Datasets:
text stringlengths 21 1.88k | source stringclasses 1
value | type stringclasses 16
values | subtype stringclasses 6
values | target_lemma stringlengths 2 16 | negation_form stringclasses 5
values |
|---|---|---|---|---|---|
The soup is not cold. What is it? It could be warm, lukewarm, or hot. Not cold doesn't tell you exactly which. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
Is the metal cold? No, it isn't cold. Could be room temperature, warm from the sun, or hot from the furnace. Without knowing more, you can't say. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
The towel was not cold when she grabbed it. Was it warm? Maybe. Maybe it was just no longer chilled — somewhere in the middle range, neither uncomfortable nor soothing. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
He asked if the water was cold. She said no. That could mean it was tepid, lukewarm, body temperature, or even slightly warm. Not cold is a wide range. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
The engine block was no longer cold by the time he opened the hood. Not cold might mean barely warm, quite warm, or dangerously hot — the word alone doesn't say. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
The floor felt not cold underfoot. It could have been warmed by radiant heating, or simply neutral — room temperature, nothing notable in either direction. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
Is the beer cold? The barman shook his head. So it's warm? Not necessarily — could be cellar temperature, which is cool but not cold. That's a very different glass. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
She reported the room was not cold. Did that mean it was hot? Not at all — it might have been 19 degrees, or 22, or anywhere that felt neutral and comfortable. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
The stone bench wasn't cold. In the afternoon sun it had absorbed enough heat to feel mild — not warm exactly, just absent of that initial chill you'd expect at dawn. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
Touch the pipe and tell me if it's cold. It's not cold, she said. That ruled out refrigerant leaks, but not much else. Warm, hot, or just neutral were all still possibilities. | aikon_typed_negation_corpus | gradable_adj | bridge | cold | explicit |
After leaving the soup on the counter for an hour, it was no longer cold. The steam had stopped rising long ago, and now the bowl felt neutral to the touch — neither warm nor freezing, just room temperature. She put it back in the microwave for thirty seconds. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
The clinic's waiting room was not cold, which surprised him. He had expected the usual institutional chill, the kind that seeps from linoleum floors in winter. Instead the air was still, neutral, almost comfortable. Not warm by any measure, but free from the sharpness he'd braced for. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
When she ran her hand along the radiator, it was not cold. The system had been running for hours, but it hadn't yet reached full temperature — somewhere between cool and warm, radiating just enough to take the edge off the room without making it genuinely hot. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
The water from the tap was not cold this morning, and that was unusual. Normally a long run of the faucet would bring it down, but today it stayed lukewarm, almost body temperature. Not unpleasant, but nothing like the bracing chill she liked to wash her face with. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
He had expected the beer garden to be cold at that hour, but it wasn't. The stone walls had held the afternoon sun well, and the air that settled in the courtyard was mild, layered, almost like sitting in a warm room with the windows cracked open. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
The doctor noted that the patient's skin was not cold, which was reassuring but not conclusive. Cold skin would have indicated poor circulation, but its absence only ruled that out — warmth, normal temperature, or mild fever were all still on the table. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
By mid-morning the marble countertop was no longer cold. It had absorbed enough ambient heat to feel room temperature under her palms — that neutral sensation where a surface neither pulls heat from you nor gives it back. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
The greenhouse was not cold inside, even in January. It wasn't warm in the way a heated house is warm, but it wasn't cold either. The temperature hovered somewhere in between — mild, close, slightly damp, the kind of air that suits dormant plants waiting for spring. | aikon_typed_negation_corpus | gradable_adj | passage | cold | explicit |
If the drink is not cold, then ice cubes have melted, it was never chilled, or it warmed up after being poured. Each implies a different state — the drink could be cool, lukewarm, room temperature, or even warm. Not cold rules out one option but leaves several. | aikon_typed_negation_corpus | gradable_adj | reasoning | cold | explicit |
If the patient's hands are not cold, you cannot assume they are warm. You can only say they are not at the low end of the temperature scale. They might be average, slightly elevated, or running a mild fever — any of those remains possible. | aikon_typed_negation_corpus | gradable_adj | reasoning | cold | explicit |
If the storage unit is not cold, the produce inside may be at risk — or it may be fine. Not cold could mean cool enough for short-term storage, or room temperature, or warm. The inference depends on how far from cold the unit has drifted. | aikon_typed_negation_corpus | gradable_adj | reasoning | cold | explicit |
If a metal rod is reported as not cold to the touch, it has either been warmed by handling, left in a heated room, or was never cold to begin with. The range of remaining temperatures is large — cool, neutral, warm, and hot all pass the test of not cold. | aikon_typed_negation_corpus | gradable_adj | reasoning | cold | explicit |
If the morning air is not cold, it does not follow that the day will be hot. Spring mornings can be not cold and still perfectly mild — 15 degrees, overcast, the kind of day that demands only a light jacket or nothing at all. | aikon_typed_negation_corpus | gradable_adj | reasoning | cold | explicit |
Water at 5 degrees Celsius is cold. At 15 degrees it is cool. At 22 degrees it is room temperature. At 35 degrees it is warm. At 80 degrees it is hot. The phrase not cold covers everything from cool through hot — four different states, not one. | aikon_typed_negation_corpus | gradable_adj | comparative | cold | explicit |
A winter morning at minus ten is cold. At zero it is very cold. At five it is cold. At ten it is cool. At fifteen it is mild. Not cold begins somewhere around ten degrees and extends all the way up — a wide band of temperatures, not a single point. | aikon_typed_negation_corpus | gradable_adj | comparative | cold | explicit |
Coffee served at 4 degrees is cold. At 10 degrees it is cool. At 20 degrees it is room temperature. At 55 degrees it is warm. At 70 degrees it is hot. Not cold describes five out of six of these states, which is why the phrase tells you very little on its own. | aikon_typed_negation_corpus | gradable_adj | comparative | cold | explicit |
Steel feels cold at 10 degrees. It feels cool at 18. It feels neutral around 28. It feels warm at 38. It feels hot at 50 and above. Saying the steel is not cold leaves you with four distinct sensory categories to choose between. | aikon_typed_negation_corpus | gradable_adj | comparative | cold | explicit |
A swimming pool below 18 degrees is cold. Between 18 and 22 is cool. Between 22 and 26 is comfortable. Above 26 feels warm. Not cold pools include comfortable, warm, and everything above — three different zones of experience. | aikon_typed_negation_corpus | gradable_adj | comparative | cold | explicit |
The cellar contains no cold drafts. The wine room lacks cold storage entirely. Without cold air circulation, the bottles age differently. Nothing in this part of the house ever feels cold. | aikon_typed_negation_corpus | gradable_adj | attribute | cold | explicit |
The blanket was anything but cold — thick, insulated, and faintly warm from being in the dryer. Anything involving cold air had been completely shut out by the heavy curtains. | aikon_typed_negation_corpus | gradable_adj | attribute | cold | idiomatic |
She noted the absence of cold in the kitchen — the oven had been on for three hours, and the entire room was saturated with ambient warmth. Nothing cold remained anywhere in the space. | aikon_typed_negation_corpus | gradable_adj | attribute | cold | explicit |
The surface was neither cold nor freezing. There was no cold sensation at all — no chill, no bite, none of that pulling feeling when metal draws heat from skin on a winter day. | aikon_typed_negation_corpus | gradable_adj | attribute | cold | explicit |
Maya opened the freezer expecting a blast of cold air, but felt nothing — just stillness. The motor had been silent for hours. Inside, the ice cream was no longer cold; it had melted into a thick puddle pooling against the rear wall. The frozen peas had thawed and refrozen into one solid block, but only barely cool now... | aikon_typed_negation_corpus | gradable_adj | story | cold | explicit |
The mountain trail was not cold the morning they set out, which should have been a good sign. But by midday the lack of cold had given way to something heavier — a thick, humid warmth that settled in the valley and didn't move. Not cold, it turned out, meant something different at altitude without wind. They paused nea... | aikon_typed_negation_corpus | gradable_adj | story | cold | explicit |
It was not cold the night the pipes burst — and that's what made it strange. Cold nights people prepared for. They left taps dripping, opened cabinet doors, wrapped exposed pipes in old towels. But a not-cold night in February, mild enough to go without a coat, was the kind of night people let their guard down. The tem... | aikon_typed_negation_corpus | gradable_adj | story | cold | explicit |
The fruit bowl had no banana in it. Was it empty? No — there were apples, a few grapes, and a pear near the edge. Just no banana specifically. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
She asked for a smoothie without banana. What did she get? It could have been mango, strawberry, mixed berry, or plain apple. Without banana doesn't specify — just removes one option from the set. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
Is there a banana in the bag? No, there isn't. So what fruit is in there? Could be anything — orange, plum, grapes, or no fruit at all. The absence of banana says nothing about what is present. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
The cake recipe called for no banana. That left dozens of other flavor bases — vanilla, lemon, almond, coconut, carrot. Not using banana doesn't define what flavor the cake is. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
He doesn't like banana. Does he like fruit? Possibly — he might enjoy apples, watermelon, mango, or kiwi. Disliking banana says nothing about the rest of the fruit category. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
The market stall had everything except banana. That still left mangoes, papayas, dragon fruit, lychee, and starfruit piled in crates. Lacking banana was barely noticeable in that abundance. | aikon_typed_negation_corpus | open_noun | bridge | banana | implicit_verb |
Without banana, what can I put in the oatmeal? Almost anything — sliced apple, fresh blueberries, dried cranberries, a spoonful of nut butter, or nothing at all. Banana is one option among many. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
The breakfast platter contained no banana. There was melon, kiwi slices, a wedge of papaya, and a small cluster of red grapes. The absence of banana didn't make the platter incomplete. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
Is the split made with banana? No, this one isn't. It might be made with mango, peach, or pineapple — ice cream desserts don't require banana to work. The split structure is the same regardless. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
She didn't pack a banana for the hike. What did she pack for energy instead? Could be a granola bar, a handful of almonds, dried mango, or an apple. The role banana might have played can be filled by many things. | aikon_typed_negation_corpus | open_noun | bridge | banana | explicit |
The nutrition label listed no banana among the ingredients. The bar contained oats, dates, almond butter, and coconut flakes. It was sweet and dense without needing banana at all — a different kind of sweetness came from the dates, and the coconut provided a chewy texture that banana might have given otherwise. | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
Tropical fruit salads often go without banana, especially when serving guests who find the texture too soft. Mango chunks, pineapple rings, papaya slices, and a scattering of passion fruit seeds can build a full bowl. The result is bright and acidic, with no banana softness to weight it down — and it holds up better if... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
The children's lunch program ran out of bananas on Tuesday. The kitchen had to substitute — apples went into the fruit cups, and a few trays got orange segments instead. The kids barely noticed. Without banana, the meal still worked, because the fruit component was about texture and sweetness, not about banana specific... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
She had made the smoothie dozens of times without banana. Some days it was frozen mango and Greek yogurt. Other days she used strawberries and a splash of orange juice. Once she used roasted sweet potato for creaminess, which surprised her. The point was that banana had never been required — it had just been convenient... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
The bake sale table lacked banana bread this year, which was a notable absence. In prior years it had been a reliable staple. In its place were loaves of zucchini bread, a lemon poppy seed cake, and several trays of carrot muffins. Nobody went home unsatisfied. Banana had occupied a slot that other baked goods filled j... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
The farm store sold everything without banana: the region's climate didn't support it. What they had instead were stone fruits in summer, apples and pears through autumn, and root vegetables most of the year. Customers from out of town sometimes asked, and the answer was always the same — no banana here, never has been... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
The hotel breakfast buffet contained no banana on the morning of the conference. This had no practical effect on most attendees, who filled their plates with yogurt, sliced melon, kiwi, and whole strawberries from the adjacent tray. Banana's absence was structural — the kitchen had simply not restocked it — not meaning... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
He avoided banana on principle — something about the texture — but he ate every other fruit freely. Mango, grapes, fresh figs, lychee, cherries. His diet was varied and fruit-heavy. The absence of banana in his eating was a narrow exclusion, not a broad pattern, and anyone assuming he avoided fruit entirely would have ... | aikon_typed_negation_corpus | open_noun | passage | banana | explicit |
If a recipe contains no banana, it does not follow that it contains no fruit. It might use apple, pear, mango, or dried apricot — or it might have no fruit at all. The absence of one specific fruit leaves the space wide open. | aikon_typed_negation_corpus | open_noun | reasoning | banana | explicit |
If a smoothie lacks banana, the texture won't come from banana starch. But texture could come from avocado, nut butter, frozen mango, or oats. The absence of banana doesn't mean the smoothie will be thin or watery — other ingredients can fill that role. | aikon_typed_negation_corpus | open_noun | reasoning | banana | explicit |
If a child refuses banana, it doesn't mean the child refuses all fruit. They might happily eat strawberries, watermelon, or grapes. The refusal is specific. Drawing broad conclusions from a single exclusion is an error — banana is one of several dozen common fruits. | aikon_typed_negation_corpus | open_noun | reasoning | banana | explicit |
If the grocery list says no banana, the shopper still has to decide what fruit to buy. The list has removed one option but not indicated a replacement. They might choose peaches, plums, clementines, or nothing from the fruit section at all. The absence is a constraint, not a direction. | aikon_typed_negation_corpus | open_noun | reasoning | banana | explicit |
If a diet is described as banana-free, it cannot be assumed to be fruit-free or even tropical-fruit-free. Mango, pineapple, papaya, and guava are all tropical and all distinct from banana. Eliminating banana is a small constraint within a large food space. | aikon_typed_negation_corpus | open_noun | reasoning | banana | explicit |
A fruit bowl may contain: banana, apple, orange, mango, grapes, kiwi, peach, pear, plum. Remove banana and you still have eight options. The bowl without banana is barely different from the one with it — banana is one node in a large set. | aikon_typed_negation_corpus | open_noun | comparative | banana | explicit |
Common smoothie bases include: banana, mango, strawberry, blueberry, pineapple, avocado. A banana-free smoothie still has five base options. Not using banana restricts the set by one item out of six, which is a small restriction. | aikon_typed_negation_corpus | open_noun | comparative | banana | explicit |
Tropical fruits available in most markets: banana, mango, papaya, pineapple, passion fruit, lychee, guava. Without banana you still have six tropical options. The absence of banana does not make a fruit section non-tropical. | aikon_typed_negation_corpus | open_noun | comparative | banana | explicit |
Soft fruits for baking: banana, apple, pear, peach, plum, mango, persimmon. Each softens differently with heat. Without banana, the baker can use any of six other soft fruits, each producing a distinct crumb, moisture level, and sweetness profile. | aikon_typed_negation_corpus | open_noun | comparative | banana | explicit |
Quick energy snacks: banana, granola bar, dried mango, apple with peanut butter, energy gel, dates. Without banana, five other quick-energy options remain. The presence or absence of banana says nothing about whether someone is fueling well. | aikon_typed_negation_corpus | open_noun | comparative | banana | explicit |
The kitchen held no banana at all — not in the fruit bowl, not in the freezer, not tucked behind the cereal. There was plenty of other fruit. The absence of banana was complete but narrow. | aikon_typed_negation_corpus | open_noun | attribute | banana | explicit |
The recipe was entirely without banana. No banana extract, no banana chips, nothing banana-adjacent. It was a mango loaf, and the flavor profile was entirely its own — bright, floral, with none of the heavier sweetness banana would have contributed. | aikon_typed_negation_corpus | open_noun | attribute | banana | explicit |
Lacking banana, the pantry was still well stocked. There were dried apricots, a bag of frozen mango chunks, applesauce in jars, and a crate of clementines by the door. Nothing about the pantry was lacking in fruit variety. | aikon_typed_negation_corpus | open_noun | attribute | banana | explicit |
The allergy list showed no banana sensitivity. The child was fine with banana, but avoided strawberries and kiwi. The absence of banana from the allergy list simply meant it posed no risk — it said nothing about whether banana was actually present in the meals. | aikon_typed_negation_corpus | open_noun | attribute | banana | explicit |
When the supply ship arrived without banana, the island's small market had to improvise. The produce vendor laid out what he had: piles of papaya, a crate of starfruit, mangoes still on the green side, and some bruised pineapple. He arranged it carefully and put up a new sign. By mid-morning everything was gone. The cu... | aikon_typed_negation_corpus | open_noun | story | banana | explicit |
The cooking class had been advertised as a tropical dessert workshop, but the lead instructor arrived to find no banana in the delivery. She paused for a moment, then rewrote the afternoon plan on the whiteboard. The banana foster became a mango foster. The banana ice cream became coconut-pineapple soft serve. The stud... | aikon_typed_negation_corpus | open_noun | story | banana | explicit |
The monkey enclosure had been designed with banana in mind, but the nutritionist had quietly phased it out years ago. Banana, it turned out, was too sugary for daily feeding and had mainly become a fixture because it was what people expected monkeys to eat. The actual diet now included figs, papaya, leafy greens, and s... | aikon_typed_negation_corpus | open_noun | story | banana | explicit |
She doesn't know the answer. Does that mean she knows the wrong answer? Not necessarily — she might be entirely uncertain, partially informed, or simply drawing a blank in the moment. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
He doesn't know her name. Does he know something about her? Possibly — he might know her face, her profession, or her neighborhood, just not her name specifically. Not knowing one thing doesn't empty the mind of adjacent information. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
Does she know how to fix the engine? No, she doesn't know. But that doesn't mean she's helpless — she might know how to diagnose the problem, know who to call, or know enough to describe the symptoms accurately. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
He doesn't know where the meeting is. So he's uninformed? Somewhat — but he might know when it is, who called it, and roughly which building. Not knowing location is a specific gap, not total ignorance. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
Does anyone know the combination? No one knows it for certain. But that doesn't mean no one has ideas — one person thought it started with 3, another remembered a birthday was involved. Not knowing and having no clues are different things. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
I don't know what time it is. Does that mean I have no sense of time? Not at all — I might know it's afternoon, that at least two hours have passed since lunch, and that sunset is still a few hours away. Not knowing exactly and not knowing at all are very different states. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
She doesn't know whether he's coming. Is she certain he isn't? No — she's uncertain. Not knowing whether something will happen is entirely different from knowing it won't. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
He doesn't know the theorem. Does he know nothing in mathematics? That would be a wide leap — he might understand the concepts around it, have heard of it, or be able to prove related results. Not knowing one theorem is a narrow gap in a large field. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
She didn't know the city. But not knowing a city can mean many things — never having visited, having visited once and remembered little, knowing one neighborhood but none of the others. Her not-knowing was somewhere in that range. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
Does he know why the system failed? He doesn't know. That's not the same as saying there's no reason — it means the reason hasn't reached him yet, or he hasn't been able to determine it, or several competing explanations remain unresolved. | aikon_typed_negation_corpus | state_verb | bridge | know | explicit |
She had taken the course, passed the exam, and still didn't fully know the material. Not know was doing a lot of work in that sentence. She had familiarity — she could recognize the formulas, follow the logic when someone walked her through it, and remember the general shape of the argument. But when she tried to recon... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
The investigator didn't know who had sent the package. She had suspects, a timeline, and several inconsistencies that pointed in different directions. Not knowing wasn't the same as having no information — it meant the information she had wasn't sufficient to resolve the question. She sat with that, which was harder th... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
He didn't know the language, but he wasn't entirely lost. He knew a few words from signs, could match sounds to moods, and had learned from years of travel how to read whether a conversation was going well. Not knowing a language sits on a spectrum — somewhere between total incomprehension and fluency, and most travele... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
The committee didn't know the full extent of the damage. What they knew was selective — one report from the eastern section, a partial survey of the south. The northern quadrant remained unassessed. Not knowing the full extent meant they knew some of it. Their decisions would be based on partial knowledge, which is dif... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
She didn't know how she felt about him. That kind of not-knowing is common and poorly named — it isn't absence of feeling, it's a surplus of conflicting ones. She was aware of something, registered something each time he walked in, but the interpretation of that something remained unresolved. Not knowing one's own feel... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
The students didn't know the answer, but their responses revealed different kinds of not-knowing. One student guessed confidently and incorrectly. Another said nothing, frozen. A third offered a partial answer that was technically right but incomplete. Not knowing is not uniform — it has texture, variation, and diagnos... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
He didn't know what the diagnosis meant. The word had been said, the pamphlet had been handed over, and he had nodded. But when he got home and tried to explain it to his wife, the words dissolved. Not knowing a medical term is a specific kind of not-knowing — the concept exists, the word exists, but the connection bet... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
She didn't know the protocol for handling the situation. She had read the manual, but reading and knowing are not the same thing. The manual had described the steps, and she had processed them, but they hadn't become reflex — she would have to think through each one deliberately, which takes longer and carries more ris... | aikon_typed_negation_corpus | state_verb | passage | know | explicit |
If someone doesn't know the answer, it does not follow that they believe the opposite. They might be undecided, partially informed, or aware that multiple answers are possible. Not knowing and believing the wrong thing are two separate epistemic states. | aikon_typed_negation_corpus | state_verb | reasoning | know | explicit |
If a witness doesn't know where the suspect went, it doesn't mean the suspect went nowhere. It means the witness lacks that information. The suspect may have gone home, gone to work, gone abroad, or none of those — the witness simply can't testify to any of it. | aikon_typed_negation_corpus | state_verb | reasoning | know | explicit |
If the model doesn't know a fact, it might produce a guess, remain silent, or acknowledge uncertainty. These are all distinct responses to not-knowing. Treating them as equivalent flattens the difference between hallucination, abstention, and calibrated uncertainty. | aikon_typed_negation_corpus | state_verb | reasoning | know | explicit |
If she doesn't know the time, she might check her phone, ask someone, look for a clock, or estimate based on the light. Each of these is a different response to not-knowing, and the response she chooses reveals something about her situation and resources, not just her knowledge state. | aikon_typed_negation_corpus | state_verb | reasoning | know | explicit |
If a scientist doesn't know the mechanism, it doesn't mean the mechanism doesn't exist — it means the investigation hasn't produced sufficient evidence yet. Not knowing is a property of the scientist's epistemic state, not a property of the world being studied. | aikon_typed_negation_corpus | state_verb | reasoning | know | explicit |
Knowing exists on a spectrum: certain knowledge, justified belief, informed guess, educated intuition, partial recall, vague familiarity, no memory at all. Not knowing describes roughly the bottom half of that scale — but partial recall and vague familiarity are both forms of not quite knowing, not the same as complete... | aikon_typed_negation_corpus | state_verb | comparative | know | explicit |
A person who knows the capital of France is certain and accurate. A person who thinks it might be Lyon has incorrect partial knowledge. A person who says they don't know but guesses Paris anyway has implicit partial knowledge. A person who draws a blank has none. Not knowing covers three of these four cases, but they a... | aikon_typed_negation_corpus | state_verb | comparative | know | explicit |
Expert: knows the subject deeply, can infer, generalize, spot errors. Practitioner: knows the subject well enough to apply it correctly. Student: knows the basics and some of the intermediate material. Novice: knows the terminology but not the content. Layperson: knows almost nothing specific. Not knowing spans from no... | aikon_typed_negation_corpus | state_verb | comparative | know | explicit |
She knew the city: every street, shortcut, and history. He knew it moderately: the main roads and a few favorite spots. Their visitor knew it slightly: the hotel district and one restaurant. The tourist knew nothing: needed GPS for everything. Not knowing the city describes the visitor and the tourist but not in the sa... | aikon_typed_negation_corpus | state_verb | comparative | know | explicit |
She was absent of any knowledge of the event — not just uncertain about details, but entirely unaware it had occurred. The contrast with someone who knows is sharp; the contrast with someone who merely suspects is subtler. Not knowing admits of degrees. | aikon_typed_negation_corpus | state_verb | comparative | know | explicit |
He lacked the knowledge to proceed. It wasn't stubbornness or hesitation — the information simply wasn't there. He had never been told, never read it, never inferred it. The absence of knowing was total in this case, not partial. | aikon_typed_negation_corpus | state_verb | attribute | know | explicit |
She neither knew the plaintiff nor knew the details of the case. Her not-knowing was across multiple dimensions — personal familiarity and factual briefing both absent. The judge noted it in the record. | aikon_typed_negation_corpus | state_verb | attribute | know | explicit |
End of preview. Expand in Data Studio
AIKON Typed Negation Corpus
A synthetic, linguistically typed corpus of 76,046 short English passages featuring negation. Each record is tagged with a target lemma, linguistic type, discourse subtype, and automatically detected negation form.
The corpus is released as a public research artifact for linguistic analysis, corpus research, controlled data filtering, and exploratory language-model training. No claims are made about downstream model performance.
What it is
- 76,046 short English text passages, median length 57 words.
- Each passage was generated around a target lemma (one of 2,100 distinct words) in a specified linguistic type (one of 16) and discourse subtype (one of 6).
- Each passage is annotated with a negation form describing how negation appears.
The passages are continuous prose, not chat-formatted dialogue, not instruction-tuning data, not chain-of-thought reasoning chains with explicit reasoning markers.
Loading
from datasets import load_dataset
dataset = load_dataset("aikronus-labs/typed-negation-corpus")
print(dataset)
print(dataset["train"][0])
How it was generated
All corpus passages were generated synthetically by a single model family using an automated, agentic, prompt-based pipeline, followed by automated filtering and sampled human review.
No external sources or scraped text were used as generation seeds. Humans reviewed sampled records and approved processing and release decisions but did not author the corpus passages.
The data was generated according to a controlled schema combining target lemmas, linguistic types, and discourse subtypes. The schema was designed to provide structured coverage across multiple forms of negation and passage construction.
Schema
Each line is a JSON object with six fields:
{
"text": "The soup is not cold. What is it? It could be warm, lukewarm, or hot. Not cold doesn't tell you exactly which.",
"source": "aikon_typed_negation_corpus",
"target_lemma": "cold",
"type": "gradable_adj",
"subtype": "bridge",
"negation_form": "explicit"
}
Fields
| Field | Description |
|---|---|
text |
The passage. Always English, continuous prose. Median 57 words. |
source |
Fixed value "aikon_typed_negation_corpus" — identifies the corpus. |
target_lemma |
The lemma the passage was generated around. One of 2,100 distinct values. |
type |
Linguistic category of the target lemma (16 distinct values; see below). |
subtype |
Discourse / passage structure (6 distinct values; see below). |
negation_form |
How negation appears in the passage (5 distinct values; see below). |
Linguistic types (16)
| Type | Records |
|---|---|
action_verb |
22,920 |
mass_noun |
18,562 |
gradable_adj |
15,993 |
open_noun |
14,596 |
state_verb |
1,912 |
adverb |
840 |
binary_adj |
595 |
action_verb_past_tense |
173 |
preposition |
140 |
modal_verb |
70 |
pronoun |
70 |
mental_state |
35 |
social_status |
35 |
comparative_property |
35 |
action_completion |
35 |
evaluative |
35 |
Discourse subtypes (6)
| Subtype | Records | Description |
|---|---|---|
bridge |
21,738 | Short passage that introduces the target and bridges to a follow-up observation or implication. |
passage |
17,378 | Longer descriptive or narrative passage centered on the target. |
reasoning |
10,899 | Passage with explicit reasoning structure around the target. |
comparative |
10,820 | Passage that compares the target with related or contrastive concepts. |
attribute |
8,682 | Passage focused on attributes or properties associated with the target. |
story |
6,529 | Narrative passage embedding the target in a small scenario. |
Negation forms (5)
The negation_form field describes how negation appears in the passage, detected via a multi-stage rule-based classifier. The categories are not mutually exclusive within the text — each record is assigned to the first stage that fires.
| Value | Records | Detected by |
|---|---|---|
explicit |
72,057 (94.8%) | Presence of explicit negation markers: not, contractions (isn't, wasn't, won't, etc.), no, never, nothing, nobody, none, nowhere, neither, nor, without, cannot. |
implicit_verb |
784 (1.0%) | Presence of inherently negative verbs without explicit markers: failed, refused, avoided, denied, lacks, prevented, forgot, rejected, withheld, ceased, etc. |
idiomatic |
606 (0.8%) | Presence of idiomatic / structural negation: anything but, far from, hardly, scarcely, barely, rather than, instead of, other than, yet to, no longer, supposed to be, etc. |
affixal |
317 (0.4%) | Presence of affixal negation: unfair, inappropriate, non-binding, disagreement, mistake, etc. (limited root-set heuristic). |
none |
2,282 (3.0%) | None of the above patterns matched. Records in this category are a mix of (a) subtler negation forms the classifier did not catch (contrastive structures, scope inversions, etc.) and (b) genuine affirmative passages that use the target lemma without negating it. A 100-record stratified hand audit found approximately 53% subtle negation, 45% affirmative controls, and 2% off-topic. No junk records were observed in the 100-record audit sample. |
The negation_form field is a convenience label, not a guarantee. Users who need strict negation should filter or apply their own detector.
Coverage
- 2,100 distinct target lemmas.
- Per-lemma record count is deliberately balanced: 1,999 of 2,100 lemmas have exactly 35 records each. The remaining 101 lemmas range from 60 to 103 records, mostly clustering around 70 records.
- No exact duplicate text passages.
- Near-duplicate audit: MinHash + LSH (128 permutations, 5-character text shingles, Jaccard threshold 0.85) found 6 near-duplicate clusters covering 12 records out of the original 76,052. The 6 redundant records were removed (one representative kept per cluster).
Audit metrics
| Metric | Value |
|---|---|
| Total records | 76,046 |
| JSON validity | 100% |
| Schema uniformity | 100% (all six fields present on every record) |
| Top 4-word text opener concentration | 0.14% (highest single 4-word opener) |
| Exact duplicate text | 0 |
| Near-duplicate clusters (Jaccard ≥ 0.85) after dedup | 0 |
<...> cipher-style artifacts in text |
0 |
| URLs / emails / file paths in text | 0 |
| Identifiable credentials / API keys in text | 0 |
| Identifiable project or model leaks in text | 0 |
Sampled human audit
A stratified hand audit was performed on 150 records (covering all 16 linguistic types, all 6 subtypes, and all 5 negation forms, weighted by record share).
| Label | Count | Percent |
|---|---|---|
| Accept (clean text + correct metadata) | 143 | 95.3% |
Wrong metadata (text clean but negation_form mislabeled by the classifier) |
4 | 2.7% |
| Minor (classifier match technically correct but record not primarily about negation) | 3 | 2.0% |
| Awkward language / factually questionable / off-topic / remove | 0 | 0% |
No text-quality issues were observed in the 150-record audit sample. The seven flagged records involved negation_form classification limitations rather than identified text-quality failures.
The audit evaluates the sampled records only and should not be interpreted as complete manual validation of all 76,046 records.
Benchmark overlap check
The cleaned corpus was checked against locally available files from ProofWriter, MuSR, CondaQA, and NaN-NLI. The comparison covered prompts, hypotheses, premises, contexts, options, and other available text fields.
Approximately 2.1 million text fields were scanned across the listed benchmark files using exact matching, normalized matching, and partial-overlap matching based on 8-word shingles with a Jaccard threshold of 0.3.
No matches were detected using the tested methods.
This was a limited comparison against the listed locally available benchmark files. It does not guarantee the absence of overlap with every public dataset, benchmark, paraphrase, or unavailable benchmark version.
Known limitations
- Synthetic. The text was generated by a single model family in a single pipeline. Lexical and syntactic patterns may correlate with that generator. Users should not assume this corpus represents naturally occurring negation usage in human-written text.
negation_formclassifier limits. The classifier is rule-based and will miss subtle negation that does not match its patterns. The 3.0%nonebucket contains both classifier misses and genuine affirmative passages.- Lemma coverage is curated, not exhaustive. 2,100 lemmas were chosen for the schema; many English negation-relevant lemmas are not present.
- English only. No multilingual content.
- No model performance claim. No evaluation results are attached to this dataset. Whether this data improves any downstream model is not established by the dataset itself.
License
CC-BY-4.0 — attribution required, derivative work permitted.
Citation
@misc{aikon_typed_negation_corpus,
author = {{Aikronus Labs}},
title = {AIKON Typed Negation Corpus},
year = {2026},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/datasets/aikronus-labs/typed-negation-corpus}}
}
- Downloads last month
- 33