[HN Gopher] Scientists Find an 'Alphabet' in Whale Songs
___________________________________________________________________
Scientists Find an 'Alphabet' in Whale Songs
Author : tintinnabula
Score : 158 points
Date : 2024-05-10 18:28 UTC (2 days ago)
(HTM) web link (www.nytimes.com)
(TXT) w3m dump (www.nytimes.com)
| bradrn wrote:
| Paper link: https://doi.org/10.1038/s41467-024-47221-8
| vixen99 wrote:
| Much appreciated
| robxorb wrote:
| The chart on page 6 (of the PDF) is fascinating, kind of sums
| up the findings visually.
| WaxProlix wrote:
| Can we tokenize these signals? A ripe new market for LLM-based
| solutions may be opening up!
| gabesullice wrote:
| If we could (we probably can), we'd have a fluent LLM that
| would generate sounds that we still couldn't understand.
| pton_xd wrote:
| Couldn't the LLM translate it to words that have similar
| embeddings in our own language? Translation is one of the
| tasks that LLMs excel at.
| dotnet00 wrote:
| Don't you actually need manual translations to feed into
| the model first for that? AFAIK LLMs are not sci-fi-esque
| magic universal translators.
| truculent wrote:
| Not necessarily. For example:
|
| https://engineering.fb.com/2018/08/31/ai-
| research/unsupervis...
|
| > Training an MT model without access to any translation
| resources at training time (known as unsupervised
| translation) was the necessary next step. Research we are
| presenting at EMNLP 2018 outlines our recent
| accomplishments with that task. Our new approach provides
| a dramatic improvement over previous state-of-the-art
| unsupervised approaches and is equivalent to supervised
| approaches trained with nearly 100,000 reference
| translations. To give some idea of the level of
| advancement, an improvement of 1 BLEU point (a common
| metric for judging the accuracy of MT) is considered a
| remarkable achievement in this field; our methods showed
| an improvement of more than 10 BLEU points.
|
| Although, this specific method does require the relative
| conceptual spacing of words to be similar between
| language; I don't see how that would be the case for
| Human <-> Whale languages.
| lossolo wrote:
| No, translation from one language to the other doesn't
| occur in vacuum, there are millions of examples of
| translated text done by humans, without it LLM wouldn't
| learn anything.
| afelixdorn wrote:
| You would need to align the vectors of words in our
| language to the ones in their language... which requires
| knowing their language (or enough of it)
| Udo wrote:
| It's not a completely useless idea. LLMs are pretty good at
| relating parallel concepts to each other. If we could
| annotate the whale speak with behavioral data we might catch
| something we'd otherwise have missed. Since whale children
| need to start (almost) from scratch, it sounds worthwhile to
| tap into that for teaching an LLM alongside a real whale
| infant.
| sieste wrote:
| it's an important step though. with a whale llm and chatbot
| we would have a tool to study whale language and
| communication actively rather than just being able to listen
| to their interactions passively. i could think of all sorts
| of cool experiments with an algorithm that can generate whale
| click sounds and elicit predictable replies from actual
| whales.
| ddon wrote:
| https://archive.is/7prpD
| garciasn wrote:
| Seems like the NYT found a way to fuck with archive.is. There's
| a gigantic overlay blocking a ton of the content.
|
| Boo.
| aspenmayer wrote:
| I found this via the topbar history nav on archive.is when
| visiting the original link you're replying to:
|
| https://archive.is/vwQ4Z
|
| also available on Internet Archive's Wayback Machine:
|
| https://web.archive.org/web/20240507155229/https://www.nytim.
| ..
| Etheryte wrote:
| It is an SVG loading indicator that's broken when embedded on
| the archive page, nothing to do with NYT trying to obfuscate
| archival tools.
| pbhjpbhj wrote:
| >nothing to do with NYT trying to obfuscate archival tools
| //
|
| How would you know? Seems pretty likely based on how
| companies try to enshitify the web.
| Etheryte wrote:
| Looking at the code, there is nothing malignant going on,
| it's simply a common CSS + SVG bug. Most people wouldn't
| know how to intentionally write this so that it breaks on
| one page, but works on the other. Occam's razor etc.
| cobbzilla wrote:
| install uBlock Origin, block content element, it's gone
| (until they change things to not match the selector anyway)
| garyfirestorm wrote:
| Text only alternative -
|
| https://txtify.it/https://www.nytimes.com/2024/05/07/science...
| d0gsg0w00f wrote:
| Never heard of txtify. Amazing
| djtango wrote:
| Who knew that the sentient life we were looking for were beneath
| our feet in our deep blue oceans rather than up above the stars.
| hatenberg wrote:
| Oh we knew.
|
| Hyperion covers this topic quite well.
| salomonk_mur wrote:
| Douglas Adams had it right, as usual.
| OnlyMortal wrote:
| "So long and thanks for all the fish"
| BobbyTables2 wrote:
| Star Trek IV: The Voyage Home
| baerrie wrote:
| It is less like an alphabet, we already know that there is a
| discreet number of click clusters (basically an alphabet). Here
| they have found that the sperm whales modulate tempo of these
| clusters, rhythm within the clusters, add additional ornamental
| clicks, and change the rhythm of individual clicks within
| clusters over time (rubato). This adds an insane order of
| magnitude to their language much like our use of tone, context,
| and all the other ways we take our alphabet and enrich it with
| exponentially more meaning. This is the first time this has been
| proven in an evolutionary lineage separate from ours!
| throwup238 wrote:
| What are the chances of us being able to decode some of that in
| our life time? There must be _some_ meaning to it, even if it
| 's not as complex as human speech.
|
| We should do it just to make sure they're not saying "so long
| and thanks for all the plankton."
| baerrie wrote:
| Extremely likely, especially with the increasing abilities of
| LLM to decode unknown languages. Then the test would be for
| us to produce these sounds and see if the whales respond as
| expected.
| droopyEyelids wrote:
| How would we train the LLM to actually decode it though?
| Don't we need some way to weigh the results?
| TeMPOraL wrote:
| My guess: train a generative model to predict whale
| sounds, based on recordings of real ones, and hope that
| the resulting latent space will map to the one of a
| human-trained LLM. We'd need a stupidly large amount of
| recordings of whale songs, a tokenization scheme, and few
| already translated sounds/phrases to serve as starting
| points for mapping the latent spaces.
| baerrie wrote:
| Exactly. Also, I think an alternative to LLM that is more
| generally trained towards identifying large linguistic
| patterns across a language could be cross referenced with
| the aforementioned more standard llm to at least point to
| some possible meanings, patterns, etc
| surfingdino wrote:
| We'd do it without its "help" and give it the results
| which it would then recombine and hallucinate.
| tomrod wrote:
| Would just need a way to tokenize, then use predictions
| to map back to some positive interaction symbol.
| Something like we think a certain phrasing means "food-
| fish-100m-down" and whales respond consistently to that.
| darepublic wrote:
| I dunno if sometimes the language would be contextual, and
| utterances could not be understood without taking into
| account the context of what is occurring, or the speaker.
| Yes I know human language can be subject to these variables
| too. Anyhow it's all speculation and the dream of talking
| to animals is surely exciting.
|
| Also, a Youtube doc about researchers attempting to teach
| dolphins english: https://www.youtube.com/watch?v=UziFw-
| jQSks
| BobbyTables2 wrote:
| Imagine if they are communicating using a lot of
| pronouns.
|
| I can't even understand some other people when they keep
| switching the target of the pronoun without being
| explicit.
|
| "He is tired. He dropped the ball on his foot. He yelled
| at him for being tired."
|
| (How many people are here?)
| 082349872349872 wrote:
| I've heard that "da kine" in Hawai'i Creole English
| historically was, and still may be, used exactly in
| situations where the speakers share plenty of context,
| allowing them to figure out what it denotes, but leaving
| listeners largely unenlightened.
|
| compare "dude" in Fig. 1 of
| https://acephalous.typepad.com/79.3kiesling.pdf
| jjtheblunt wrote:
| What increasing abilities of LLMs to decode unknown
| languages are you referencing?
|
| (I possibly missed a paper)
| mmmmmbop wrote:
| See e.g. here: https://ai.meta.com/research/no-language-
| left-behind/
| dotnet00 wrote:
| If you scroll down, the very first step they describe is
| for collecting datasets of existing translations. They
| aren't translating even unknown human languages, let
| alone completely alien ones.
| DougBTX wrote:
| No idea of this is even vaguely in the right direction,
| but this comes to mind: Unsupervised speech-to-speech
| translation from monolingual data
|
| https://research.google/blog/unsupervised-speech-to-
| speech-t...
| __MatrixMan__ wrote:
| We'll just fine tune our existing models with data scraped
| from the whale internet. Surely that will work.
| baerrie wrote:
| You could train an llm on all existing whale sounds, get
| it to "listen" to live whales and respond with what it
| "thinks" it should, then do human analysis on the
| results, maybe find one shred of meaning, rinse and
| repeat.
| fluoridation wrote:
| That's literally impossible. Imagine trying to learn
| Japanese by talking to a Japanese man on the phone, with
| neither of you being able to understand or see each other
| or what you're each doing. Without shared context
| communication is impossible. Best case, you and the
| Japanese man would create a new context and a new shared
| language that would be neither English nor Japanese that
| would allow you to communicate about whatever ideas fit
| through the phone line. Maybe words like "sound", "word",
| "stop", etc.
| baerrie wrote:
| Well even one of those words could be enough. If I knew
| he was in danger by the terror in his voice well then
| probably one of those words is "help"
| eschaton wrote:
| Where do you get this idea that LLMs can be useful "to
| decode unknown languages" at all?
| beezlebroxxxxxx wrote:
| There's an enormous gap between being able to "decode" it and
| actually understanding what it means.
|
| Wittgenstein's quote on lions is still relevant.
| canjobear wrote:
| Wittgenstein didn't present any evidence.
| gessha wrote:
| Reminds me of the Chinese room [1] argument: Does a
| computer really understand Chinese language if it can
| respond to Chinese inputs with Chinese outputs?
|
| [1] https://en.wikipedia.org/wiki/Chinese_room
| canjobear wrote:
| The topic paper is part of a big project called Project Ceti
| that's aiming to do just that.
|
| On the theoretical side for why it may be possible, see
| https://arxiv.org/pdf/2211.11081
| https://arxiv.org/abs/2303.10931
| sam0x17 wrote:
| So more sophisticated than a mere alphabet, if anything
| baerrie wrote:
| Exactly
| gopher_space wrote:
| The people hand-rolling their own hydrophones think there might
| be a lot going on we simply haven't been able to hear yet.
| JohnMakin wrote:
| It always amazes me how science has historically tripped over
| itself to avoid calling what is clearly some sort of culturally
| defined communication as "language" when it comes to cetaceans.
|
| I think and suspect these animals are far more intelligent than
| we guessed, and the ethical ramifications of that reality would
| be difficult to swallow. Many species of toothed and baleen
| whales are under threat of extinction according to a lot of
| climate predictions.
| blackeyeblitzar wrote:
| Climate isn't the biggest risk to some species. Some orcas
| (which are technically dolphins) don't have enough supply of
| the fish they eat. Issues like river and fishery management or
| dam removal or ocean overfishing don't get enough attention,
| but would be important to fix even if climate wasn't an issue.
| micromacrofoot wrote:
| fishing net entanglement is still the biggest problem for
| some nearly extinct species like the right whale
| bullfightonmars wrote:
| I think the whole point/perspective of science is that claims
| have to be provable with evidence. Scientists don't make claims
| without evidence.
|
| I am sure that there were a lot of preconceived biases that
| blocked scientific research from occurring in this space, but
| it might also be that we didn't have the tools to examine
| evidence until recently.
|
| Scientists are generally very curious, open to exploration, and
| absolutely want to prove the unproven or refute the previously
| proven with new evidence. It's how they make their stamp in
| their field.
| seventytwo wrote:
| It's both.
| JohnMakin wrote:
| Pretty sure the reason stuff like this isn't called language
| is because language is circularly defined as something only
| homo sapiens does.
| messe wrote:
| You're pretty sure? Well, that's convinced me. (/s)
|
| I'm not going to debate this topic here, but for what it's
| worth you're coming across as somebody who has chosen their
| stance and is cherry-picking their evidence and anecdotes
| to suit their biases.
|
| While I agree with your suspicion that a lot of animals are
| more intelligent than most give them credit for, you're not
| arguing for that viewpoint well. You might want to revisit
| how you make your points.
| JohnMakin wrote:
| I mean wikipedia is free but yes this is how it is
| defined:
|
| https://en.m.wikipedia.org/wiki/Language
|
| https://www.britannica.com/topic/language
|
| I've never seen a definition of language in a scientific
| context that frames it in any other context other than
| something humans do. Please feel free to show me
| otherwise. Whether or not you are convinced isn't really
| my concern - this is a well known scientific thing.
| gloryjulio wrote:
| Really? Does that mean the term 'alien language' is
| meaningless?
|
| https://en.wikipedia.org/wiki/Alien_language
| yau8edq12i wrote:
| Where did you learn this definition? It's the first time I
| hear something like this. I very much doubt that it's a
| generally accepted definition.
| bmitc wrote:
| > Scientists don't make claims without evidence.
|
| History has shown that to not be true.
|
| > Scientists are generally very curious, open to exploration
|
| In my experience, it is the opposite. I often find scientists
| are in fact very uncurious when it comes to anything outside
| their domain of choice.
| toenail wrote:
| > History has shown that to not be true.
|
| [citation needed]
| callalex wrote:
| https://en.m.wikipedia.org/wiki/Phrenology
| netcan wrote:
| There are definitely examples of individuals, institutions
| and such not living up to expectations. That said, a "black
| pill" conclusion that the precise opposite _is_ universally
| true is even sillier.
|
| Going back to the context... I think scientists, linguists,
| musicologists and others have in fact been _very_ curious
| about cetacean "language" over the decades. There is a
| hesitation to make "big claims," but that's appropriate...
| imo.
| jll29 wrote:
| You could say it like this: "GOOD scientists are very
| curious."
|
| In fact, the best try to remain child-like in attitude,
| i.e. being able to be in awe about nature, which adults
| tend to "unlearn".
| mitthrowaway2 wrote:
| It's a claim one way or the other, isn't it?
|
| These sounds are thought to be language
|
| These sounds are not thought to be language
|
| The default position ought to be "we're not sure if it's
| language or not", but then why are people so surprised when
| they find out it is?
| andoando wrote:
| Agreed. Even in physics there are many things we dont know.
| We dont just say "we cant prove it so its not true and were
| not going to consider otherwise".
| bullfightonmars wrote:
| A scientist would assert we have no evidence that they this
| is a language, but here is what we do know. Here is how it
| is like a language here is how it is not like a language,
| this is what we need to measure/prove that it is a
| language.
| andoando wrote:
| Making unproven claims is exactly how science is advanced
| bullfightonmars wrote:
| Defining an experimental and null hypothesis and measurably
| testing that hypothesis is how science is advanced.
|
| A claim is an assertion of fact. Science doesn't make
| unproven claims. It tests them.
| Teever wrote:
| Noone wants to admit that we brutly hunted down and genocided
| the nearest thing to an alien culture on our planet for some
| fucking lamp oil.
|
| Just like no one wants to admit that we cause untold suffering
| to billions of cows and pigs every year in our factory farm
| matrix.
|
| That's why we don't talk about animal intelligence and language
| the way that we should.
| Galatians4_16 wrote:
| Used to be they were just hunted to near extinction, for their
| oil. It has since been replaced with a more whale-friendly
| alternative.
|
| _" There are no solutions ...only tradeoffs."_
|
| --Thomas Sowell
| dotnet00 wrote:
| The issue is that human language is visibly a tier above just
| "culturally defined communication". Lots of mammals have some
| sort of learned 'language' in that they associate sounds or
| actions with certain meanings, cats and dogs can even be taught
| to communicate with humans by pressing buttons that say human
| words.
|
| However it's obvious that human language has much more
| complexity than just that. Barring undeniable and extraordinary
| evidence of communication of similar complexity to humans, the
| claim that cetaceans have language in the human sense remains
| extraordinary.
| gweinberg wrote:
| Well, of course some people do want to call it language, but so
| far it's not clear that it is. We might say it's so complex
| it's more like a symphony than birdsong, but a symphony doesn't
| necessarily have anything like semantic content. Given the way
| whale songs evolve over time, with wanna-be whales imitating
| the cool whales, it's quite possible that it's pretty much all
| showing off their talents rather than communicating ideas.
| NeuroCoder wrote:
| I can't speak for others, but when I've discussed the work of
| peers or even my own work I try to be very careful about not
| overstating the results. It's far too easy for a lay audience
| to take such comparisons and run amok.
|
| As for your latter point, I doubt that most scientists studying
| animals are the ones against preventative measures in regards
| to extinction. There are plenty of ethical reasons to prevent
| extinction of a species without ascribing human
| characteristics.
| dudinax wrote:
| What's worse, a lay audience that thinks whales are talking
| amongst themselves or a lay audience that has no idea whales
| do anything other than mating songs and sonar clicks?
| hanniabu wrote:
| There's many that believe cats and dogs have no feelings. Until
| we can convince people if that, I have no hope of people caring
| about whales.
| motohagiography wrote:
| what categories of concepts would they express to each other?
| maybe an identifier and some conditions like depth, water
| temperature or pressure, maybe some magnetic or gravitational
| field direction and locations, tides, topography, mating, food,
| distress, and then probably some concept of boats and types of
| boats.
| ralegh wrote:
| Interspecies communication is a massively underrated field.
|
| We've bridged human cultures in the past, which is easier because
| humans do similar (ish) things, we can use sight, touch, smell,
| etc to establish common ground.
|
| We can communicate simple things with pets, though in my
| experience they learn from body language and intonation,
| understanding grammar and language feels like a step further.
|
| What's the common ground with whales?
|
| Like how eskimos have 100 words for snow, whales could have
| thousands of phrases for water, currents, temperature, storms.
| Fish, migration of different species. A language of relative
| position needed for pack hunting. They might tell stories about
| El Nino, earthquakes, tsunamis.
|
| If they have social structure we may share ideas of
| relationships, friendship, giving (food), owing, sharing,
| helping, etc.
|
| We might be able to correlate their speech with weather patterns
| and animal sightings. We could probably start a two way
| communication, I wonder if us or them would have better forecasts
| for sea conditions. They could act as a network of hundreds of
| thousands of sensors.
|
| Sperm whales travel so far, that even without maps they might
| know the shape of the continents.
|
| Very excited for the future.
| asddubs wrote:
| From what I understand that 100 words for snow thing isn't
| really true, it's just that the language uses compound words
| like german, so it's just that an adjective + noun is rolled
| into one word (e.g. powder snow would also be pulverschnee in
| german, but it's really just the word for powder and snow
| without the space)
| ralegh wrote:
| Phrases, then.
| pavlov wrote:
| Ok, so let's say there are ten qualifiers and ten base words
| for snow. That makes 100 compound words.
|
| It's still substantially more snow-related vocabulary than in
| German, it seems to me.
| FabHK wrote:
| More than you ever wanted to know:
|
| The snow words myth: progress at last - https://languagelog.l
| dc.upenn.edu/~myl/languagelog/archives/...
|
| Bad science reporting again: the Eskimos are back -
| https://languagelog.ldc.upenn.edu/nll/?p=4419
|
| "Words for snow" watch -
| https://languagelog.ldc.upenn.edu/nll/?p=3497
|
| The Great Eskimo Vocabulary Hoax - https://web.archive.org/we
| b/20181203001555/http://users.utu....
|
| "Eskimo Words for Snow: A Case Study in the Genesis and Decay
| of an Anthropological Example" -
| https://www.jstor.org/stable/677570
|
| https://en.wikipedia.org/wiki/Eskimo_words_for_snow
| robocat wrote:
| A list of 65 English words/phrases for types of snow by a
| skier: https://skimo.co/words-for-snow but missing a few
| "spring snow", "Sierra cement".
| DizzyDoo wrote:
| For anyone wanting a short piece of fiction about human-animal
| communication development, I enjoyed this one (or part one of
| one) from Lars Doucet: https://www.fortressofdoors.com/we-trade-
| with-ants-a-short-s...
| mrec wrote:
| For a blast from the past, Ursula Le Guin's _The Author of the
| Acacia Seeds_ is also fun:
|
| https://xenoflesh.wordpress.com/wp-content/uploads/2017/03/u...
|
| (Note: PDF)
| frellus wrote:
| "Gracie is pregnant"
| pegasus wrote:
| No mention of birdsong in this article leaves me hanging. Surely
| birdsong has got to be much more studied, and must contain lots
| of complexity ripe to be plumbed for hidden meaning.
___________________________________________________________________
(page generated 2024-05-12 23:00 UTC)