German lecturer with Ph.D, ..., I started using Python and R, .and stumbled about emojis ...

Wednesday, May 24, 2023

Image flow, image row. Understanding emojis? with Vilém Flusser

Little pictures in between


We used to send letters to each other. We used to write down, letter after letter, word by word, what we hoped would represent our feelings and thoughts. Only in rare exceptions would we have drawn little sketches on the paper, maybe for love or for a child.


When we receive letters nowadays, usually emails or ‘messages’ on social media, pictures pop up or we find photos or videos attached. When considering only the text, we have to understand emojis if we want to get the message.


Something in our overall communication has certainly changed, but the world of the written word itself seems to be altered. We find pixelated images between the written letters. Punctuation vanishes and leaves in its place little pictures 😔


Lying somewhere between character and image, emojis seemingly traverse the border of the alphanumeric code that we, in the Western world, have been using for approximately 4000 years now. Something in the predominant code seems to be changing, as pictures have reentered our written sentences.


When asking for the significance of the diffusion of emojis, we have to consider their habitat. They do not appear on handwritten or typed letters. They take shape on the monitors of computers and mobile phones. We know how these objects are used. We see bent over or

hunchbacked people at bus stops, on subways and, yes, driving their cars—they are incessantly watching their cell phones, scrolling down and typing now and then.



Absorbed by the flow


A student of mine (I will call her Rose) never looks at anybody—not at me the teacher and not at her classmates. She is always looking at her phone. I tell her to ‘put it away, please,’ and she does. But as soon as I start talking to somebody else, she is again looking at her cell phone.

Maybe it is my fault, but with my colleagues, she does the same.


Somehow, while sitting in the last row, she is still listening to what is happening in class and when questioned—after some minutes and with others talking in the meantime—she sometimes answers.


Once, I decided to write to her, instead of talking to her, using Facebook Messenger. Her reply arrived within a few seconds.


Rose is absorbed by the machine. Her primary channel of communication with the outside world is her mobile phone. I know that Rose views images and short texts with emojis. From time to time, I ask her, ‘Something interesting?’ She answers with responses like, ‘I am watching pets now!’ One image after the other; one series of pictures after the other.


What is happening here? Rose is both in and not in class. She is somewhere partly outside of our common space and time. What she perceives, we could describe using the words of the communication philosopher Vilém Flusser (1920–1991):


‘One adventure after the other emerges from chaos and appears on the monitor. So what matters is […] mainly to progress from surprise to surprise, from adventure to adventure,’ (Crisis 12).


Flusser described this new world of appearing images as early as 1990. According to him, the phenomenon means (as he explained in his essay on ‘The Crisis of Linearity’) the end of an epoch of human history.


The title of Flusser’s prophetic essay might be somewhat of a surprise here. Rose’s world, as an endless flow of images and texts and continuous scrolling down from post to post, seems to be perfectly linear. But the linearity that Flusser wrote about was that of written texts—rows of words. Perhaps we should talk about different types of linearity and introduce some distinction into Flusser’s concepts.



Types of linearity


There once was a German philosopher who used to say, ‘Give small change,’ when students started using very abstract concepts. When reading the inspiring texts of Vilém Flusser, this could be good advice.


In his text ‘The Crisis of Linearity’, Flusser discusses the destiny of human culture—the sense of writing, painting and pixelating, as well as the mechanism by which cameras work— jumping from the stone age into the upcoming millennium. Moving step by step, we should be

able to obtain a clear idea of what he might mean by the ‘crisis’ and precisely what the term ‘linearity’ is referring to.


Flusser starts by discussing a ‘line’ of written words. Try to imagine the scene:

I am writing. Letter after letter, word after word, I create a row of signs. Maybe you, as you follow the lines, will understand what I am saying. Afterwards, you could, if you wanted to, produce an answer or a comment by forming new rows of words.


If we had lived some forty thousand years ago, I would not have written; rather, I would have sketched a shape into the sand or drawn a picture on a stone wall in some cave. You would have watched me as I drew the picture and then you would have wiped it away or covered it

with your own drawing. You probably would not have painted your version to the right or left of mine simply because we would have had no reason for moving onward in a line when trying to express something. We know from old cave paintings that one artist did indeed paint over the picture of his predecessor.


The creation of lines when expressing ourselves is the difference between us and those who first preserved human expressions. We put one letter, one word, after the other. Why do we do this? It does not depend on an eventual difference between letter and image.


Using some pictorial sign from a hieroglyphic system, we could represent ‘I/me’ with a small image of a writer or a feather. You would see my sign and you would not cancel this small draft. Rather, you would look at me with raised eyebrows and ask, ‘What?’ I would then have to explain. I would add other signs, such as ‘will’ and then ‘go away’.


It is our need for explanations that makes us write lines of words. Words complete their predecessors. Flusser says that writing in lines is analytical. We do not watch a picture and perhaps correct it; instead, we explain what we see, word by word. We may write down what

is right and what is wrong with a picture. ‘The objective of linear writing is to critique the imagination’ (Crisis, 4). Writing, Flusser asserts, gives rise to enlightenment.


In some remote moment of our history, we left the images behind and started writing rows of words, translating and critically dissolving the images. This must have been a fundamental change ‘because our thinking, feeling, desiring, acting and even our perceiving and

conceptualizing are to a high degree shaped by the structure of the code in which we experience the world and ourselves’ (Crisis, 1).


Our word lines are endless, pointing into the future. Upon passing to a linear mode of expression, according Flusser, we must have begun to see the world not as a stable scene, but as something on the move.


‘The mood of the eternal return of the same (the magic mood) is

replaced by the dramatic mood of linear progress. Differently said: the alphabetic critique of the imagination leads to a linear, causal explanation of images’ (Crisis, 5).


The linearity of our writing, Flusser affirms, entails a linear vision of historical time rather than the ancient circular idea of time. With this, he does not say that our everyday conception of time, ordered in past, present and future, was due to the introduction of linear writing.


Probably, even in the times of mythical thinking, people distinguished the actual from the past moments and both actual and past moments from those to come. As Kant writes, time is a form of intuition or apperception (‘Anschauung’) given a priori. Without the principles of ordered succession of past, present and future, we cannot even imagine anything (Critique of Pure Reason, B47). Expressing the same idea in linguistic terms: It is very improbable that there had ever been a human language that did not, or does not, discern between present, past and future. There are different ways of doing so—for example, with words like ‘tomorrow’ or with accentuated endings like ‘à’—but the distinction will exist.


What may have changed is the general idea of time. Do we see it as a circle, as in nature, where birth is followed by growth, then by maturity and in the end by death, felt as a return to where we came from? In ancient times, this may have been the predominant image, and perhaps it still is somewhere, as in the world of the Russian peasant that Tolstoy described. In dominant Western culture, it certainly is not. Indeed, we see and study what happens in life and in history on a timeline that points into the future. We also know the inconvenience of this image: we do not know how to handle death. It just happens at some point in our timeline and then time moves on with somebody missing.


But Flusser uses his concepts quite hastily. Taking a closer look, it becomes clear that the concept of linearity alone is not sufficient for analysing the effect of linear writing. We see it as linear, but we know it is not.


We could ask linguists, who will tell us something about the deep structure of the word order, about sentence construction and dependencies. Even a program like the transformer in machine learning takes linear input and gives linear output—but what it does in between is certainly not inear. Arguing conceptually, we could advance that, within their linear order, if words are supposed to explain the preceding ones, they must refer back as well. The new word is understood because it links to the preceding ones, and it will modify their meanings. This is how Flosser’s ‘enlightenment’ works.


Thus, our text line has a double movement. In order to move on, it has to point back. That is exactly the sense by which we understand the movement of history. Only from the outside does it seem to proceed on a line, point after point. But we would not understand a single point just non its own.


If we tried analysing the ‘pointing back’ properly, we would have to look at the sense of this ‘pointing’, which means something like distinguishing itself from the other and yet continuing it. Certainly, the written line of words manifests a complex linearity.


Returning to the flow of images that my student Rose exposes herself to, the difference is now obvious. The pictures of pets she is watching are not pointing back. The little white Siamese cat does not refer back to the pink rabbit she had seen before. The new picure does not explain

or modify the preceding one.


Again, we should add that there is a structure behind this. Depending on Rose’s prior clicks and likes, some program proposes new pictures to her. Per se, there is a link between the posts, but it is not for her.


Reformulating Flusser’s thought: We are entering the world of a new form of communication. We will perceive endless chains of images, of substitutions without any reference to past and future, without explanatory effects and without analytical evolution.


Linearity will no longer be the product of a composed movement. The sense of linearity is changing.


If this is the future, then thought will have to find new ways of manifesting itself.



Emojis in between


What is the place of emojis in the evolution from complex to simple linearity? As pictures embedded in texts, they could seem to be small traitors—little intruders into the remaining realms of analytical thinking. In automated analysis, without any hesitation, they are instead assimilated into texts. Programs translate emojis into words. Websites like Emojipedia do this as well. ✈ is a substitute for aeroplane or travel, and 🦋 for a butterfly or happiness. However,

this view does not do justice to our small, pixelated companions.

Emojis behave like pictures, not like words to be defined in a dictionary. 


Empirical analysis of Italian tweets between July 2022 and May 2023 shows that ✈ is used for travel, but also for joy and as an intensifier, while 🦋 in Italian tweets in 2022 even became a proper name, together with 🐻. Some emojis acquire importance in certain cultures, such as the 😁 in Poland, where a monument has been erected for this emoji. If we look at it closely, do we not notice a note of sarcasm in this otherwise nice and friendly emoji? The 😂 seems positive, but if you use it like this, you might be identified as a boomer. Some of my students say they are using it ironically now. Some others assert that they do not


On the other hand, emojis stay within the line formed by letters. They even appear in groups, staying in line. To obtain some examples, I searched 100 Italian tweets with single emojis. I did this in Italian because I have some experience with this material. Twitter in Italy is usually dominated by gossip, which means a high frequency and variety of emojis.


 Repetitions are quite common. Searching for ✈️ (May 2023), for example, I got 💩💩🤮🤮. Each of these elements could stand alone and would still deliver the meaning (without intensification).


Take 🤭"😂😂'😂'😂'😂'😂'😂✈️✈️✈️✈️✈️⚰️ instead. If we took only a single element— perhaps the coffin—we would understand that the idea of death is involved. But, looking at the entire sequence, we see many tears of joy at the beginning. The aerplane could mean a

journey, but in Italian tweets, it often indicates happiness (‘makes me fly’). Taken together, a proper translation could be: ‘Joy and happiness are so great the writer could just die.’


There are other possibilities, of course. The important fact here is that emojis behave like words: they refer back (and forth). This is a representation of complex linearity.


It would be easy to find other examples, for example, by looking for tweets with the clown. Here, I observed the interaction between words and emojis, as in 🤡di💩. Another reall interesting looking one is 🐌<💩🐌, but I have no clue what it means



Intelligent emojis?


Things change with time, with the language and culture in which you are moving. Emoji use also changes within small groups, even individually, touching the limits of communication and demonstrating the nature of human beings. From time to time, one of us moves against probability. The others watch and see whether they can interpret what, at the moment, they do not understand.


A renewed Turing test criterion could sound like, ‘I will believe I am communicating with anhuman being if it uses an emoji that I would not have predicted but that I can, sooner or later, understand.’ This would obviously be nonsensical, as we interpret everything.


Things would be easier if we avoided the a priori distinction between human beings and machines. There is something very mechanical in most things we do (see E.T.A Hoffmann’s ‘Der Sandmann’) and something human in what machines produce. We could instead ask what it is that we understand as intelligent. It is not the production of the predicted or of nonsense.


In other words, if our students nowadays let their assignments be done by GPT-3 or 4, something is wrong with the assignments.




Literature

Vilém Flusser: Crisis of Linearity (Transla7on from German by Adelheid Mers)

German text “Krise der Linearität”, published in Absolute Vilém Flusser, Hg. v. Nils Röller and Silvia Wagnermaier, Freiburg: orange-press 2003. Found at:

hSp://bootscontemporaryartspace.org/blog/bootprint/

hSp://Vaul-dcnm.pt/joelfilip/docs/flusser.pdf

Sunday, February 5, 2023

Heat Stains on Twitter (first observations)

In order to find spots where twitter gets hot, with people writing down emphatic statements or opinions, we could start with emojis, as the creation of hashtags is not predictable, while the number of emojis is relatively limited. 

Yet, by the most common packages for Python and R we still get between 1400 and 5000 emojis. These still are far too much, for me, to loop through these lists and make a tweet search request for each of them. I decided to begin with my own list of emojis which will be enriched.

The emojis used in a very lively Italian gossip group ("#jerù"), where often anger about the stars is expressed, will be the first ones, already 34. For the beginning, I will keep within the realm of Italian tweets. As the case of Polish 😆 teaches, the use of emojis largely depends on cultures defined by languages. 

My first Italian list is : 👀 🐻 😅 🌜 👑 🌈 ☕ 🤍 😍 🥰 ♥️ 🤦 ❤️ 🌲 😂 🦋 📸 🤣 ⛰️ 💜 💚 ♀️ 😁 🔥 💖 💗 🙃 😋 🕛 😎 😭 😜 🌺 ✨.

Looking for signs of excitement, I will keep record of tweets searched with my emoji list only if on 100 tweets I get more than 15 exclamation marks. Angry and happy people love doubling and tripling these signs. 

In a tweet search done on November 12, 2022, the most prolific emoji was 🤣

In 100 tweets with this emoji, it appeared 231 times. Exclamation marks: 27 on 9234 characters, with eight double "!" and three "!!!!". 

The most frequent hashtags were 

"#taleequaleshow" "#merito" "#konpetenza" "#ottoemezzo" "#calenda" "#novax", 

i.e. two about political tv shows, two are hashtags used by a journalist (book title: “Damned Pacifists”), two about politics, with the famous hashtag "novax". Maybe this could indicate the right way to find angry people. Should we move on following the hashtags?


A hashtag "novax" search (07/ 01/ 23, n=100) results in only 15 emojis,

💯 🔝 ⚧️ 👋 👿 💉 👇 🐧 💳 💩 🤣 🤡 🏳️ 🇪🇺 🪳

with only four exclamation marks. Where has the excitation gone? We see ten times 💩, following 35 🤣.


The same for "#calenda" (an Italian politician). A search results in 18 exclamation marks, with three times double "!" and one "!!!". But, according to R and the relative emoji package, only five out of 100 tweets contain emojis, 44 in all (26 unique). Five ➡️ , five 🇮🇹, five 🇪🇺. Is it that in politics, Italians use only few emojis? This could be due to the higher age of people interested in politics here.


Let us try with other "controversial topics". A "#Salvini" search (right wing politician) results in 11 on 100 tweets with emojis. 22 🤡 , 18 😂, 13 👏 and five 🇮🇹. Other "hot" topics like #bce (European Central Bank) or "nosbarchi" (no acceptance of refugees) give similar results. People get angry about politics, but do not use emojis in these fields.


Still, by searching with emojis we can find angry people. The clown 🤡 is used when people find laughable something or somebody. Again, in Italy, we get top hashtags about soccer and about the Reality Show Big Brother VIP.

We see

feature frequency     rank 1 🤡 432      1 2 😂 72      2 3 🤣 38      3 4 🤮 26      4 5 😅 16      5 6 👇🏻 12 6 7 💩 12 6

People are angry. Searching for anger with emojis around could be helpful.

The number of unique emojis here is only 23. 
🤬 🤣 ✅ 🤦 😏 😱 😡 ⚫ 🖕 😵 💸 🤡 😂 🤢 💩 🤮 💫 ⏩ 😁 📹.
Maybe this should be the basic list of emojis, on the search of angry people, i.e. heat stains on 
twitter. 

With the middle finger, for example, tweets supposedly are quite aggressive. With an apisearch (n=10) on November 12, 2022, we receive antisemitic content, but only two exclamation marks. This number is again growing with 🤮, becoming 7 in ten tweets, and 2 double "!". 😡 gives eight exclamation marks (two double), tweets about animals rights. The 💣 brings tweets about conspiracy theories, without any "!"  
😡 search gives 42 emojis in 10 tweets, namely
🤬 🤣 💯 👇 😠 💪 🤦 ➡️ 🙏 😢 👌 ♀️ 💥 😤 👿 😱 😡 🤔 ❣️ 🇮🇹 🔻 🖕 ♂️ 🔴 👎 😈 🤞 😂 🤢 💩 🤮 😅 🥲 🤪 🥺 ‼️ 😖 ⏩ 🤨 😁 📹 ☕
The exclamation marks are here, not among the 
characters. We should getting to know emojis by the company they keep. 

Emoji diversity


For comparison, a ☕ search results in 62 unique emojis and 421 total emojis, lexical emoji diversity  .147. Lexical diversity among emojis is low in all cases, maximums are .23 with 🔥, .21 with 😎, .22 with the ❤️ and .2 with 😜. This does not depend on the simple number of emojis. (for example 254 correspond to .10, 601 to .15.). The variety of accompanying emojis rather seems to be a characteristic of the emoji itself.




technically

(ok, it is not elegant, but I had learned programming with Algol W, in 1975)

import pandas as pd

import tweepy

import csv

import emoji

import emojis

import regex

from collections import Counter

#https://stackoverflow.com/questions/49113909/split-and-count-emojis-and-words-in-agiven-string-in-python


<authentication stuff>


api = tweepy.API(auth)

with open('emojilistit22.csv','r') as mine:

    leser = csv.reader(mine, delimiter=',')

    leserl = list(*leser)

    for kw in leserl:

        print (kw)

        container = []

        tweetCount = 100

        results = api.search_tweets(kw, count=tweetCount, lang="it")

        for tweet in results:

            container.append(tweet.text)

        row_count = len(container)

        print("number of tweets ", row_count)

        filename = "exclamation_basis_" + kw + "2022-11" + ".csv"

       

        f = open(filename, 'w')

        writer = csv.writer(f)

        writer.writerow(container)

        f.close()

        zeichendf = pd.read_csv(filename)

        zeichenkette = zeichendf.to_string()

        lang = len(zeichenkette)

        print("Number of characters: ", lang)

        x = zeichenkette.count(".")

        print("Number of single points (full stops? .)", x)

prozent = x/lang*100

        print(prozent, "%")

        y = zeichenkette.count("...")

        print("Number of three points (ellipsis ...)", y)

        prozent = y/lang*100

        print(prozent, "%")

        if y>0:

            relation = x/y

            print("Relation: ", relation)

        x = zeichenkette.count("!")

        print("Number of exclamation marks", x)

        prozent = x/lang

        print(prozent, "%")

        if x>15:

            anteil = lang/x

            print (kw, "exclamation marks:", x, "on", lang, "characters", anteil, "%")

        y = zeichenkette.count("!!")

        print("Number of two exclamation marks", y)

        prozent = y/lang*100

        print(prozent, "%")

        if y>0:

            relation = x/y

            print("Relation: ", relation)

        x = zeichenkette.count("!!!")

        print("Number of three exclamation marks", x)

        y = zeichenkette.count("!!!!")

        print("Number of four exclamation marks", y)

# just having a look at the emojis, will construct an archive later

        material = str(zeichenkette)

        emoji_hier = emojis.get(material)

        print (*emoji_hier)

        zahlein = emojis.count(material, unique=True)

        zahlall = emojis.count(material)

        print (zahlein, zahlall)

 


 

Image flow, image row. Understanding emojis? with Vilém Flusser

Little pictures in between We used to send letters to each other. We used to write down, letter after letter, word by word,  what we hoped w...