Dreaming in R
A Psycho­linguistical Statistical Reverie

by “Doc” Yoomin Terrean
(with assistance from 4 ≥ n ≥ 1 anonymous psycho linguisticians)

I had been eking out a living as an under-employed linguist when, taking advice I got from strangers on the internet, I decided to try to improve my computational and corpus linguistic skills to enhance my marketability. One of the elements of my multi-faceted, multi-pronged approach^* was to take classes in the statistical computing language / environment, R. At first it was challenging, but rewarding, and there were lots of pretty pictures.

Then I started dreaming in R. As are many linguists, I’m a semi-failed semi-polyglot (I speak π languages, too!), and I’ve always looked forward to the point in my language studies where I begin dreaming in the new language.

With R, though, I was not experiencing the feeling of enhanced fluency and heightened savoir faire that I have with natural human languages. Rather, mathematics rose unbidden in my brain and did a nightmare dance. Strange numbers and symbols, and horrifying charts and graphs invaded my slumber, and I slept furiously.

I asked a number of psycholinguists for their advice, given their general acquaintance with both statistics and language acquisition. I did not find the solace I sought, but—given the ever-increasing prominence of numerical, statistical, and computational methods in the un-under-employed segments of our field—I thought I would share their collective counsel with a wider audience.

• “You’re obviously a complete nutter.”

• “It’s no surprise that the same unconscious mechanisms that consolidate and integrate foreign language knowledge are at work in processing programming language knowledge, since both represent ways of understanding and structuring ideas and information. The primary difference is likely caused by the difference in auditory and visual perceptual modalities. When dreaming in a foreign language at a level above your actual competence, your brain can seamlessly fill in the missing semantics as part of the process that ‘smooths over’ the inconsistencies and absurdities of the dream narrative. This ‘smoothing over’ is perhaps less effective with nonsense equations, graphs, and charts because there are no obvious narrative clues to provide the meaning. I wouldn’t worry about it too much if I were you.”

• “Why on earth would you learn R? Sheesh. Learn a real programming language like [REDACTED]^†!”

• “You are clearly (p < 0.001) out of your mind.”

• “Oh yeah—rDream() returns a list, and getting specific pieces of information out of those is a nightmare. You want rMatrixVision(), and be sure to pass it prolonged.fasting=TRUE”.

• “Dreams don’t mean anything, and your question has nothing to do with psycholinguistics. Get over it and get a job, hippie!”

• “I feel you. I, too, have had fragmentary R-dreams; they involved left-pointing arrows and a sense of futility.”

• “You’ve lost your marbles. Maybe you have one left, but if so, it’s flat on one side and can’t roll around as much as it used to.”

• “You have to install the dreamR package—it has a little wrapper function named neurosr() that’s basically neurosis() but with the emotion vector first and the neurex pattern second—that way dapply() can get to the emotion vector. I have no idea why neurosis puts the neurex first, but at least there’s a fix.”

• “You’d have purer dreams with Haskell, and they wouldn’t change any of your memories.”

• “Have you tried coercing your dream back into character vectors? You’re probably getting factors already.”

• “The charts you describe in your dreams have badly labeled axes, poorly chosen color schemes, and no clearly stated purpose. You need to spend less time sleeping and more time learning about statistics and data analysis.”

• “R tries to pretty up the dreams by presenting just the first four details of the floating-imagery representations, but after three or four metaphorical juxtapositions, you start getting all sorts of consistency errors. Try using inverse log imagery instead; I always use maple, and put a branch on one end to make sure I can tell when it’s inverse.”

• “R? What is wrong with you? Real scientists use MATLAB!”

• “Stop asking me the same damn question!”^‡

As luck^§ would have it, as I was leaving the Psychology Department I happened to take a wrong turn into the office of Winniphred Phroyd, Computational Freudian Psychologist. She talked to me about my problems, asked me about my childhood and my motherboard, and she helped me understand my p-value–envy. That helped, but not really enough.

In the end, though, she solved all my problems, after a fashion, when she showed me the following R-or-Shock™ Personalized Graph-o-Blot℠ Multiplot:

I don’t know what it means, but it gave me quite a shock all right, and I had a seizure right there in Dr. Phroyd’s office. When I came to, I decided to abandon my quest for statistical, numerical, and computational enlightenment. Instead, I’ve decided to devote myself to a life of abstract theoretical asceticism as a monk in the order of The Brothers of Austere Syntax—free from the temptations and evils of data, analysis, and anything having to do with numbers. I will soon take an official and sacred vow of innumeracy, and I’m even going to swear off subscripts. Nothing more concrete than Greek letters for me.

Good luck to the rest of you in your wordy, worldly pursuits, though. Peace out. ✌ ☮ ䷊

	The SpecGram Inquisition—Pete Bleackley—with Inquisitor Generalis Jonathan Downie
	Decrypt the Missives from the Cosmos—A Puzzle in Many Parts—Trey Jones
	SpecGram Vol CLXX, No 3 Contents