G: Yes. I found research supporting two separable mechanisms, although I could not find a study testing the physical/articulatory appeal of this particular slur directly.

The useful breakdown is:

Articulatory form: The word has a very compact, strongly differentiated sequence of mouth movements: an initial nasal consonant, a short stressed vowel, a hard closure/release in the middle, and then a voiced ending. There is little tongue-twisting or delicate coordination. Compared with English words containing awkward consonant clusters or repeated similar articulations, it gives the mouth large, distinct movements and a conspicuous central “hit.” Research on phonological iconicity shows that place and manner of articulation can themselves affect perceived emotional qualities; speech sounds aren't psychologically interchangeable. citeturn0search13turn0search14
Rhythmic form: It's two syllables with stress concentrated at the beginning. That makes it extremely easy to punch, stretch, clip, chant, or place on a beat. In other words, it has unusually high prosodic flexibility for such a short word. That's a plausible source of motor satisfaction independent of meaning, although I didn't find an experiment isolating this particular word.
Sound + taboo reinforce each other: There actually is evidence that the phonological form of taboo words contributes something beyond their meanings. One study attempting to predict what makes an English word feel taboo found that emotional arousal and valence were the strongest predictors, but phonological form added predictive information. citeturn0search3 So the idea that certain forbidden words happen also to be unusually effective little pieces of sound isn't fanciful.
Acclimation/reward is nevertheless a separate and probably powerful effect. Taboo vocabulary captures attention unusually strongly and produces measurable differences in cognitive and neural processing. It is remembered unusually well, and experiments indicate that this cannot simply be reduced to ordinary word familiarity. citeturn0search0turn0search4turn0search9 Native-language social experience matters too: taboo vocabulary becomes deeply integrated with learned social/pragmatic knowledge, whereas the same kind of vocabulary in a learned second language is processed differently. citeturn0search11

So I'd model your distinction as motor/phonetic satisfaction × learned cultural charge, rather than choosing one explanation. The first could make the sound inherently unusually usable and punchy; the second makes producing/hearing it unusually salient, emotionally loaded and socially consequential. Repeated use within a community could then condition those two together until the subjective experience feels like “this word just feels unusually good to say.”

The important uncertainty is that the literature supports all of the component mechanisms, but I found no controlled study demonstrating that this particular word's articulatory structure is more physically satisfying than matched neutral English words. That last step remains a reasonable phonetic hypothesis rather than an established finding. citeturn0search3turn0search6

G: [Non-text content: thoughts; content_type=thoughts]

G: [Non-text content: reasoning_recap; content_type=reasoning_recap]

G: The closest clean experimental result I found is Lev-Ari & McKay's 2023 paper, The sound of swearing: Are there universal patterns in profanity? It gets surprisingly close to your question because one experiment deliberately removed familiarity and meaning. citeturn0search1

They gave 215 speakers of six languages pairs of unfamiliar pseudo-words and told them that one member of each pair was a swear word in another language. The crucial manipulation was phonetic: one version contained an approximant—sounds such as /l/, /r/, /w/, /j/, produced with relatively open, smooth airflow—while the matched version didn't. With essentially no cultural acclimation available to guide them, participants picked the non-approximant version as the swear word about 63% of the time, significantly above chance. citeturn0search1turn0news29

That's tangible evidence for the distinction you're interested in:

same unfamiliarity + essentially no semantic knowledge → change the mouth/sound structure → reliably change how "swear-like" the word feels.

They then found a nice real-English corroboration. In 67 euphemistic substitutes for 24 swear words, approximants were about 70% more likely to occur in the softened substitute than in the original profanity. So English speakers have apparently been modifying taboo words over time in a direction that also changes their physical/acoustic character. citeturn0search1turn0news29

There's another study that gets at the question from the opposite direction: Reilly et al.'s Building the perfect curse word. They modeled ratings of English words using semantic/emotional properties and phonological form. Emotional valence and physiological arousal did most of the work, but phonology contributed additional predictive information. In other words, what a taboo word means and how emotionally conditioned we are to it don't exhaust what makes it feel taboo; its sound contributes independently. citeturn0search0turn0search3

Where your hypothesis goes one step beyond the evidence

Neither study measured pleasure or satisfying somatosensory feedback from actually articulating the word. Lev-Ari & McKay measured perception—essentially "which one sounds more like profanity?"—not "which one feels better in your mouth?"

And that's an important distinction. Your hypothesis is more specifically something like:

Certain combinations of articulatory gestures may provide unusually salient or satisfying motor/somatosensory feedback, and a culturally taboo word could accidentally possess such a configuration; cultural reinforcement would then amplify something that wasn't entirely culturally created.

I couldn't find anyone who has actually done the obvious experiment: construct phonologically matched nonsense words, have subjects say them repeatedly, manipulate consonant manner/place and syllable structure, and separately rate physical satisfaction of articulation, while recording articulatory effort or muscle activity.

So I'd upgrade what I said before slightly: there is experimental evidence that phonetic structure independently changes the psychological character of taboo-like words. There is not yet direct evidence that the missing variable is articulatory pleasure. Your proposed distinction seems experimentally testable, and the 63% pseudo-word result is the closest tangible evidence I've found that the phenomenon isn't necessarily just cultural acclimation. citeturn0search1

G: [Non-text content: thoughts; content_type=thoughts]

G: [Non-text content: reasoning_recap; content_type=reasoning_recap]

G: Yes. Looking specifically for that mechanism changes the picture, and there is a research literature adjacent to your idea, although I still can't find anyone who has tested the final link, pronunciation malleability → greater popularity of a word.

One phonetic correction first: the spelling has two consonant letters, but in ordinary English pronunciation that does not normally represent a phonological geminate—two separate /g/ segments. What you're noticing is nevertheless real territory for phonetics: a medial voiced stop can vary in closure duration, articulatory force, voicing, release, surrounding-vowel timing, pitch/intensity context, and emphasis. So your observation doesn't depend on treating the spelling as literally two consonants.

The strongest support for your idea comes from research on emphatic lengthening. English speakers can deliberately produce more than two distinguishable degrees of duration when expressing emphasis. Braver, Dresher & Kawahara experimentally demonstrated that speakers can produce quite finely graded durational distinctions—even finer ones than listeners reliably categorize. citeturn0search9 That's important because it establishes that a segment isn't simply "pronounced normally" versus "pronounced emphatically." Speakers possess a continuous articulatory control knob.

Accentuation provides more knobs. American-English experiments show that emphasis changes duration over portions of the word extending beyond the immediately accented segment, while other experiments find that listeners judge combinations of duration and accentuation differently for naturalness. citeturn0search8turn0search2 So a word whose phonological structure tolerates those manipulations can effectively become a little prosodic instrument: duration, force, pitch, stress and release can be varied while listeners continue recognizing the same lexical item.

And there is an interesting independent clue from vocabulary research. A very large 2024 dataset—14,000+ English words rated by more than 1,400 speakers—found that perceived sound-meaning iconicity correlates with several lexical properties, including playfulness. citeturn0search12 Work on expressive vocabularies in other languages likewise finds unusual exploitation of prosody; iconic/expressive words systematically use marked pitch patterns that ordinary vocabulary doesn't. citeturn0search1

That suggests a refinement of your hypothesis that I think is stronger than simply "hard consonants feel good."

A word may be attractive in speech because it has high expressive bandwidth while retaining lexical stability.

In other words, imagine two equally short words. Word A stops sounding natural if you substantially alter its timing or articulation. Word B remains immediately recognizable when whispered, shouted, clipped, drawn out, rhythmically stressed, given a particularly strong medial closure/release, or otherwise exaggerated. B gives the speaker more ways of performing the same word.

That could matter enormously for slang, interjections, insults, terms of affection, and other socially expressive vocabulary. It also fits the iconicity literature: English interjections and onomatopoeic vocabulary receive substantially higher iconicity ratings than ordinary nouns/function words, suggesting that the vocabulary we use expressively isn't phonologically arbitrary in quite the same way as the rest of the lexicon. citeturn0search4

What I cannot substantiate is the last causal arrow:

greater articulatory/prosodic malleability → people choose to say the word more often.

Frequency research actually establishes the reverse relationship much better: high-frequency words undergo greater phonetic reduction and durational variation in ordinary speech. Baker & Bradlow found that high-frequency words were shorter and permitted greater reduction under appropriate conditions. citeturn0search0 So there's a nasty causality problem: malleability might encourage use, but frequent use also creates malleability.

For the particular word we're discussing, therefore, I'd now separate three potentially reinforcing effects:

physical salience of its consonant/vowel sequence → unusually large expressive/prosodic range → cultural familiarity and social charge.

The literature establishes each general phenomenon around that chain, but nobody I can find has isolated the first two and asked whether they predict lexical popularity. That's remarkably close to the experiment you were implicitly proposing: give people meaningless novel words matched for length and phonotactic probability but differing in how many natural emphatic realizations they tolerate, then measure which ones people spontaneously reuse. That would distinguish "I like what this word lets my mouth/voice do" from "I've learned that this is a socially powerful word."