The Evolutionary Origin of Predication and Syntax

1. Introduction

This article addresses the origin of syntax and therefore presents a hypothesis about language evolution. As is well known, the only way to make progress in the study of language evolution is to formulate hypotheses and attempt to evaluate them using every resource available to us.

Predication and syntax are features of exclusively human communication. The question of how they emerged in evolution —or, more precisely, in gene–culture coevolution— may therefore also shed light on other uniquely human characteristics that can be linked in one way or another with fully developed language. These links cannot be explored here, although I will refer to some of them at various points.

The proposal advanced in this article focuses on vocal communication. Our present-day language, and very probably language at its origins as well, is certainly multimodal: we communicate not only through words but also through gestures. Nevertheless (in Section 2), I will explain my reasons for concentrating on the auditory modality.

Vocal communication is particularly well suited to the functions of requesting something or calling someone over. These functions, especially calling, often occur when the distance between producer and addressee is greater than is suitable for purely gestural communication. Thus, I propose (in Section 3) that, before the emergence of predication, language consisted only of pre-words used exclusively for requesting something or calling someone over, pre-words that were always used alone in each message, and with the same intonation every time. In this connection, I will examine the causal relationship between language and concepts and attempt to reverse the traditional view of that relationship.

According to the hypothesis, predication evolutionarily emerged because of a particular dialogical relation (Section 4). A chance hearer of a call addressed to someone who was absent, or the addressee of a request for something that was unavailable, might have been puzzled by what they had just heard. This would lead that person to reject, correct, or update the mental content they had just heard expressed by the other individual. Predication thus emerged, together with a syntax in an as-yet ungrammaticalized form.

The emergence of predication subsequently required—although, in evolutionary terms, not much later—another major change (Section 5). Pre-words did not possess the two differentiated and mutually detachable planes of phonetic articulation and intonation. A genuine word, by contrast, must be perceived as the same word both when it requests something or calls someone over and when it functions as a component of an intonational pattern encompassing at least two words. This required some other element of the sound to remain identical while intonation changed. In other words, articulatory-phonetic sequences had to be kept invariant by being learned and imitated with complete fidelity and precision.

Section 6 examines the relationship between Theory-of-Mind and the dialogical reply that constitutes the origin of predication.

In Section 7, the hypothesis turns to the predications used in the last communicative function of language to emerge: speaking to oneself.

In Section 8, we will return to the emergence of the two differentiated and mutually detachable planes of phonetic articulation and intonation. The need for this differentiation gave rise to the well-known fact that the two planes of word sound are processed in different cerebral hemispheres, although they must immediately be unified. Considering this fact within the framework of the hypothesis may prove informative. More specifically, with the aid of paleogeneticists, it might allow us to derive a tentative inference (or ‘hypothetical deduction’) concerning when syntax emerged.

Section 9 seeks to ground our hopes for the long-term outcomes of research on language evolution and, more broadly, evolutionary general anthropology.

2. Multimodality

We communicate not only through words but also through manual, facial, and whole-body gestures. When interacting with someone who does not know our language, we may even use pantomimes resembling children’s symbolic play. Human communication today therefore clearly operates in both the auditory and visual modalities, and it is highly plausible that this was also true at its origins. The hypothesis developed here will nevertheless focus on predominantly vocal communication.

I will begin, however, by acknowledging that highly distinguished scholars maintain that “gesture supports the emergence of communication systems”. The major figure Michael Corballis (2002) must be cited here, as must the group from Nicolaus Copernicus University (Poland) whose work focuses on pantomime ¹.

What I find particularly important in this respect is that humans share ostensive and emotional gestures with the great apes. Indeed, among adult humans, purely gestural predicative communication most often takes the following form. First, the producer establishes eye contact with the addressee; next, the producer points to and looks at—or merely looks at—an object in the environment; finally, the producer looks back at the addressee and accompanies this gaze with a gesture expressing an emotion. We all know how to produce and understand purely gestural communication equivalent to “This speaker is a bore” or “This feast looks appetizing”. These emotional gestures are admittedly highly semiotized ². More specifically, they are semiotized to a degree perhaps exceeding the maximum attainable by chimpanzees. Even if this difference is accepted, however, an evolutionary continuity with the signs of great apes remains, one that at first sight appears to favor a gestural origin of human language.

Since the 1990s, it has also been argued that gestural modality may provide further clues to the origin of language. This claim draws attention to emerging sign languages, which are progressively created when a group of deaf people live together over a period of years (see Senghas & Coppola 2001; Senghas et al. 2004). As Sandler et al. (2022) state, these languages “are the only human languages that can emerge de novo at any time. (…) Spoken creole languages are also young, but are different from emerging sign languages, in that the speakers of pidgins from which creoles are assumed to have descended already had native languages.”

Despite all this, I remain convinced that the vocal modality (or, more precisely, the predominantly vocal one) may be more fruitful for studying language evolution. Several considerations lead me to focus the hypothesis on the auditory-vocal modality:

• Attention to both child development and pragmatics has led us to recognize the great importance of intonation. This is the most important of all these considerations.
• The historical development of grammaticalization has proceeded only through the vocal channel.
• Among people without hearing or vocal impairments, purely gestural communication is restricted to specific contexts in which the aim is for the message to reach only the addressee. I return to the example “This speaker is a bore”. Purely vocal communication, by contrast, is always entirely possible.

Finally, multimodality is now, and surely always has been, functionally differentiated. Certain communicative functions are much better suited to one modality than to the other. Messages used to ask for something and –even more so– to call someone over, typically involve a distance between producer and addressee that exceeds what is appropriate for purely gestural communication. Such communication would therefore have relied primarily on vocal modality.

3. The Type of Communication Immediately Preceding the Emergence of Predication and Syntax

At the outset, then, the vocal modality would have been used to ask for something and to call someone over, and it would also have relied on intonation. Even today, both as producers and as recipients, we continue to use intonation as the distinctive cue for differentiating calls, requests or commands, predications, and questions.

What the hypothesis proposed in this article emphasizes above all is that, unlike our words, these vocal signs

• would have been used only for the communicative functions of requesting a particular object or calling a particular individual over,
• would always have occurred one at a time—that is, always in holophrastic messages,
• and would always have been produced with the same intonation.
• In other words, these signs would not have been genuine words, but pre-words ³.

In human language, by contrast, the word ‘mama’ may of course be used as a call and the word ‘water’ as a request, but these words can also

• fulfil many different syntactic roles,
• be pronounced with different intonational patterns or as different parts of an intonational pattern,
• and, most importantly, be combined with other words. Genuine words are therefore always ‘words for syntax’. Moreover, as syntax became increasingly grammaticalized, words came to be assigned to one or another “part of speech” or “word class,” and, in this way, they came to require one another reciprocally—especially nouns and verbs.

We have not yet considered a feature of pre-words that is particularly important to the hypothesis and therefore requires closer attention. Pre-words did not designate any referent; they served only to perform concrete acts of requesting an object or calling someone over. These signs certainly had what we might call ‘concrete worldly connections,’ enabling them to call a particular individual over or request a specific object. Those worldly connections, however, did not in any way turn them into referential signs to which a predicate could be added.

A further step is needed to characterize this phase of vocal pre-words. Words –especially words with meanings that can be evoked, such as nouns, verbs, and adjectives– have commonly been thought to derive from concepts formed by the mind independently of communication. The most explicit and elaborate version of this idea is ‘mentalese’ (Fodor 1975). Even without going that far, most scholars of language evolution regard concepts as the origin and model of content words.

I propose reversing this relationship. In Bejarano 2014 (a review of Hurford 2010) I criticized above all his acceptance of the conventional view. I argue that we never think of the multiple features contained in a perception as independent steps that jointly compose a higher-order unit. Perceptions simultaneously activate an enormous array of features. Size, distance, movement, action, direction, and so forth are just as constitutive of the perceptual unit as recognition of the ‘wolf’ or ‘apple’ pattern, and none of these features is apprehended in perception independently of the others. Indeed, progressively assembling perception would be not merely useless but counterproductive. Consider what would happen if the perception of a tiger had successively to add the features ‘large’, ‘coming toward me,’ and ‘rapidly.’ It therefore seems more likely that the concept, as a meaning that can be thought independently, emerged only after—and as a consequence of—the ‘semantics for syntax,’ or semantics of ‘parts of speech,’ created by human language.

This proposal –namely, language creates concepts that can be thought independently– closely resembles the ‘label-feedback hypothesis’. Lupyan 2012 defines it as follows: “Labels act as implicit category markers, highlighting commonalities among items that share a label”. Recently, he has demonstrated the power of his hypothesis even more convincingly. He did so primarily by focusing on the present: “This conception of language not only helps explain the effects of language on reasoning, categorization, and perception, but also the recent astonishing advances in artificial intelligence in which exposure to natural language transforms general-purpose neural networks into human-like minds”: Lupyan 2026. However, working within the well-established field of experimental semiotics, this author projects his hypothesis back onto the origins of language somewhat less explicitly than I would have preferred.

In summary, I propose that the original use of the vocal modality produced not genuine words, but instruments inextricably tied to a particular communicative goal and necessarily accompanied by their particular intonation. This is how they would have been learned by each generation. Detaching these signifiers from their function and intonation must therefore have been difficult.

4. How Could the Difficulties Surrounding the Original Emergence of Predication Be Overcome?

What has been said above about the enormous difference between pre-words and words seems to lead us into a dead end. Yet we know that access to vocal predication was not only possible but spectacularly successful. It should be recalled that, among people without vocal or hearing impairments, communication based solely on gesture is confined to very few occasions and highly specific contexts. We must therefore explain how, according to the hypothesis, genuine words and syntax emerged.

The most plausible possibility is that the addressee of a request, or the recipient of a call addressed to someone, was puzzled by what they had just heard. Today we might express this puzzlement by saying “How can this person be addressing a calling to someone who left several days ago?” or “How can this person ask me to bring something we no longer have?”. The recipient would then reply. Let us consider an example.

In Bejarano 2008, I described the following scene ⁴. “An adult and a child are playing with wooden blocks to build an increasingly tall tower. The adult then asks the child, who is holding the box of blocks, ‘Give me more blocks! ‘More!’ At that moment, the child, still holding the box, sees that it is empty and says in Spanish: ‘Más (= more) no.’”

This is, of course, very different from what the evolutionary origin would have been: the child had been surrounded by genuine words for more than two years. It can nevertheless serve as an example of the solution I am proposing. Whereas my “Más” was a message with conative intonation and function, the “Más” constituting the first part of the child’s reply was not conative at all; it served only to specify the point to which the second part applied. The second part, “no,” far from being a refusal to do what I had requested—that is, far from being the rejection-negation that the child had been using long before then—provided information about the object of the request. More specifically, it conveyed information still unknown to the addressee. A further feature must also be considered. ‘Más’ and ‘no’ can certainly often appear as the sole word in a message. The child’s reply, however, was a composition in which the two meanings formed a single intonational pattern.

In the preceding paragraph, I referred to the addressee’s ‘lack of knowledge’. This point, I propose, is crucial to the origin of predication and must be developed further. The thema (or, later, with the development of grammaticalization, the subject of the verb, as in ‘The blocks have run out’) does not correspond to the speaker’s own belief about the referent. Rather, it corresponds to the false, incomplete, or outdated belief that, in the speaker’s view, the addressee holds about that referent. The apprehension of false belief –the emblem of Theory of Mind– is central to my proposal concerning the origin of syntax; here, however (unlike in most Theory-of-Mind research ⁵) that belief has been heard. (Note, please, that it is not necessary to linguistically assert the predication corresponding to a belief for the hearer to become informed of that belief.) The rhema or predicate, for its part, must be understood as the feature that corrects, completes, or updates the addressee’s belief. I have defended this view of predication throughout my research career using various approaches. Chronologically, the first of these was a critical analysis of Frege’s “Sense and Reference”.

The hypothesis concerning the original emergence of predication has now been presented, but it can also be described in other terms. Repetition by another person of the conative, presyntactic message would have stripped that message—that pre-word—of its force and self-sufficiency. This would have constituted an extraordinary instance of ‘bleaching,’ a transformation of a magnitude never subsequently repeated in the languages for which the grammatical technical term ‘bleaching’ was coined ^6. In this original bleaching, the genuine word emerged from the pre-word, and syntactic composition emerged with it.

I wish to emphasize that both ideas employed in this proposal—first, the clash between one individual’s mental content and that of another, and second, the emergence of predication in dialogical reply—lead us to regard dialogue as the cause that brought syntax into being. Proposals of this kind are now becoming increasingly widespread. Carpendale et al. (2024), for example, explain children’s achievements not in terms of an internal capacity but “as a distributed accomplishment emerging within shared practices.” My focus is primarily evolutionary rather than developmental, and the achievement that concerns me is the genesis of syntax (and, simultaneously, of concepts, as discussed above). Despite these differences in focus, I am pleased to observe that the general trend has now changed almost completely.

5. The New Requirement Generated by the Emergence of Predication

A genuine word must be perceived as the same word both when it serves as a call addressed to a particular individual or a request for something specific and when it functions as part of an intonational pattern encompassing at least two words. This required some other element of the sound to remain identical while intonation changed. The sound of genuine words therefore had to acquire two differentiated and mutually detachable planes: intonation and the articulatory-phonetic sequence. In other words, articulatory-phonetic sequences had to be kept invariant by being imitated with complete fidelity and precision.

This type of imitation was an evolutionary novelty. To appreciate its significance, we may begin by considering the research conducted at Stockholm University over approximately the past decade on the high neural cost of both representing and imitating motor sequences (see, e.g., Lind & Jon-And 2025). These types of imitation and learning were probably difficult to acquire. By contrast, earlier forms of cultural learning, although they could involve several steps, were assisted at each stage by the affordances presented by the object to which the task was applied. We can now turn to something—mentioned in Jon-And et al. 2025—that is more specific and directly relevant here, namely, imitation of the articulatory-phonetic sequence. In this type of imitation, it is clear that no such assistance provided by the object’s affordances is available; everything depends on the ability to represent the sequence.

This extremely demanding imitation of motor sequences has been undoubtedly applied to other tasks. But I propose that it first emerged in language. Note that, according to the hypothesis, the detachment of the articulatory-phonetic sequence from intonation —that is, the outcome of precise articulatory-phonetic imitation— provided the enormous adaptive advantage of making syntax less costly ⁷.

To explain more clearly how important, according to the hypothesis, the precise imitation of articulatory-phonetic sequences was, we can establish the following causal chain.

• One, the earliest vocal signs lacked this precise imitation of articulatory-phonetic sequences.
• Two, without such imitation, detaching intonation from the signs was virtually impossible.
• Three, without this detachment, even after the evolutionary emergence of predication and syntax, understanding them remained difficult.
• Four, following the evolutionary emergence of syntax, the precise imitation of articulatory-phonetic sequences must have evolved relatively quickly on an evolutionary timescale.

The distribution of present-day language across the two cerebral hemispheres should also be noted ⁸. In most individuals, the production and reception of articulatory-phonetic sequences are the exclusive responsibility of the left hemisphere. The right hemisphere, by contrast, is invariably involved in both emotional prosody and linguistic prosody, the latter serving to disambiguate speech acts and syntax. (Given this division of prosody, the term ‘intonation,’ formerly used to refer to linguistic prosody, could be dispensed with. I will nevertheless sometimes retain the older term here for convenient reference to the intonational pattern.)

The separation of the two major elements that constitute present-day linguistic signs clearly allows each to be processed with great efficiency. It is equally clear, however, that this new cerebral distribution must have represented a major change.

Before proceeding to the next section, I will comment on two points. Is there anything resembling a pre-word in present-day language? My answer is that there is nothing genuinely comparable. The only possible candidates that occur to me are purely expressive cries prompted by extreme emotion. In such cries—unlike linguistic interjections—the articulatory-phonetic sequence cannot be separated from intonation because there is no articulatory-phonetic sequence. In this respect, they bear some resemblance to pre-words. Yet there is a fundamental difference between these cries and ancient pre-words: cries are purely expressive and therefore lack communicative intentionality, whereas pre-words were genuine communicative instruments.

More generally, no pre-word could have survived as a “living fossil” ⁹. In light of the hypothesis proposed in this article, the following qualification can be made. Pre-words were used only to call someone over or to request something. By contrast, when we learn, for example, a person’s proper name, we assign it a referential meaning, whether in a narrated world or in the real world. Likewise, we learn words together with the other words with which they have been combined. In this way, we acquire them as what linguistics call “parts of speech” ¹⁰. Pre-words, however, could not have constituted ‘parts of speech’. In short, according to the hypothesis, living fossils can survive only from periods in which grammaticalization was still weak, not from the very origin of predication and syntax.

The other point concerns Gázquez et al. (2026), who propose that “hemispheric specialisation for language—especially production—can exert pressure on praxis pathways and manual skill asymmetry, more strongly than the reverse”. I find this issue interesting. However, rather than comparing two influences operating in opposite directions, we might suggest that development and evolution involve a spiral of steps caused alternately by hemispheric specialization and by manual asymmetry ¹¹.

6. The Relationship Between Theory-of-Mind and the Origin of Predication

In Section 4, we saw that, according to the hypothesis, the apprehension of false belief (the emblem of Theory-of-Mind) is central to the original emergence of predication and syntax. More specifically, the dialogical reply that gave rise to the predicative communicative function resulted from reply-producer having heard a call addressed to someone who was absent or a request for an item that was no longer available. Certainly, Theory-of-Mind encompasses other highly diverse domains. The domain of interest here, however, may be the most neurally demanding.

After being puzzled by such messages, the producer of the impending dialogical reply simultaneously maintains two different mental contents concerning the same entity. Other Theory-of-Mind activities, by contrast, are based merely on imagining oneself in another’s circumstances. This may occur in the form practiced by animals—probably through vicarious expectations (Bejarano 2025, subsection 3.2)—or in the highly elaborate form of the 'addiction to narratives’ (see Sperber 1985 / Sterelny 2001). In these activities, the subject is momentarily identified with the individual to whom attention is directed; the costly duality concerning the same entity is therefore unnecessary.

How could the simultaneous duality of these contents –different from one another yet referring to one and the same entity– have originally emerged? I propose that this costly duality requires the subject engaged in Theory-of-Mind activity to be entirely unable, at that moment, to identify with the individual attended to and whose mental state the subject is apprehending. I further propose that this requirement is fulfilled in the hypothesis about the origin of predication presented above: Individual A cannot identify with Individual B while Individual B is speaking to Individual A.

7. Is There Any Way to Estimate When Predication and Syntax Originated?

The hypothesis presented above does not specify the evolutionary period in which predication and syntax emerged. It may nevertheless allow us to draw an inference that offers a tentative clue. For the moment, however, let us begin by focusing on the evolutionary emergence of an intonational pattern encompassing at least two words.

The function of such an intonational pattern corresponds perfectly to that of the appearance of the human eye. Hearing this intonational pattern and observing the producer’s eye movement serve to bind tightly together the different elements of syntactic predication within each modality. In gestural predications, such as the example equivalent to “This speaker is a bore”, the conspicuous horizontal movement of the iris across the broad white sclera leads the addressee to connect the producer’s gaze toward the object with the combined emotional gesture and gaze directed toward him –toward the addressee. This helps the addressee understand the emotional gesture as the predicate that the producer wishes to communicate about the indicated object ¹². Analogously, in the vocal modality, the shared intonation turns the first meaning—the word “more” in the example described above—into the basis to which the predicate “no,” which corrects or updates it, is applied. Both resources help the recipient achieve syntactic integration.

We argued above, in Section 5, that the evolutionary emergence of vocal predication subsequently required —although, in evolutionary terms, not much later — an important change: the processing of intonation encompassing two words—that is, the intonation whose function we have just emphasized—and the processing of the articulatory-phonetic sequence constituting each word had to become separate, because each operated on different units. For both processes to remain efficient, they would have had to be located in different cerebral hemispheres; therefore, the immediate integration of their outputs would also have become indispensable. This probably required a major strengthening of the corpus callosum connecting the two hemispheres. Friederici et al. (2002) and Sammler et al. (2010) show that speech comprehension requires syntax and prosody to be reintegrated through interhemispheric interaction via the corpus callosum.

This provides the tentative clue—the inference derived from the hypothesis—that we were seeking. The gene or genes responsible for the larger size of the human corpus callosum relative to that of other hominins may perhaps be identified in present-day humans. In that case, our question about when predication emerged could then be reformulated as follows: assuming that this gene is identified in present-day humans, would paleogeneticists find it in extinct Homo species? And if so, in exactly which extinct species?

8. Concluding Remarks: A Reflection on Interdisciplinarity

Evolutionary general anthropology should study all characteristically human capacities together. These are numerous: language, reasoning, the capacity for moral choice in the most demanding sense, an attraction to fiction, imagination, curiosity capable of identifying problems to be solved, the sense of humor, etc. Such inquiry should also approach these capacities from every relevant scientific field. Put differently, in this domain we would do well to follow the maxim: “If you want to solve a puzzle, you must work with all the pieces”. Unfortunately, this conviction, which many of us ultimately share, is difficult to put into practice.

The difficulties are enormous. No one can acquire more than a superficial and ultimately useless familiarity with every field involved. Teams bringing together specialists from several related fields can certainly be highly productive. At present, however, a project aimed at assembling specialists from every relevant field would probably achieve little. Genuine interdisciplinarity culminates in a synthesis, not merely in an aggregate of studies on different topics.

These remarks may appear pessimistic, but they are not so in the long term. Indeed, although for now we must accept that the ideal strategy is not yet feasible, I am confident that this will change. Researchers will then have access to a far greater body of findings on which to draw. Moreover, future AI will provide an external memory of everything preserved from what has been said or writte; and, much more importantly, the capacity to identify, through associations across this immense body of human ‘semantic archives’, some element that might “make sense in relation to the very small fraction (of possible actions) that lead to successful discoveries” (Yaman et al. 2016), or, in other words, some element that might fit the still-empty profile of the solution we happen to be seeking at that moment. Embracing this long-term hope, however, requires us—despite our humble recognition of our limitations—to work with enthusiasm in the present.