A month or two ago, while writing Crown of Crows, I suddenly and unexpectedly found myself in desperate need of a language for dragons. Not really having time to come up with one completely from scratch, I asked myself if there were some quick way I could create something that sounded reptilian and exotic, but also grounded in linguistic theory and historical linguistics.
You can read a quick summary of the result here on Patreon (and the full details, for those who want to create their own Draconic words, is behind the paywall). Here I offer another perspective, looking at how Draconic might have influenced various languages throughout human history.
I should emphasize here that, while all the details of history and linguistics below are true, dragons do not, in fact, exist; and any evidence that they do, or of their influence on human language and society, is completely made up and false.
I hope.
When Dragons Shaped Human Speech
Abstract: Ancient Celtic, Arabic, and Japanese all lost the sound /p/ through remarkably similar processes. Proto-Indo-European developed an unusually elaborate three-way velar distinction. Medieval bestiaries described dragons as real creatures with specific habitats—Ethiopia, India, and remote cave systems—until the 18th century. The Greek word for dragon, drákōn, is derived from a root meaning “to see.” These facts, taken together, suggest the obvious question: What if dragons’ phonological preferences shaped human languages through sustained contact?
It has now been two decades since archaeologists Ted Amicus and Adam Chiang, comparing fossils and other remains (bones of prey, treasure, and other middens) gathered from the high, arid mountains of Tibet, Iran, Macedonia, and eastern British Columbia, presented compelling evidence that dragons (or draconis volaris), flying reptilian megafauna, actually existed, and may in fact have survived long into the historical era. This astonishing result was comparable to the discovery of the actual giant squid, the Komodo dragon, or the gorilla, all of which were once thought to be mythical. But partially reconstructed skeletons now stand in the British Museum and Smithsonian, and the shocked public is gradually coming round to the realization that these creatures of myth and nightmare once truly ruled our skies.
It is not known for certain if dragons were intelligent or used human speech, as they do in our myths and stories. However, intricate patterns of scratch marks in the caves where bones and other remains were found are highly suggestive of symbolic writing, and a partial translation of the inscriptions at Donggar provided by W. Chao of Beijing is compelling. Contemporary medieval accounts suggest that the dragon(s) carried on extensive negotiations with nearby human populations on the matter of resource usage (cattle, sheep, gold, virgins, etc.) and pacts of nonaggression.
This article presents a speculative but linguistically grounded hypothesis: that dragons, as historical creatures rather than pure mythology, influenced human speech patterns through substrate and contact effects during the ancient and medieval periods. The evidence lies in cross-linguistic patterns of labial consonant loss, enriched velar and uvular systems, and geographic correlations between dragon legend concentration and distinctive phonological features. If true, the linguistic and historical record preserves traces of a vanished but extensive social web between human and dragon.
Labial Erasure: Coincidence?
The most striking pattern emerges in labial consonants, particularly /p/. Three major language families underwent nearly identical processes of /p/ loss during periods when dragon belief was strongest, a fact viewed as coincidence by conventional analyses.
Celtic languages provide the clearest timeline. All Celtic languages lost Proto-Indo-European *p during the Proto-Celtic stage, approximately 1300-800 BCE. The phonological trajectory followed predictable lenition: *p > *pʰ (aspiration) > *ɸ (voiceless bilabial fricative) > *h > Ø (complete deletion). By the time Celtic languages diverged into distinct branches, /p/ had vanished entirely from inherited vocabulary. The famous P-Celtic versus Q-Celtic distinction reflects not *p retention but *p replacement: P-Celtic languages filled the phonological gap by shifting Proto-Celtic *kʷ (from PIE labiovelars) into a new /p/, while Q-Celtic languages retained *kʷ as /k/, leaving them with no /p/ phoneme for centuries.
It is worth noting, as elaborated elsewhere, that Celtic languages derived from another language with a major labial gap: Proto-Indo-European itself. PIE famously had no *b phoneme. The Celtic daughter languages filled in this gap, converting both *bʰ and *gʷ to *b, but the instability of labial consonants and their slipperiness (despite the ease and non-markedness of these sounds) is quite striking.
Arabic underwent a parallel transformation. Proto-Semitic *p became /f/ (voiceless labiodental fricative) in Arabic, Ethiopian Semitic, and Modern South Arabian languages through the same lenition mechanism: bilabial stop > labiodental fricative. Though precise dating remains elusive, this shift predates the earliest Arabic inscriptions (4th-6th century CE) and likely occurred during the prehistoric Proto-Arabic period. The change is exceptionless in native vocabulary: Proto-Semitic *p-ūm “mouth” became Arabic fam, following the pattern across all inherited roots. (Compare also “Palestine” vs. “Philistines”, “Persia” vs. “Farsi”.) Significantly, this shift occurred in precisely those Semitic languages spoken in regions with documented dragon traditions (Arabia, Ethiopia, Southern Arabia) while Northwest Semitic languages (Hebrew, Aramaic, Ugaritic), spoken north of the dry mountainous desert regions, retained /p/.
Japanese provides a third example. Old Japanese possessed *p, but during the transition to Early Middle Japanese, *p lenited to /ɸ/ (voiceless bilabial fricative), then to /h/ in most environments. The chronology spans roughly 700-1200 CE, overlapping with Japan’s adoption and elaboration of Chinese dragon mythology. Intervocalic /ɸ/ voiced to /w/ and often disappeared entirely, following the Celtic trajectory. Modern Japanese retains /p/ only in preserved geminates (historically *pp) and loanwords; native words show /h/ where etymological *p once stood.
The standard explanation invokes a coincidental independent development. But the geographic and temporal correlations deserve scrutiny. These changes cluster in regions and periods with documented dragon traditions, often near the arid and / or remote, high-altitude regions where dragon remains have been found. Celtic (Pannonian, Alpine, Apennine) Europe during the late Bronze Age, the Arabian Peninsula and Horn of Africa in late antiquity, and medieval Japan all represent epicenters of dragon belief when measured by textual density and artistic representation.
Velar vs Alveolar Consonantal Inventories
While labial consonants weakened or vanished, languages in dragon-associated regions often developed unusually rich inventories of velar and uvular consonants—sounds produced at the back of the vocal tract. This is an interesting parallel development that demands explanation.
While Draconic may have had an extensive sub-velar phonemic inventory, it is more likely (based on contemporary accounts) that they had a rich set of dental, alveolar, and palatal coarticulated consonants. With a prehensile forked tongue, dragons could pronounce a wide variety of combinations of these consonants at the same time, a feat no human mouth could match. How, then, would humans have pronounced words with these sounds?
Here we propose that humans mapped the Draconic horizontal coarticulation of consonants onto the vertical human vocal tract, supplementing dental and alveolar consonants with velar, uvular, glottal, and other subvelar consonants as needed.
Arabic phonology, for example, displays remarkable elaboration of back consonants. Beyond simple /k/, Arabic contrasts voiceless uvular stop /q/, uvular fricatives /χ/ and /ʁ/ (dialectal), voiceless pharyngeal fricative /ħ/, and voiced pharyngeal fricative /ʕ/. The emphatic (pharyngealized) consonants /sˤ/, /dˤ/, /tˤ/, /ðˤ/ involve secondary articulation retracting the tongue toward the pharynx. This back-loaded phonology creates a sound system dominated by velar-uvular-pharyngeal articulations. It is proposed that this extraordinarily rich sub-velar inventory was developed to approximate the rich super-velar (alveolar) phonemic distinctions available to dragons.
Proto-Indo-European presents an even more striking case with its reconstructed three-way velar distinction. Traditional PIE phonology posits palatovelars (*ḱ, *ǵ, *ǵʰ), plain velars (*k, *g, *gʰ), and labiovelars (*kʷ, *gʷ, *gʷʰ): nine dorsal consonants contrasting only by secondary articulation. Modern theories suggest the “plain velars” may have been uvular, giving PIE a palatal-velar-uvular three-way contrast unparalleled in most language families. Why would the proto-language of Indo-European develop such elaborate dorsal articulation? The Draconic influence hypothesis suggests contact with a community that had a consonantal inventory that could only be approximated by full use of the human throat.
The Caucasus region, legendary home to numerous dragon-guarding tales in Georgian and Armenian tradition, hosts languages with the world’s most extreme velar/uvular inventories. Ubykh (extinct 1992) possessed twenty distinct uvular consonants within a total inventory of 84 consonants—the largest non-click consonant system documented. Northwest Caucasian languages generally show similar patterns: minimal vowels (often just two), elaborate consonant systems, and maximal exploitation of the velar-uvular-pharyngeal axis. And the Caucasus is an arid mountainous region postulated to be prime dragon habitat.
Etymology as Cultural Memory
The Greek word drákōn preserves what may be linguistic memory of dragon-human contact. Derived from the verb dérkomai “to see clearly,” through the aorist stem drak-, the noun drákōn functions as an agentive noun: “the seeing one,” “the watcher,” “the deadly-eyed.” This etymology is by far the most probable, yet its implications remain unexplored. Why would Greeks name a creature for its vision?
Medieval bestiaries provide context. Dragons were described as cave-dwelling ambush predators, guardians of treasures and sacred sites. They are said to have unblinking stares and supernatural watchfulness. Beyond that, however, they are noted for their remarkable cleverness and insight, renowned for riddles, wordplay, and uncanny observations of the character of the humans that challenged them. (It is possible that the animal known as the “sphinx” was originally a dragon.) Their nigh-immortal lifetimes and flawless memories naturally added to their legendary wisdom. And the mapping between “vision” and “insight” is an ancient metaphor found in tales, poetry, and myth worldwide.
The PIE root *derḱ- “to see” appears across Indo-European branches: Sanskrit dṛś- “to see,” Old Irish derc “eye,” Welsh drych “aspect,” Germanic turhtaz “bright/clear” (semantic shift: “seeing clearly” → “bright”). Only Greek seems to have maintained the root as the name a creature, suggesting that Greek-speaking populations had direct observational experience requiring the vocabulary item. The semantic connection between vision and dragon implies that exceptional sight (and insight) was the dragon’s most salient characteristic in human encounters. And indeed there is some evidence that *derḱ is derived from the dragons’ own word for themselves.
Effect of Species Contact
Languages frequently absorb features from nearby populations speaking different languages during contact periods. One very clear case involves the Dravidian substrate in Indo-Aryan. Sanskrit developed retroflex consonants (ṭ, ḍ, ṇ) absent from other Indo-European branches, matching the elaborate coronal systems of Dravidian languages indigenous to India. Approximately 88 Vedic Sanskrit words show “unconditioned retroflexes” (retroflex consonants appearing in positions inexplicable by Sanskrit’s internal phonology), suggesting loanwords or substrate influence from retroflex-using populations. If human language contact leaves such clear phonological traces, contact with a non-human but speaking species would likely create similar observable patterns.
The labial-loss pattern fits contact-induced change profiles. Small articulatory adjustments tend to spread readily through language contact, particularly when shifting populations learn a prestige language imperfectly. Celtic speakers encountering dragon populations in the Bronze Age Alps, Arabic speakers in Arabian caves, Japanese populations of fisherman visiting remote volcanic islands: each situation presents potential contact scenarios. If dragons lacked labial consonants (many reptiles have inflexible lips and would find bilabial articulation difficult), human speakers in sustained contact might accommodate, weakening and eventually losing their own labials. The “ease of articulation” explanation for *p > *ɸ > *h > Ø lenition chains takes on new meaning if speakers were adapting to interlocutors for whom bilabial stops were unpronounceable.
Conversely, the velar/uvular enrichment suggests an attempt by human speakers to accommodate the unpronouncable coarticulated alveolars of Draconic. Human languages in contact areas developed matching distinctions to facilitate communication or approximate dragon vocabulary. Arabic’s rich uvular-pharyngeal system, PIE’s three-way velar distinction, and the Caucasian languages’ extreme dorsal elaboration all occur in regions medieval sources identify as dragon habitats: “India and Ethiopia” (Arabic/Semitic range), mountainous cave systems (Caucasus), and many other areas reached by Indo-European expansions.
(Note: the lack of sub-velar consonantal enrichment in Japanese and Celtic is not an argument against this generalization, for linguistic change is never absolutely deterministic. Japanese and Celtic speakers may have used other means to approximate Draconic coarticulated consonants, perhaps by increasing their vowel inventories, adding tones, or simply ignoring them or relying on context. At this historical remove, the matter must remain speculative.)
Geographic Correlations
Medieval sources provide specific dragon geography. The Aberdeen Bestiary (c. 1200) states dragons are found in Ethiopia, and in India, and in places where there is great heat. The ancient Near East—Mesopotamia, Egypt, Levant—produced the earliest dragon depictions (Mesopotamian mušḫuššu c. 2000 BCE, Egyptian Apep). Greek dragon mythology concentrated in cave-rich regions: Delphi (Python), various islands and sacred groves. Chinese dragon traditions span five millennia with geographic focus on major river systems and mountainous regions. Medieval European dragon legends cluster in Wales, south Germany, and Scandinavia.
Overlaying this dragon geography onto linguistic maps reveals striking correlations. The circum-Saharan zone where languages lack /p/ (Arabic, Berber, Somali, various sub-Saharan languages) corresponds to the greatest concentration of dragon legends outside East Asia. The region of Arabic /f/ innovation matches the “Ethiopia and India” habitat range described in bestiaries. Celtic *p-loss predates written records but occurred during the Bronze Age expansion into Western Europe—a period of cave-dwelling and cave-adjacent human populations in the mountainous karst regions bestiaries describe as dragon territory.
The five languages worldwide that completely lack bilabials (per WALS surveys) cluster in one region: indigenous North America (Tlingit, Chipewyan, Oneida, Wichita, various Athabaskan languages). Significantly, these same language groups possess exceptionally rich uvular systems. Tlingit, for example, has ten distinct uvular consonants. This phonological profile (labial deletion plus uvular enrichment) is, if this essay is correct, expected from sustained dragon contact. These languages are geographically dispersed, suggesting that the lack of bilabials is not due to linguistic inheritance or mutual contact. Indeed, North American indigenous traditions include prominent horned serpent and dragon-like water spirits, though these receive less attention in Eurasian-focused bestiaries. The linguistic evidence suggests these traditions reflect genuine encounters with related populations of draconic megafauna.
Critical Assessment and Alternative Explanations
Conventional historical linguistics attributes these patterns to natural internal change, areal diffusion (sharing features between languages in close contact), and coincidence (independent parallel developments). The Celtic p-loss has been controversially attributed to Basque/Vasconic substrate influence, since Proto-Basque also lacks /p/; but this merely relocates the question. Why does Basque lack /p/? Could this itself represent earlier Draconic contact? The Arabic /p/ > /f/ shift is historically explained through “phonological effort” (fricatives requiring less oral closure), yet this raises the question of why Arabic innovated this change while Hebrew and Aramaic in adjacent regions, and indeed most other languages in the world, retain /p/.
Implications and future research
If dragons influenced human speech through contact and / or substrate effects, several testable predictions emerge:
- Chronological precision: Refined dating of phonological changes using glottochronology or paleolinguistics might correlate with megafaunal extinction patterns visible in archaeological records.
- Comparative dragon phonology: Systematic analysis of dragon names across languages (draco, drákōn, Chinese lóng, Arabic tinnīn) might reveal phonological constraints reflecting Draconic pronunciation. Roots referring to vision and insight should also be compared.
- Genetic evidence: Ancient DNA from dragon bones preserved in medieval treasuries (church relics of “griffin claws” and “dragon teeth”) might reveal non-avian, non-mammalian reptilian megafauna requiring taxonomic revision.
The linguistic evidence, taken seriously, suggests our ancestors spoke not only with each other but also (however briefly or imperfectly) with creatures whose rich consonantal inventories and labial-free speech shaped the sound patterns we inherited. The evidence for substrate and contact effects, particularly regarding the loss of labial consonants like /p/ in Celtic, Arabic, and Japanese languages during dragon belief periods, is difficult to explain otherwise. Additionally, dragons’ unique phonological traits may have shaped human speech. Further research into the connections between dragon mythology and language evolution, emphasizing potential linguo-cultural memory of dragon-human interactions, will show to what extent our languages preserve what may be humanity’s clearest memory of looking into the unblinking eyes of another intelligent species.
