In a study published in Frontiers in Language Sciences, Kiran Pala of the University of Eastern Finland, Vasudevan Nedumpozhimana of Trinity College Dublin, and colleagues at the Indian Institute of Technology Ropar introduce a framework they call “fulfillers”—semantic roles that are never spoken aloud but are grammatically, discursively, or culturally implied. Their proposal challenges the foundations of how computational systems and linguistic theories represent meaning, suggesting that what a sentence leaves out may matter as much as what it says.

Semantic role labeling, the process by which linguists and computer scientists assign structured meaning to words, has long relied on frameworks like PropBank and FrameNet. These systems excel at tagging explicit arguments: who did what to whom, where, and when. But they stumble when meaning depends on inference. In pro-drop languages like Hindi, Bangla, and Polish, subjects routinely vanish from sentences because verb endings make them recoverable. In topic-drop languages like Japanese, Korean, and Malayalam, even subjects and topics can disappear entirely, their identity carried by discourse context alone.

Traditional annotation schemes handle these gaps inconsistently at best. A classical system might mark an implicit subject if verb morphology makes it recoverable, but it has no mechanism for capturing the richer layers of meaning that listeners automatically reconstruct—the motivation behind an action, the emotional tone accompanying it, the cultural script that gives it sense. The researchers call this the difference between a bare event with an unfilled slot and the full, situated understanding a human listener actually experiences.

The fulfiller concept, which Pala first proposed in earlier work on cognitive semantics, goes beyond zero anaphora or null arguments. Those constructs are limited to syntactically licensed omissions—cases where grammar itself predicts what is missing. Fulfillers encompass something broader: inferences grounded in cultural scripts, affective states, and shared experiential knowledge. When an English speaker hears “she went to the store,” they don’t just recover the subject; they implicitly understand the routine of shopping, perhaps the mode of transport, the purpose of the errand. None of this is stated, yet all of it shapes interpretation.

To test whether fulfillers could be systematically identified rather than merely intuited, the team built a multilingual corpus of 194 sentences spanning English, Hindi, Malayalam, Bangla, and Bagri—a western Indo-Aryan language spoken in Rajasthan, Haryana, and Punjab. The corpus was deliberately asymmetric: English and Hindi were sampled at larger scale to support contrasts between idiomatic and pro-drop constructions, while the other languages served as targeted typological points of comparison. Two trained annotators, supervised by a semanticist, worked through the data using a layered annotation framework that distinguishes grammatical, discourse-pragmatic, and cultural-conceptual fulfillers.

The annotation process was rigorous but acknowledged its inherent subjectivity. Annotators first worked independently on a shared subset, resolving disagreements through discussion and achieving roughly 85 percent agreement on whether a fulfiller was present. Formal inter-annotator agreement metrics like Krippendorff’s alpha were deferred to a later phase, with the initial priority being guideline refinement rather than numerical scoring. The researchers also proposed complementary approaches—probabilistic annotation, confidence scores, and task-based evaluation—as ways to handle the framework’s inherently interpretive nature.

The corpus revealed a striking pattern. In pro-drop and topic-drop constructions, discourse-pragmatic fulfillers dominated: missing agents were recovered almost entirely through grammar and situational context, with culture playing little or no role. In idiomatic expressions, the picture reversed. Cultural and experiential fulfillers took over, drawing on shared knowledge that no amount of discourse tracking could supply. The Hindi idiom “Lakshman rekha langhana”—crossing the Lakshman line—derives its meaning not from verb agreement or context but from a mythological script rooted in the Ramayana. The Malayalam expression “kollam kandavanu illam venda”—one who has seen Kollam needs no home—relies on the historical salience of Kollam as a wealthy trading port.

This split has real consequences for how we understand linguistic typology. The existing literature treats referential density—how much of an argument a language leaves for context to supply—as a single property carried mainly by verb morphology. But the researchers found that the same language can behave as strongly context-dependent in everyday clauses while drawing on entirely different cultural resources in its idiomatic layer. A typology built only around the grammatical axis gets Hindi’s pro-drop behavior right and has nothing to say about its idioms. The two dimensions, grammatical and cultural, coexist within a single language rather than reducing to one parameter.

The implications extend beyond theory into practical applications. For machine translation, a fulfiller-annotated resource could flag expressions where literal translation loses cultural content—precisely the cases where current systems fail silently. The Bagri idiom “ganv-ganv gogo, ganv-ganv khejdi”—in every village the deity, in every village the khejri tree—means almost nothing to an English speaker, yet a literal rendering would score well on fluency metrics while losing nearly all its meaning. By marking fulfillers at the annotation stage, such failures become testable rather than invisible. The researchers acknowledge limitations: their corpus is regionally focused on Indo-Aryan and Dravidian languages, and they call for future work testing the framework on typologically distinct languages like Finnish and Turkish. But their core argument stands: absence is not a gap in language but a structured, inferable part of it—one that our models of meaning have been missing all along.

Subject of Research: A cognitive-semantic framework for annotating implicit, culturally and grammatically inferred meaning in multilingual language data.

Article Title: Semantic labeling with fulfillers in multilingual contexts

Article References: Pala, K., Nedumpozhimana, V., Shalu, S., Silu, P., & Choudhary, K. K. (2026). Semantic labeling with fulfillers in multilingual contexts. Frontiers in Language Sciences, 5, Article 1688982. https://doi.org/10.3389/flang.2026.1688982

Image Credits: AI Generated

DOI: 10.3389/flang.2026.1688982

Keywords: semantics, linguistic typology, pro-drop languages, idioms, semantic role labeling, cognitive semantics, multilingual corpora, machine translation, annotation framework, cross-linguistic variation, implicit meaning, natural language processing

The researchers recorded the activity of hundreds of individual hippocampal neurons in patients who had electrodes implanted for clinical reasons, typically to localize seizure activity before epilepsy surgery. While the patients listened to narrative speech, the team captured the firing patterns of these neurons with millisecond precision. This rare opportunity to eavesdrop on the living human brain at single-cell resolution allowed the investigators to ask a deceptively simple question: when a person hears a word, what do hippocampal neurons actually do with its meaning?

To answer it, the team turned to an encoding model approach, a technique in which computational models of language are used to predict neural responses. Crucially, the researchers controlled for lower-level features of the speech signal, including the phonemic content of words and their grammatical roles. Even after stripping away these confounds, semantic information remained robustly encoded in the population activity. In other words, the hippocampal neurons were not merely responding to sounds or syntactic structure; they were representing what the words meant.

One of the study’s most intriguing findings concerns the selectivity of individual neurons. Rather than acting as specialized detectors for single words or narrow categories, hippocampal neurons displayed complex, mixed selectivity, responding to multiple words drawn from multiple semantic categories. This kind of distributed, high-dimensional coding has been championed in the population doctrine of cognitive neuroscience, which holds that complex cognitive content is carried not by individual cells but by patterns of activity spread across many neurons. The hippocampal semantic code fits this doctrine well: meaning emerges from the joint pattern of firing across the population, much as a chord carries information that no single note contains.

The parallels with large language models run deeper still. In models such as Word2Vec, BERT and GPT-2, words are represented as vectors in a high-dimensional space, where the distance between vectors reflects semantic similarity. The researchers found that the distance between neural population responses in the hippocampus correlated with semantic distance computed from these embedding spaces. Words with similar meanings evoked population activity patterns that were closer together in the neural space, while unrelated words produced more distant patterns. The geometry of meaning, it seems, is not unique to machines trained on oceans of text; the human hippocampus appears to organize word meanings along comparable geometric lines.

Yet the neural code was not a simple copy of any single model. Among the embedding schemes tested, neural population activity aligned most closely with GPT-2, a contextual transformer model that represents words differently depending on the surrounding sentence. This alignment suggests that hippocampal semantic representations are heavily contextualized: the same word does not always elicit the same neural pattern, but rather a pattern shaped by the linguistic moment in which it appears. Supporting this interpretation, variation in the neural response patterns across occurrences of a word correlated with polysemy measures derived from language models, meaning that words with many context-dependent senses produced more variable neural representations than words with a single, stable meaning.

The study also uncovered a subtle and technically fascinating phenomenon the authors call contrastive coding. For pairs of semantically similar words, the relationship between semantic distance and neural distance inverted: instead of nearby meanings producing nearby neural patterns, the hippocampus pushed similar representations apart. The researchers attribute this counterintuitive pattern to the noise-mitigating benefits of contrastive coding, a principle familiar from machine learning systems that learn by pushing similar examples apart to make them more distinguishable. In the brain, where neural signals are inherently noisy, deliberately separating representations of similar concepts could prevent them from being confused with one another. This echoes earlier findings that overlapping spatial memories in the hippocampus trigger representational repulsion, suggesting the structure may apply a similar differentiation strategy to meaning as it does to space and memory.

These results reshape how neuroscientists think about the hippocampus itself. For decades, the structure has been studied primarily as the engine of episodic memory, the faculty that binds the who, what, where and when of personal experience. But a growing body of evidence, including prior work showing that hippocampal neurons reactivate conceptual representations when pronouns refer back to earlier nouns, and that the region supports the flexible use of language in conversation, points to a broader role. The new findings suggest that representing word meaning during comprehension is not a peripheral sideline for the hippocampus but a core computational function, one that may provide the semantic scaffolding upon which narrative memories are built.

The methodological rigor of the study strengthens its conclusions. The team analyzed responses across thousands of words, validated their models against shuffled controls, and demonstrated that semantic embeddings explained significantly more variance in neural activity than phonetic features alone. They also showed that the semantic-neural distance relationship held across individual patients and that removing function words and pronouns did not eliminate the effect. The data and analysis code have been made publicly available, allowing other researchers to scrutinize and extend the findings, a transparency increasingly expected in high-stakes human neuroscience.

The broader implications reach toward the frontier where neuroscience and artificial intelligence converge. If the human brain and large language models have independently arrived at similar solutions for representing meaning, then the vector-space approach that underpins modern AI may capture something fundamental about how meaning must be organized to be useful. Conversely, the brain’s contrastive tricks and contextual flexibility could inspire the next generation of models. For now, the study delivers a vivid answer to an ancient question: when you hear a word, its meaning does not live in a single neuron or a single brain region, but in a shimmering, high-dimensional pattern distributed across the hippocampal population, a pattern whose geometry looks remarkably like the embeddings of a machine that learned to speak.

Subject of Research: Semantic encoding of spoken words by hippocampal neuron populations in humans

Article Title: A population code for semantics in human hippocampus

Article References: Franch, M., Mickiewicz, E. A., Belanger, J. L., Joiner, B., Katlowitz, K. A., Zhu, H., Chavez, A. G., Chericoni, A., Paulo, D. L., Goldman, A. M., Krishnan, V., Maheshwari, A., Bartoli, E., Kemmer, S., Piantadosi, S. T., Sheth, S. A., Provenza, N. R., Hayden, B. Y., & Hennig, J. A. (2026). A population code for semantics in human hippocampus. Nature Neuroscience. https://doi.org/10.1038/s41593-026-02436-4

Image Credits: AI Generated

DOI: 10.1038/s41593-026-02436-4

Keywords: hippocampus, semantics, population coding, single-neuron recordings, large language models, GPT-2, word embeddings, language comprehension, polysemy, contrastive coding, neural encoding, human neuroscience