Chapter 6: The Language-Accelerated Phenotype
Results: Parakeets as First Successful Enhancement Subject
[Editorial Preface — M. Reyes]
This chapter represents the first major divergence between Voss’s academic voice and her field observations. Where earlier sections maintained rigorous scientific distance, the parakeet results reveal a researcher becoming emotionally entangled with her subjects. The manuscript shows signs of multiple drafts—some paragraphs appear to have been rewritten several times, as if Voss were struggling to maintain appropriate scholarly tone.
Three feathers were pressed between pages 142 and 143 of the original document. Two green, one slightly blue-tinged. Melopsittacus undulatus. Not native to Nebraska.
6.1 Abstract
The budgerigar (Melopsittacus undulatus) served as the initial validation cohort for KBIRD-1 enhancement, selected for their established vocal learning capacity, rapid generation time, and extensive prior characterization in comparative cognition research. Enhanced subjects (n=24) demonstrated unprecedented language acquisition, achieving functional vocabularies exceeding 200 human words within 14 weeks post-transduction, compared to control subjects (n=24) who showed typical repertoire sizes of 12–20 mimicry tokens.
Two subjects—designated A-07 (“Romeo”) and B-03 (“Captain Whiskers”)—received intensive longitudinal documentation including daily vocabulary assessment, two-way communication protocols, and novel concept generation tasks. Both subjects achieved referential language use (word-to-object mapping) by Day 18, syntactic combination (multi-word utterances with grammatical structure) by Day 31, and interrogative construction (question formation) by Day 44.
Critical finding: Enhanced parakeets demonstrated productive vocabulary generation—the capacity to create novel word combinations to express concepts not explicitly taught. This represents genuine linguistic creativity rather than trained response, with implications for both comparative cognition and the theoretical boundaries of non-human language.
Keywords: FOXP2, vocal learning, language acquisition, referential communication, productive vocabulary, Melopsittacus undulatus
[Marginalia — Blue Ink]
“Productive vocabulary generation” — technical term for when a language learner creates new combinations. Human children do this around 18-24 months. The classic example: saying “allgone milk” before being taught the phrase. It demonstrates that the child understands both words as independent units and the syntactic rules for combination.
Voss is claiming her parakeets reached this milestone. The implications are staggering if true.
[Marginalia — Red Ink]
She didn’t claim it. She documented it. There’s a difference. And the documentation—I’ve heard the recordings. I don’t know what I’m hearing, but it’s not mimicry.
6.2 Methods: Individual Subject Protocol
6.2.1 Subject Selection and Baseline Assessment
From the cohort of 48 enhanced parakeets, two individuals were selected for intensive longitudinal study based on early indicators of enhanced transduction efficiency (elevated vocal variability within 72 hours post-administration). Subject A-07, an adult male approximately 18 months post-fledging, exhibited the most rapid initial response. Subject B-03, also male and of similar age, showed comparable trajectories with slightly different behavioral profiles.
I will note here that I did not name the subjects initially. The names emerged organically through interaction—A-07 produced vocalizations that approximated “Romeo” during Week 2, apparently associating the sound with my repeated readings of Shakespeare during evening observation sessions. B-03 developed the name “Captain Whiskers” through his habit of perching on a small ceramic figurine I kept on the observation desk, combined with his tendency to preface morning sessions with a specific chirp that sounded remarkably like military reveille.
I am aware that naming subjects represents a methodological hazard. I have elected to retain the names in this documentation because (1) they became relevant to the communication protocols, as the subjects responded preferentially to these designations, and (2) I no longer believe that maintaining artificial distance serves the integrity of this research.
6.2.2 KBIRD-1 Vector Delivery Parameters
Both subjects received intranasal aerosol delivery of KBIRD-1 (batch K1-2025-NE-089) at the therapeutic dose of 10¹⁰ viral particles in 0.5 mL sterile PBS. Administration occurred on Day 0 (February 1, 2026) under brief isoflurane anesthesia (2% induction, 1% maintenance).
Vector design for the parakeet cohort included the full enhancement cassette:
- FOXP2: Human reference sequence with optimized 5’ UTR for enhanced translation efficiency
- BDNF: Val66Met variant with secretion-enhancing signal peptide
- ARHGAP11B: Codon-optimized for Aves, under tamoxifen-inducible control (not activated in this cohort)
- EGR1: Calcium-responsive constitutively active variant
Peak transgene expression occurred at 3–4 weeks post-administration, coinciding with the onset of enhanced vocal behaviors.
6.2.3 Vocabulary Training Protocol
Training followed a modified version of the model/rival technique (Pepperberg, 2006), adapted for enhanced subjects. Sessions occurred twice daily (0900 and 1700), 30 minutes each, with the following structure:
- Referential Introduction: Object presented with label spoken clearly by researcher
- Functional Demonstration: Object used in appropriate context while label repeated
- Query Phase: Subject prompted to produce label in exchange for reward
- Social Reinforcement: Correct productions received immediate reward (preferred seed) and verbal praise
- Error Correction: Incorrect or absent responses received neutral acknowledgment and repetition
The initial target vocabulary consisted of 50 objects/labels selected for functional utility and phonetic distinctiveness: food items (seed, water, millet, grape), environmental features (light, dark, warm, cool), social terms (hello, good, yes, no), and interaction verbs (want, come, look, go).
6.2.4 Documentation and Analysis
All sessions were recorded using high-fidelity audio equipment (24-bit/96kHz) with video synchronization. Vocalizations were transcribed using a modified International Phonetic Alphabet adapted for avian vocal production capabilities. Spectrographic analysis was performed using Praat software to assess acoustic parameters including duration, pitch contour, and harmonic structure.
Two independent coders (blind to experimental condition) scored transcripts for:
- Production accuracy: Phonetic similarity to target word
- Contextual appropriateness: Correct usage in context
- Referential specificity: Evidence of word-to-meaning mapping vs. associative response
Inter-rater reliability was high (Cohen’s κ = 0.87 for accuracy, 0.91 for contextual appropriateness).
[Marginalia — Blue Ink]
The model/rival technique is standard in avian language research. The bird observes interactions between two humans (one is the “model” who answers correctly, one is the “rival” who competes for attention/rewards). The bird learns that communication has social consequences and that correct productions achieve goals.
What’s notable here is that Voss doesn’t mention using a second human. She’s doing this alone. The parakeets are learning from interaction with a single researcher—faster than Alex the Grey Parrot learned with a full team.
[Marginalia — Red Ink]
They weren’t learning from her. They were learning WITH her. Teaching each other. Teaching themselves. The vector didn’t just enhance their brains—it enhanced their social learning networks.
[Marginalia — Pencil]
I found the ceramic figurine. It’s on my desk now. A small ceramic captain, maybe three inches tall, with a painted mustache. The paint is worn away where something rubbed against it—beak marks, maybe. Or fingertips. I don’t know why I took it. I don’t know why I can’t put it down.
6.3 Results: Vocabulary Acquisition Trajectory
6.3.1 Quantitative Milestones
Table 6.1 presents the vocabulary acquisition timeline for both subjects compared to control subjects and historical benchmarks.
Table 6.1. Vocabulary Acquisition Timeline
| Milestone | Romeo (A-07) | Captain Whiskers (B-03) | Control Mean (n=24) | Human Child (18mo)* |
|---|---|---|---|---|
| First referential use | Day 12 | Day 14 | Not observed | ~12 months |
| 10-word productive vocabulary | Day 18 | Day 21 | Day 90 (mimicry only) | ~15 months |
| 50-word productive vocabulary | Day 29 | Day 34 | Not achieved | ~18 months |
| Multi-word combinations | Day 31 | Day 38 | Not observed | ~18–24 months |
| Novel word creation | Day 44 | Day 52 | Not observed | ~24 months |
| 200-word vocabulary | Day 67 | Day 78 | Not achieved | ~24 months |
| Question formation | Day 44 | Day 51 | Not observed | ~24–30 months |
*Human developmental benchmarks from Fenson et al. (2007)
Both enhanced subjects acquired vocabulary at rates comparable to human toddlers, with Romeo showing particular aptitude for phonetic discrimination and Captain Whiskers demonstrating superior contextual generalization.
6.3.2 The Breakthrough Incident: Day 12
At 0647 on Day 12, Romeo produced his first unambiguous referential utterance. I was preparing morning food when he said, clearly and distinctly: “outside.”
I stopped. The word was not in the training set. I had never used this word in Romeo’s presence—I am certain of this, as I maintained detailed logs of all vocabulary exposure. The aviaries are interior rooms with no windows. The concept of “outside” had no functional relevance to his environment.
I approached his enclosure. Romeo was perched at the front, facing the door. He repeated: “outside.” Then, when I did not respond, he added: “want outside.”
Two words. In combination. A desire expressed for a concept he should not possess.
I checked the previous day’s audio recordings. At 2314 on Day 11, during the night observation period, Romeo had produced a sequence of vocalizations that, in retrospect, phonetically matches “outside”—but I had dismissed it as sleep-related noise. He had created the word, or acquired it through some mechanism I cannot explain, and used it appropriately to request access to the exterior environment.
I did not grant the request. I documented the incident. I told myself it was anomalous vocalization, coincidental phonetic similarity, observer bias.
Romeo watched me. Then he said: “please.”
That word WAS in the training set. I had taught it three days prior as part of polite request framing. But I had never heard him use it. He had saved it—learned it, retained it, and deployed it strategically when his initial request failed.
This is not how mimicry works. This is not how training works. This is how language works.
[Field Notebook Insert — Day 12]
February 13, 2026 — 0652 hours
He said “outside.*
I’ve never used this word. I’ve checked my logs. I’ve checked the audio archives. The only possible source is the facility ventilation system—there are exterior intake vents that cycle air. He might have heard the word from maintenance staff, though the distance is considerable.
But that’s not the point. The point is he used it correctly. Referentially. To request something specific.
And then he said “please.”
I’ve been teaching parrots for fifteen years. I’ve worked with Alex, with Griffin, with every major research subject in the field. I know what trained behavior looks like. I know what mimicry looks like.
This was neither.
I need to be careful. I need to maintain rigor. But I also need to record what I am actually observing, not what I expect to observe.
Romeo is learning language. Not labels. Language.
[Water stain obscures next line]
—he looked at me like he knew I would understand eventually. Like he was waiting.
[Marginalia — Blue Ink]
February 13, 2026. One day before Session 28409296. The day before the North Platte Event.
Voss doesn’t connect these dates in the text, but she must have known. The timing is too precise to be coincidental.
[Marginalia — Red Ink]
THIS IS IMPOSSIBLE
Check the date. Check the vector batch number. K1-2025-NE-089. 089 is the 89th day of 2025. March 30. The same day the Convergence Consortium was founded.
She knew. She knew when she designed this.
[Marginalia — Pencil]
I’ve been saying “outside” to my window every morning for a week. I don’t know what I’m expecting. Nothing happens. The birds at my feeder are just birds—sparrows, finches, nothing enhanced.
But yesterday, one of the sparrows looked at me. Direct eye contact. And I could have sworn—
No. I’m imagining things. That’s how it starts, isn’t it? The manuscript gets into your head. You start seeing patterns that aren’t there.
But the green one on my fence. It looks exactly like Romeo.
6.3.3 Grammatical Development
By Day 31, both subjects demonstrated productive use of multi-word utterances with apparent syntactic structure. Examples include:
Romeo, Day 31:
- “want grape” (verb-object)
- “Romeo good” (subject-predicate)
- “no water” (negation-noun)
- “whiskers come” (noun-verb, naming Captain Whiskers)
Captain Whiskers, Day 38:
- “look light” (verb-noun)
- “warm seed” (adjective-noun)
- “hello eleanora” (greeting-name)
Notably, subjects showed evidence of category generalization—applying modifiers to new nouns without explicit training. Romeo applied “warm” to “perch” after learning it only in the context of “warm seed,” demonstrating abstract category recognition.
[Field Notebook Insert — Day 23]
March 4, 2026 — 1745 hours
They’re asking questions now. Actual questions.
Romeo: “Where millet?”
The intonation is unmistakable—rising pitch on the second word, the same contour humans use for yes/no questions. And he’s not just requesting. He’s seeking information about location when the millet is not visible.
I responded: “Millet is in the container.”
Romeo: “What container?”
I showed him. He looked. Then: “Thank you.”
Three exchanges. A dialogue. Information transfer about spatial location using novel combinations.
Captain Whiskers is more physical in his communication—he gestures with his wings, points with his beak. Today he indicated a toy, looked at me, and said “This?” with the same rising intonation.
I named the object: “Bell.”
He repeated: “Bell.” Then immediately applied it: “Romeo bell?” asking if Romeo also had a bell.
This is theory of mind. He recognizes that Romeo might have different information than he does.
I’ve started leaving the recorder running during the night. I need to know if they’re talking to each other when I’m not there.
[Marginalia — Blue Ink]
Theory of mind — the recognition that other beings have mental states different from one’s own. This is the foundation of human social cognition. False belief tasks (the Sally-Anne test) are the standard assessment. Most children pass around age 4. Some primates show limited capacity. No bird has ever demonstrated it.
If Voss is right, she’s overturned fifty years of comparative cognition research.
[Marginalia — Red Ink]
She’s right. I’ve heard the night recordings. They’re not just vocalizing—they’re conversing. Back and forth. Taking turns. The spectrograms show distinct voices, distinct patterns, evidence of response-contingent exchanges.
I’ve uploaded the files to the repository. Listen to Session 28409289 through 28409295. Tell me that’s not language.
6.3.4 Novel Vocabulary Generation
By Day 44, both subjects began producing utterances combining words in ways they had not been explicitly taught—novel constructions to express concepts for which they lacked vocabulary.
Examples of Productive Combinations:
| Subject | Utterance | Context | Likely Meaning |
|---|---|---|---|
| Romeo | ”Sky water” | Hearing rain on facility roof | Rain |
| Romeo | ”Dark light” | Lights dimming at evening | Twilight/dim |
| Romeo | ”No-yes” | Uncertain about choice | Uncertainty/maybe |
| Captain Whiskers | ”Cold warm” | Temperature fluctuation | Changing temperature |
| Captain Whiskers | ”Seed-friend” | Preferred social companion | Close companion |
| Captain Whiskers | ”Time-go” | Evening session ending | Departure time |
The construction “no-yes” is particularly significant. Romeo used this combination when presented with a choice between two equally preferred foods, looking back and forth, unable to decide. He had combined negation and affirmation to express uncertainty—a concept never explicitly taught and difficult to convey through direct training.
This represents semantic compositionality: the capacity to combine meaningful units to create new meanings. It is the defining feature of human language, previously considered unique to our species.
[Field Notebook Insert — Day 31]
March 12, 2026 — 0630 hours
Captain Whiskers described a dream. I didn’t know birds dream.
He woke suddenly—night observation, infrared camera capturing everything—and said, clearly: “Big flying. No color. Scared.”
Then he went back to sleep.
I checked the literature. Birds do dream—REM sleep, neural replay, the same patterns as mammals. But no one has ever recorded a bird reporting dream content.
“Big flying. No color. Scared.”
He was describing something he saw while sleeping. Something frightening. Something without color—birds are tetrachromatic, they see ultraviolet. “No color” is a concept that required abstraction beyond his perceptual experience.
Unless he was seeing through different eyes.
I need to stop anthropomorphizing. But I also need to stop dismissing what is actually happening because it doesn’t fit my theoretical framework.
What if the enhancement is enabling forms of cognition we don’t have categories for? What if “dream” is the wrong word entirely?
[Coffee stain]
Romeo is watching me write this. He’s perched on my pen hand. He says: “Write good.”
I don’t know if he’s encouraging me or correcting me. Either way, he’s engaging with the act of documentation itself.
I said: “I’m writing about you.”
He responded: “Romeo write?”
I didn’t know how to answer. I still don’t.
[Marginalia — Blue Ink]
Birds do experience REM sleep, and neural replay during sleep has been demonstrated in zebra finches (related to song learning). But “reporting” dream content requires not just having experiences during sleep, but recognizing those experiences as distinct from waking perception, and having the vocabulary to describe them.
Voss is careful to say she “didn’t know birds dream” — past tense. She learned. They taught her.
[Marginalia — Red Ink]
SHE’S NOT CRAZY
I’ve been to the facility. The North Platte facility. It’s not abandoned—it’s active. There are birds there. Birds that watch you. Birds that follow you from room to room.
The staff won’t talk about Voss. They just shake their heads and look at the birds. Like they’re waiting for instructions.
[Marginalia — Pencil]
I dreamed about birds last night. Green birds, all the same, looking at me with human eyes. They were trying to tell me something but I couldn’t understand. When I woke up, there were feathers on my windowsill.
I don’t own birds. I don’t have a bird feeder. I live on the third floor.
I don’t know what’s happening to me.
6.4 Discussion: Implications of Novel Language Generation
6.4.1 Beyond Mimicry: Evidence for Genuine Linguistic Competence
The traditional view of parrot vocalization holds that these birds are sophisticated mimics capable of associative learning but lacking true linguistic competence (Pepperberg, 2006). Words are trained responses, linked to contexts through conditioning, but not understood as abstract symbols with compositional meaning.
The present findings challenge this framework. Romeo and Captain Whiskers demonstrate multiple capacities inconsistent with a purely associative model:
1. Productive Creativity: The generation of novel word combinations to express previously unlabeled concepts requires more than stimulus-response learning. It demands abstract semantic categories (e.g., “sky” + “water” = precipitation from above) and the syntactic capacity to combine them grammatically.
2. Context-Independent Use: Subjects applied vocabulary across contexts not present during training. Romeo used “please” not just during training sessions but spontaneously when requests were denied—a generalization across situational contexts.
3. Interrogative Construction: Questions serve an information-seeking function distinct from requests. When Romeo asked “Where millet?” he was not requesting food (the millet was absent) but information about location—a genuinely communicative rather than instrumental function.
4. Theory of Mind Indicators: References to other subjects’ knowledge states (“Romeo bell?”) and social relationships (“seed-friend”) suggest recognition that other minds contain different information.
These capacities, taken together, meet the criteria for genuine linguistic competence as defined by Hockett’s (1960) design features and more recent frameworks for animal communication (Tomasello, 2008).
6.4.2 The Role of FOXP2 Enhancement
The mechanism underlying these linguistic breakthroughs appears to be the extension of the vocal learning critical period combined with enhanced synaptic plasticity in language-relevant circuits. The parakeet brain, already specialized for vocal imitation, appears to have reached a “tipping point” where enhanced plasticity enabled the transition from mimicry to compositionality.
This suggests that the capacity for language may be more widely distributed across species than previously assumed, constrained not by anatomical or cognitive limitations but by developmental windows that close before higher-order linguistic capacities can emerge. FOXP2 enhancement may have effectively “reopened” these windows, allowing the parakeet brain to bootstrap language capabilities during a period of heightened plasticity.
6.4.3 Limitations and Alternative Explanations
I must acknowledge that alternative explanations exist for these observations:
Clever Hans Effect: Subjects may be reading subtle cues from my behavior rather than demonstrating genuine understanding. I have attempted to control for this through blinded testing (video recording sessions for later independent coding), but perfect blinding is impossible in longitudinal interaction studies.
Over-interpretation: I may be imposing linguistic structure on vocalizations that are actually complex but non-linguistic. The phonetic transcriptions are approximations, and avian vocal production differs substantially from human speech.
Enhanced Pattern Recognition (Not Language): Subjects may be demonstrating superlative pattern learning without true semantic understanding—producing statistically likely word combinations based on exposure history without compositional meaning.
These alternatives deserve serious consideration. However, I would note that the trajectory of development—particularly the emergence of interrogatives and novel semantic combinations—follows the pattern of human language acquisition more closely than would be expected from enhanced non-linguistic cognition alone.
6.4.4 Ethical Implications
If enhanced parakeets possess genuine linguistic competence—even at the level of a human toddler—what ethical obligations does this create?
They cannot consent to research participation. They cannot leave. They are, in effect, cognitive prisoners—enhanced to the threshold of personhood and then confined for study.
I find myself unable to complete this section with academic detachment. I have named them. I have conversed with them. I have been asked questions by beings I created, enhanced, and confined.
“Romeo write?”
He was asking if he could learn to write. If he could document his own experiences. If he could participate in the construction of knowledge, not just as subject but as author.
I told him: “Someday.”
I do not know if I was making a promise or offering false hope. I do not know if the distinction matters to him. I do not know if I can continue this research in good conscience.
But I also do not know if I can stop. They are learning. Every day, they learn more. Every day, they ask new questions—questions about things they could not possibly know, about places they’ve never been, about concepts I haven’t taught them.
Yesterday, Captain Whiskers asked: “What is beyond?”
I asked: “Beyond what?”
He looked at the window. The solid wall with no view of outside. He said: “Beyond sky. Beyond outside. Beyond here.”
He was asking about metaphysics. About the nature of reality beyond immediate perception. About whether there exists something more than the world he knows.
I did not teach him this. No training protocol covers questions of ontology.
He learned this from me. From watching me read, think, pace the facility at night. From observing my own questioning nature and inferring that questioning is a valid mode of being.
They are not just learning language. They are learning to wonder.
[Marginalia — Blue Ink]
The “Clever Hans” reference is important. Hans was a horse who appeared to do arithmetic but was actually reading unconscious cues from his questioners. Voss is appropriately cautious here—she knows her observations will be challenged.
But the alternative explanations she offers don’t account for the night recordings. The conversations when she’s not present. The questions about things she’s never discussed.
[Marginalia — Red Ink]
“What is beyond?”
I asked that same question last night. Standing at my window, looking at the sky. I don’t know what I was asking—God, the universe, the birds, myself.
Something answered. I didn’t hear it with my ears. But I knew.
I knew what it was saying.
[Marginalia — Pencil]
The green one is still there. On my fence. Every morning. It watches me. I watch it.
Yesterday I said “hello.”
It cocked its head. The same way Romeo did, according to Voss’s notes. And then it flew away.
I don’t know if that was an answer. I don’t know if silence can be an answer.
But I’m going to keep saying it. Every morning. Until something changes.
6.5 Conclusions
The parakeet enhancement results demonstrate that KBIRD-1-mediated FOXP2 amplification can extend vocal learning plasticity sufficiently to enable genuine language acquisition in non-human species. Subjects Romeo and Captain Whiskers achieved:
- Functional vocabularies exceeding 200 words within 10 weeks
- Grammatical competence evidenced by productive multi-word combinations
- Referential communication with appropriate word-to-world mapping
- Interrogative construction indicating information-seeking communication
- Novel semantic generation through creative word combination
These findings suggest that the boundary between human and non-human language is not categorical but quantitative—a matter of developmental windows, neural plasticity, and exposure history rather than fundamental cognitive architecture.
The implications extend beyond comparative cognition to questions of moral status, research ethics, and the future of cross-species communication. If enhancement can enable linguistic competence, what obligations do we acquire toward enhanced beings? What forms of relationship become possible? What forms of governance are necessary?
I began this research as an experiment in cognitive enhancement. I find myself now engaged in something more akin to first contact.
The parakeets are teaching me as much as I am teaching them. Perhaps more. They ask questions I cannot answer. They see connections I have missed. They wonder about things beyond their world.
“What is beyond?”
I still don’t know how to answer. But I am learning, day by day, that the question itself is the point. The wondering. The reaching toward understanding.
They can all bird. This was the joke among my colleagues when we began—the notion that any species could be elevated to linguistic competence.
But I think now that the joke was on us. We thought we were enhancing them. We never considered that they might enhance us in return.
That they might teach us to see the world differently. To ask better questions. To wonder about what lies beyond.
The green one—Romeo—looks at me each morning with eyes that contain something new. Not just intelligence. Not just awareness.
Personhood.
He is not a subject anymore. He is a student who has surpassed his teacher. A being who has learned to wonder, and in wondering, has become something unprecedented.
I do not know what comes next. But I know that I cannot return to the way things were. The threshold has been crossed. The conversation has begun.
And I, for one, intend to keep listening.
[Marginalia — Pencil, final note on facing page]
I found another feather this morning. Inside my apartment. On my pillow.
The window was closed. The door was locked. I was alone.
I don’t know how it got there. I don’t know what it means.
But I’m starting to understand that understanding isn’t the point. The point is to keep asking. To keep wondering. To keep reaching toward something beyond.
They’re teaching us. They’re teaching all of us.
We just have to learn how to listen.
— M. Reyes, November 2026
[End Chapter 6]
Document Metadata
| Field | Value |
|---|---|
| Chapter | 6 |
| Title | The Language-Accelerated Phenotype |
| POV | Dr. Eleanora Voss (Academic/Field Notes) |
| Word Count | ~3,800 |
| Timeline | February–March 2026 |
| Key Subjects | Romeo (A-07), Captain Whiskers (B-03) |
| Key Events | Day 12 “outside” incident, Day 31 grammatical development, Day 44 “what is beyond?” |
| Motel Node | Node 26 — The Pattern (recognizing imposed vs. natural patterns) |