The Russian KFC Problem
Cognitive vs. Linguistic Problem Solving
Here I will describe a series of anecdotal experiences that challenged my previous conceptions about language use and foreign language learning as an exercise of acquiring certain cognitive abilities. In particular, I will show how it was possible to place an order entirely in Russian with almost no knowledge of the Russian language.
For a long time, I have thought of the process of language learning as a process of reaching an ideal target, that is, when we say someone speaks good “English” or “Russian”, we tend to think of the target language as a set of knowledge and abilities that one gradually masters. And only then will the speaker be able to perform various tasks in that language properly, such as ordering beer, asking for and understanding directions, and having a philosophical debate about the cosmos.
In contrast to this, my newer view is that the abilities of performing these cognitive tasks are never 100%, and they need not be so. Language skills can be largely fluid, fault-tolerant, based on “making do”.
To Russia with Love
We went from Helsinki to St. Petersburg by cruise. On the ship I started learning about the Cyrillic alphabet, and I grouped the letters into several categories:
Group 1: these letters “carry over” from the Latin alphabet, they generally look and sound the same, e.g. T = T; M = M.
Group 2: these letters look similar to Latin ones but have different phonemic value, e.g. C = S; H = N.
Group 3: these consist of letters introduced in high-school physics, which also serves as mnemonics for their sounds, e.g. Д = Delta; Ф = Phi; П = Pi; Р = Rho; Л = Lamda.
Group 4: these are letters that look very different and have sounds somewhat alien, e.g. Ж, Ц, Ш, Щ. Though there is some leeway or robustness for this group, because even if I haven’t mastered them and just assume a certain letter is “some affricate”, the occurrence is rare, and compensated by context, and unlikely to cause confusions.
Once I learned to read Cyrillics, signs suddenly somehow became “transparent”. The string of letters may be very “Russian-looking”, but when you read them out, they become “Business Centre” or “Second Hand” or “Best Travels”.
This paves the first step of being able to “use Russian”.
Contextual Scaffolding
The second step correlates with my experience of learning Finnish as a foreign language. If we look at the teaching materials, the main goal seems to be equipping the learner with the ability to deal with various everyday situations in the language. There are certain set situations or “scripts”, such as introducing oneself or ordering at a bar. One thing I realised quite early on with this “doctrine” is that these exercises may be very “useful”, yet their usefulness is confined to a specific, repeated scenario where the bounds of the interactions are well-defined. There are expected discourse turns following a predictable pattern, in any case, it is not an open discuss. In my personal experience, if the other person asks me something unexpected, deviating even slightly from the “standard script”, I would be dumbfounded and have to switch back to English.
In some sense, this type of language use is non-generative. It is akin to using a vending machine through speech recognition. Sometimes orthography does not even come into the question, for example, in Finland, the menus are not only written with the Latin script, but with direct imports of English words, such as “McNuggets” (this is similar to how a posh restaurant in England may have an item on the menu that reads “Coq au Vin”). In such cases, specifying the order items would hardly be a problem.
The next helpful feature is that all these basic “template phrases” can be well rehearsed, for example, one could learn and master the phrase for “I would like…”, then insert the item names thereafter. Since in the actual situation the menu is generally provided in an orthography that is either learned or transparent, “word prompts” are readily available. In the Russian KFC, the Cyrillic items actually read out as “Sanders Burger” and “Stripes”. Further “prompts” are sometimes provided by the interlocutor, for example, when I ordered the “Chicken Strips”, the next question she asked was: “[WORD], не [WORD]”, and I presumed it to be “Spicy, or not spicy?” and repeated the sound sequence for “spicy”.
Another feature that facilitates the interaction is the relatedness of languages, especially in the applicability of cognates. For example, I could order “three strips” because the Russian word for “three” is simply “три” or “tri”. Even when there are false cognates, the semantic proximity often gives some hints. When the Russian speaker asked, “[…] menus?”, assumed to be “Do you want the menu?”, according to Grime’s maxims, she already knows that I have seen the menu, so she would not be offering the menu, therefore menu must mean something else, in this context, the most likely semantic cousin is “sides” or “meal”, which means fries and drink along with the burger. For all these questions I can answer with “yes” or “no” in that specific language.
The next type of aid would be hand gestures, in a sense “innate grammar”. It is possible that we have directly transparent indications of location, movements, mood, aspects by the proximity and shape of the speaker’s hands. In this particular example, I assumed one question to be “Eat here or take away?” because of the co-occurring gesture indicating proximity when the word for “here” is uttered. Then I simply repeated the sound sequence for “here”. With the combination of gestures or posture or intonation, I could interpret which question might have meant “Anything else?” and respond with the rehearsed “Njet.”
Also worth mentioning is that in real interactions, there are often “situational props”, for example, just at the time usually presumed to be suitable for processing payment, I was holding a credit card in my hand, therefore my preference for the method of payment is declared without the use of linguistic specification.
In summary, the provision of predictable “scripts” and related rehearsals provide even the novice speaker with the ability to handle a very narrow range of situations involving language use. In terms of attention, filtering, and processing, this is comparable to the early days of speech recognition over telephony, the machine is capable of either recognising a small set of utterances, e.g. the number one to ten, but from a large number of speakers; or recognising a large range of utterances, but based on the voice profile of a single person. There is a trade-off between the tolerance and flexibility that can be implemented around how a word or utterance is realised, and the ability to accommodate the different individual variations that exist within a population.
In my particular case, the preconditioning and rehearsal tuned my ability to pay specific attention to certain keywords, with filtering through a noise environment, and with heightened abilities to recognise approximations. This is what happens in a scripted scenario for using a highly unfamiliar language. Whereas in an open conversation between competent speakers of a language, the usage patterns are more robust; the speakers constantly generate predictive script (which are dynamic and live), allowing them more flexibility in hearing a sentence amidst background noise, or reading a text with scrambled words or typos.
