid the uncertainties and risk of communication. When speaking with – or to – a conscious being, there is always a chance that this being will not respond or react in the way one wishes. This applies to communication with human beings (as everybody knows) just as much as it applies to communication with deities, even if this exchange is, in fact, unilateral.

If I ask a colleague to turn down the heater, there is a chance I’ll be turned down, whereas if I do it myself, I can at least expect the room temperature to go down (if environmental conditions don’t prevent it). Similarly, a deity may decide to fulfil or not fulfil the wishes for which a devotee has prayed. In other words, when applying the superlinguistic principle, you are not trying to convince, persuade, or cajole your interlocutor to be benevolent and act in accordance with your intentions or wishes. Rather, you expect your words and sounds to become effective through a different, more powerful and secure, mechanism.

Another reason to resort to the superlinguistic principle is because your goal cannot possibly be achieved through normal communication, be it the cure from a physical ailment, the attainment of superhuman powers, a better reincarnation after death, and so forth. Such goals may, at best, be granted by a superhuman being, but as mentioned above, communication with superhuman beings is not without risks. Moreover, depending on the religion, there might not be a superhuman being that is willing to fulfil all kinds of wishes, because the wish may be unethical or against other divinely-sanctioned regulations (e.g., wishing for the death of a relative, obtaining superhuman powers or forbidden substances, etc.).

The presence or absence of belief in the superlinguistic principle is a factor that may help in collecting related or comparable entities, and one could even define “mantra” as a text (especially if it is comparatively brief) that is believed to function according to the superlinguistic principle. If we adopt this perspective, we will begin to find superlinguistic, mantra-like texts in many regions and cultures. All over the world, there are formulas, spells, prayers, and so forth that are believed to be superior to normal speech. The belief in the superlinguistic principle seems to be at least as widespread as the belief in superhuman beings.

Collecting and comparing superlinguistic texts not only promises to lead to interesting insights, but will also reveal that the impression that all “mantras” taken together form a unique cultural phenomenon is an illusion. While many individual mantras, mantra systems, and mantra theories are indeed unique and worth studying, superlinguistic texts are a universal phenomenon. Their development appears to be a reaction to one of the numerous limitations that characterize human existence, namely the inadequacy of ordinary language. Superlinguistic texts are supposed to fulfil the “the dream of a natural language,” as Robert Yelle put it (Explaining Mantras, Routledge: 2003, p. 2); they speak to the desire of humans to maximize their agency.

About Dominik A. Haas

Dominik A. Haas is the author of a monograph on the history of the Gāyatrī-Mantra as well as a translation and text-linguistic study of the Kaṭha-Upaniṣad. He is currently a member of the FWF-funded Cluster of Excellence “EurAsian Transformations” at the Austrian Academy of Sciences and an external collaborator of ERC-funded synergy project “MANTRAMS” (Vienna, Tübingen, Oxford).