Language and Meaning

by Tom Pittman


Professionally, I am a computational linguist.

In the world of linguistics (the study of human languages), a computational linguist is a linguist who uses computers -- usually in the sense of word processors and databases to manipulate data otherwise organized, but increasingly also with reference to sophisticated computer tools for doing linguistic kinds of things such as analyzing the phonology of a previously undocumented language. In that sense, I am a computational linguist because I am developing computer software to facilitate Bible translation into remote languages.

In the world of computers, a computational linguist is a computer professional who uses linguistic theories originally developed for the study of human languages, but applies them to the artificial linguistic world of programming languages. In that sense I am also a computational linguist, because my PhD dissertation emphasized the preservation of correct semantics (meaning) across the translation of computer programs from (just barely) human-readable languages like Basic and C++ and Fortran and spreadsheets, into the arcane ones and zeros of machine language. Semantics in this context signifies that the program which the programmer envisioned and wrote, representing the precise and sequential manipulation of text and/or numbers, should in fact behave exactly as intended when the computer operates on those numbers as directed by the translated machine language which is the only language the computer understands.

The translation process effected by compilers in the computer environment is exactly the same as the process the same word "translation" denotes out there in the world of natural languages. The only difference is that in the computer environment the languages are artificial and precisely defined, while in the natural language world languages are ambiguous and changing, and meaning tends to be only an approximation. Meaning is only an approximation in the computer world also, but we pretend otherwise.

The purpose of this essay is to explore the nature of meaning and how it is conveyed by the linguistic elements we use to that end. I can identify three such linguistic elements, which for this paper I will identify as Syntax, Semantics, and Context. I intend to look at these elements in effecting the transmission of a Message from the mind of the Author (or speaker; the medium of transmission is irrelevant to the consideration here) to the mind of the Reader (or listener), without any prejudice as to whether either party is in fact a machine. Of course the "mind" of a machine is somewhat metaphorical, and refers more precisely to the effect of the Message on the physical conditions (hardware and software) driven by that message in the case of a computer Reader, or else the physical conditions (sensors, other software) whose interaction generates the message in the case of a computer Author.
 

Message

It is the nature of the message that it is intended to communicate something from the Author to the Reader not previously known by the Reader. Intentionality is important here, but in the case of a computer Author, we reach through the programming to the intents of the programmer who wrote the program acting as Author, or the engineer who designed the physical elements causing or creating the Message. Thus a pressure dial on a steam boiler communicates a message in pounds per square inch because the engineer who designed the boiler determined that the operator needs to know that pressure.

A graphing calculator generates a message which is a curve on its x-y coordinate screen because the user intended to see the shape of that curve and therefore entered the terms of the formula into the program; furthermore, the programmer who wrote the calculator firmware intended to make that curve visible to the user, as did the engineer who designed the pixels of the screen to be controlled by the firmware. The Author of the particular curve is the user who chose the terms, but the Author of the visual message carrying that curve was the programmer and engineer working together.

Some messages -- such as art and music -- communicate only nonverbal emotions, for that is what the artist or musician intended. Some messages -- particularly jokes -- are intentionally ambiguous: the clash between the two senses of a pun is the nature of the message. Some messages can carry multiple non-clashing meanings, such as a fine illuminated manuscript, which contains both text (what the words say) and the artistic visual presentation of the calligraphy and miniatures.
 

Syntax

Syntax refers to word placement, punctuation, and inflection, whatever is required by the grammar of the language to produce well-formed sentences which make sense in the Message. In English the pseudo-sentence "blue wind in Hit the." is ungrammatical -- so much so that it is impossible to make any sense of it. Nothing is communicated. English -- indeed most languages, including most computer languages -- has enough redundancy built into the syntactical requirements that many ungrammatical sentences can be partly understood anyway. I had to put significant effort into constructing a truly meaningless example here.

A more reasonable example is "The boy hit the girl" which has a different meaning than exactly the same words in a slightly different order, "The girl hit the boy". In other, more fully inflected, languages like Latin or Greek, the word order is irrelevant to the basic meaning, because the noun inflection (also considered part of the syntax) tells the Reader which word is the perpetrator and which is the victim. In English we inflect the noun only for plural and possessive case, but verbs are inflected for aspect and tense, such as past or continuative.

In our boiler dial example, the position of the needle is the syntax.

Semantics

Semantics refers to the meaning of the words, typically the dictionary sense. In the boiler example, it would be the numbers on the dial, which thus assign a pressure value to the angle of the needle. Artistic or musical messages might have no established semantics, if the artist is not intending to evoke a particular emotion. Most artists do have that intent, so they choose their colors or tones carefully to give the intended result. The colors and tones are syntax, but the emotions are semantics. In different contexts red can communicate an angry or cheerful emotion; blue usually communicates peace.

Context

Context is whatever body of knowledge shared by the Author and the Reader, and to which the Author makes implicit reference. This is best seen by a couple of examples.
A man and his wife are packing to go on a trip. She tells him that her carry-on valise is ready. He replies, "The car is open."
What message did he intend to communicate? Something about the state of the vehicle? Hardly. She probably already knew that. He was inviting her to carry the valise out to the car. The shared knowledge he implicitly referred to is the fact that to go on a trip, you need to first move the luggage from the house to the car, and if one of the two is otherwise occupied (with his own packing), it is reasonable for the other to start carrying.

I recently finished reading Neil Stephenson's Cryptonomicon, a novel about second- and third-order efforts to hide messages from prying eyes. The first-order effort is to encrypt the message, but if the opponent knows you are encrypting the message, he can use various known methods to break the encryption. Then, of course, the attacker wants to keep the primary agent from realizing that he is being spied upon, so he in turn uses third-order strategems to conceal that fact. It makes a marvelously complex story line. Not far from the end, the hero Randy has been thrown into jail with his laptop computer in an poorly concealed effort to get him to decrypt some vital information on his computer, which the opponent is surrepticiously monitoring. Into the adjacent cell another character -- sort of a guardian angel -- has managed to get himself thrown. He now tells Randy:

With any luck, Randy, you and I can make a bridge -- as long as you are just standing there pontificating anyway. [p.785, author's italics]
The bridge, he subsequently tells Randy, refers to the game. The point of this strange sentence has nothing at all to do with the dictionary sense of any of the words -- not even the game of bridge. Its sole purpose is to communicate to Randy that he should now begin the process of communicating by means of an encryption technique called "Pontifex" previously sent to him by email, which uses bridge hands in a card deck to encode and communicate the decryption key, so as to conceal from the jailhouse eavesdroppers that they are doing so.

In both of these communications, the entire message is contextual, in one case as an abbreviation, and the other to hide its intent from third parties.

These examples are extreme. Most of the time context provides only part of the message. How large a part varies from time to time, and from context to context, but never is it completely absent. Indeed, the dictionary both Author and Reader carry in their heads, which defines the meaning of the words (semantics), and the grammar of the language they communicate in (syntax), are necessary and intrinsic parts of that context. Beyond that, the additional communication context (including facial gestures and tone of voice) disambiguates words with multiple senses.
 

Balance

Some people place the highest value on the context carried by nonverbal channels such as facial expression, and expect it to override the plain sense of the syntax and semantics; others (like myself) often try to capture as much of the meaning as possible in the syntax and semantics, so that context mostly becomes irrelevant. Even when context is intended to be a large part of the message, if the shared information turns out to be smaller than the Author expected, he might recover from the transmission error by retransmitting the same message in a more context-free mode. In the Stephenson novel, the key words were italicized to draw attention to their contextual referents. When Randy expressed confusion, additional material was added to emphasize the contextual intent. This was of course mostly for the benefit of the novel's readership, so we would correctly understand the message being communicated. Stephenson then goes on to describe Randy's efforts to decrypt the first message received, so that nobody is left in the dark. Thus Stephenson begins with a wholly contextual message, which he repeats several times using more and more explicit semantics, until nothing is left to context but the dictionary.

Legal documents by nature must leave as little as possible to context. Otherwise some zealous judge may presume to invent a context the Author did not intend. Poetry is often the reverse, where the poet seeks to evoke a great deal of context in the mind of the Reader -- sometimes with odd side effects. I think it was a Browning sonnet, when the poet was later asked to explain it, he replied, "When I wrote that, only God and Robert Browning understood it. Now only God does." [quoted in the Amazon review of The Barretts of Wimpole Street]

I have a particular interest in Bible translation. I studied Greek and Hebrew so better understand exactly what the Author intended to communicate. Apart from a few badly translated words of little consequence, most English Bible translations are reasonably accurate. Very little is lost in translation. There is, however, a certain amount of cultural context that the individual writers (Author) assumed, which is not in the text. Most of the imagery in the Apocalypse directly links to Old Testament texts. Thus the "Mark of the Beast" (666) can be easily understood to have commercial implications when you see that same number referring to the amount of gold imported into Solomon's kingdom in one year within a context of runaway inflation. Similarly, much of Jesus' ministry has implicit links to Old Testament prophecies. Sometimes the gospel Authors recognize their Readers' unfamiliarity with those references and directly cite them; other times the Reader is just expected to figure it out. Mostly the Authors knew they were writing to a general audience, so what they left to context is minimal.

I try to do the same with my writing. You need to know I'm a Christian, so occasionally there will be a covert reference to the Bible in my writing. Most of the time I eschew unstated context, and it annoys me when people try to assume I will figure out the context in their cryptic communications. Thus I tend to favor email over telephone calls, so there is no back-channel contradicting the intended message. When somebody insistes on verbal communication with me, it usually tells me that they are not thinking clearly and seek the protection of "he said, she said" deniability for their blunders.
 

Tom Pittman
First draft 2008 October 23