• View
• Edit
• History
• Print

Language

Readings

• Winograd and Flores, Language, listening, and commitment (Chapter 5).

Additional Resources and Readings

• Jorge Luis Borges, The Analytical Language Of John Wilkins. Translated from the Spanish 'El idioma analítico de John Wilkins' by Lilia Graciela Vázquez; edited by Jan Frederik Solem with assistance from Bjørn Are Davidsen and Rolf Andersen.

Ladder of Inference - Mark Schar

Responses

Language and Software – Robert Graebert

The fifth chapter in the book brings home the message that sentences are only meaningful in a shared context as they are not ultimately definable in the objective world. When the shared background is not given, a conversation can break down requiring further dialog. The shared context is important so that active listening can trigger interpretation beyond the conveying of information. The context or background itself is defined shared by the shared being in the world.

In my opinion this is applicable to HCI aspects of software applications. In this case the “speech acts” are between the application displaying information and the inputs by the user. Especially with technical applications it is very common for there to be a breakdown of the conversation between the application and the user. In a way the application is a projection of the background of the designers and programmers on the one hand and the user on the other.

From experience training users on an application the first task is to establish the language with the implicit metaphors the application uses to communicate with the user. Once this is established the user is less likely to get into a breakdown situation: feeling completely lost and not knowing how to proceed. There is no simple remedy such as a starting a new conversation with the other conversation participant to resolve the conflict with the application (ignoring support options in this case). I think this is where the quality of any self-help system integrated with the application is tested.

Only once a common language is (re-) established and the user “connects” with the application the full productivity can be achieved. At that point the application becomes a part of the background for the user not being perceived explicitly but becomes invisible for the execution of a task. At this point, we can start reducing information presented to the user as he/she is able to interpret the utterance of the software.

It seems that in this context Dreyfuss’s “Disturbance” and the “Breakdown” mentioned in the current reading are actually very similar. These are the times when we start becoming aware of the things or conversations and we need to take action to resolve it.

The question I have for Terry is if current research in HCI takes these aspects into consideration? Can we boil this down into general guidelines for bringing new users to an application without overwhelming the experienced user who already shares the context?

Pointing/Evolution - Chris Anderson

Having taken Professor Herb Clark’s class on “Language and Thought”, much of the content of this chapter was review – but with more theoretical background. Prof. Clark introduced us to many of Grice’s axioms as well as Austin and Searle’s formulation of speech act theory, but did so without acquainting us to the fundamental philosophy behind the idea that communication is not knowledge transfer but a social process.

One interesting topic we did cover, however, was the notion of using non-linguistic (in the traditional sense) cues as part of conversation. Pointing, for instance, bears as many complexities in analysis as spoken word. If I say “I want that” and point at a book, do I mean a) I want that exact copy of the book b) I want that book, but it could be another copy c) something about what the book is about or d) something else entirely? In the thrownness of conversation, however, we can disambiguate this instantly. The same sort of understanding must happen even if nothing is spoken at all.

To the phenomenological/hermeneutic view of language’s credit, it does a wonderful job of not turning pointing or any other non-verbal cue into a problem. A rationalistic perspective might have to assign pointing to a new class of actions in addition to speech because their meanings do not necessarily overlap. What “objective meaning” is there to a finger pointed in a direction? Who can say that it is pointing at an object at all?

I am also reminded of a statement that Professor Russ Altman made in the first class of CS274?: “nothing in biology makes sense without evolution”. He is making the same kind of claim as “nothing exists without language”. That is, to say that we can study biology without studying evolution is not really false. It feels like a tautology, or that we are begging the question. Biology is nothing without evolution – it would simply not be biology. Similarly, we could phrase the second assertion as “saying something exists does not make any sense without language”. Existence, when we refer to it in language, is a linguistic concept. The rationalistic perspective tends to separate concepts like these, when in reality they are one in the same.

Language - Forest Peterson

The context of background information to frame language is illustrated by these writing assignments. The website presents the writing assignment as a requirement to respond to one or the other of the weekly topics. But, I have noticed that the topics do not have equal questions prompts, one has had the questions while the other only presents a reading. Based on past context of coursework, I assume, that the default question prompts given on the main page apply to the topics that are not given specific prompts. Falling back on my comment from last week, I could use a random chance to represent this unknown, using a 2 sided die, i.e., a coin, the answer is found.

Obviously, since I am answering topic two with the default prompt, I have thrown a 1. I would like to see this aspect of ambiguity better explicated. Possibly each time conditions such as those presented in “Listening in a background” the listener is throwing die of sides 0-->n, 0 representing no ambiguity and as n-->infinity ambiguity approaches unrepresentable. The key point presented is that a representation is possible to some level, although very small. For example, the listener hears there is water in the refrigerator, based on “sufficient coupling” the listener throws a die and determines incorrectly that there is drinking water, rather than the correct coupling that the speaker is a cad and so they are likely being overly specific, hence, eggplant water. In the same way the reading provides the implication that categorizing representations of the world into a machine interpretable form requires throwing of the die to represent the ambiguity. Machines are so poor at ambiguity that even ambiguity generators are finite.

The representation of the world as a formal model and the empirical observation of the world to develop and/or validate the model. The ambiguity in language and context of this language determines the true observation. Capturing this context to place the observation in the proper frame appears to bee a difficult task. As a field engineer we observed project progress, but we did not formally record the context of the observation. For this reason when a future project is planned it is tradition to include on the planning team a member of the team that had recorded the data, therefore providing the recorded context in their memory and therefore making information. No research has fully addressed this issue to-date.

Language, Intentions and Origins - Seng Keat Teh

I was amused by the points brought up by Habermas, in which he asserts that "the speaker must engage himself, that is, indicate that in certain situations he will draw certain consequences for action" and "that every language act has consequences for the participants, leading to other immediate actions and to commitments for future action". I think this would very much be dependent on the intentions of the speaker, on whether the speaker intended to make a commitment of some kind. It certainly doesn't apply to politicians, especially those caught flat-footed or flip-flopping on a particular issue that they took (or not) a previous stand on. Here, politicians are masters of "speech acts" that are often evasive and ambiguous in nature, neither denying nor confirming a stance in order to maintain political expediency on hot-button topics. In order to reconcile a statement made in the past that conflicts with an act in the present, politicians attempt to keep switching the context and background with which a listener utilizes to attribute meaning to their speech acts. Of course, there is some truth in Habermas assertion that speech acts do have consequences - the listener may not deduce the politician's true stand, but he draws that the politician is being typical in his line of profession.

Searle's division of speech acts into what he claims are the five types of "illocutionary points" may not necessarily be such a clear cut means of categorization. "Directives" certainly stand to be distinctive on its own, but I felt that the categories of "assertives", "commissives", "expressives" and "declarations" to be very much overlapping. For instance, the example used in the reading of pronouncing a couple married - "you are now man and wife", is a declaration. Yet, this statement is also an assertion by the speaker, of the truth of the relationship between the groom and bridegroom, and an expression of the state of affairs regarding the marriage. The statement "I will not let this mistake happen again" is both a commisive and expressive where the speaker (if the speaker made the mistake) has unspokenly admitted and implicitly apologized for the mistake that had happened.

I do agree with the assertion that language and cognition to be fundamentally social. Yet to say the coupling of language to external reality is negligible to that of mutual coupling among language users is disagreeable. The origins of East Asian languages, for example, are symbolic sketches and simplified pictorial representations of real-world objects. The sounds made by a cat, "meow" and that of a duck, "ngap", are themselves used as words to represent these entities in spoken Cantonese. In Western languages, though, labelling of objects may have seemed arbitrary, though with roots in Latin or other ancient Western languages, the Sun could have easily been called Vol or Sol.

I would have liked to see the point made by Heidegger better explicated, that "the meaning and organization of a culture must be taken as the basic given and cannot be traced back to the meaning-giving activity of individual subjects". Did he mean that it was impossible for us to understand the origins of a language from a culture and how it came into being? I would think such information would be useful to confirm the hypothesis that language is much more a social construct than that of an objective external world, since many languages do have their origins, such as East Asian languages, as simplified pictorial representations of the real world and are thus meaning-giving activities.

After class comment --Nate

I have to admit that before this class I didn't realize how often I took for granted objectivity of things. Now that we have this more sophisticated model of how people communicate we can approach human to human interface and human to computer interface problems in a new light. We cant have some kind of universal objective truth, so the next best thing is to make sure that when two things are communicating that their relative objectivity is a close as possible. I think this is what Fernando did in his model accepting and declining tasks. I'll be interested to read more about it.

After class comment - Mark Schar

I have to admit that I had not realized how pervasive the use of metaphors had become in our everyday language. And as Terry had predicted, I'm now double thinking my own speech patterns, somehow believing that speaking without metaphors is a purer form of communication. I also appreciated the distinction between metaphor and schema - which I had not completely understood from the reading. I'm turning over comments about the interrelation of concepts between Heidegger and Lakoff - I can see more parallels than discussed in class.

After class comment - Forest Peterson

The background context providing meaning to language may have an application in communicating project feedback. If the information transfer is given in a known context then this contextual information can be inferred rather than explicitly recorded. This would reduce the data collection resources necessary. The data processing demands of project reporting would be reduced through reduced exogenous inputs, replaced by database inputs or endogenous calculations. In effect the feedback could be reduced to a "newspeak" like in 1984.

After class comment - Chris Anderson

I wanted to wait to post a comment until after we had a chance to meet with Lakoff - and I'm glad I did. A lot of our discussion in class commented on the lack of objective truth in the world, but Lakoff drove that point home. I'm not sure yet what I think about his assertion that concepts like, say, formal logic, are dependent on our physical make-up. That's quite an assertion. It has the ring of plausibility, in the same way that solipsism does, but actually applying the idea that absolutely everything is dependent on the composition of our brains in a useful way is daunting.

After class comment - Jeff Wear

I regretted that I couldn't stay for the entire talk with Lakoff, but during my short time there I was very intrigued by the discussion of his collaboration with neuroscientists at Berkeley to locate centers of "schema processing" within the brain. What a terrific validation of his conceptual theories regarding the grounding of metaphors in our bodies! Our discussion of machinic interfaces last week got me thinking about the potential development of new metaphors grounded in our use of technology. We treat our phones and other gadgets like extensions of our body and I would expect our schemas to adapt to this reality. While I imagine most metaphors are common across participants (who share the same physical makeup), these new metaphors, regarding multitasking for instance, might resonate most with a younger population, and neuroimaging could confirm this.