Language is an extremely limited tool (at least) when it comes to expressing thought. A good analogy is the attempt to capture a sunset on a canvas: The result is rarely even remotely comparable to the original, while lacking in color, luminescence, realism, not capturing the way that the red fire of the descending sun dances in the rising waves of the ocean, ... The best results are typically reached when the attempt at an accurate portrayal is abandoned in favor of something else, e.g. an attempt merely at capturing the viewer’s imagination or at emphasizing one particular aspect of the sunset.
This article discusses some of the related issues, mostly with a focus on the frustration that I, myself, often feel. The main focus is on written language; however, many of the same issues apply, m.m., to spoken language. The main difference would be that spoken language is seldom used to express complex thoughts, and thus is less affected by many of the following issues. Another difference is that the “linearity” problem described below is even greater with spoken language.
During a revisit of this page (2009-10-30, adding sections, writing an introduction), I made the pleasant observation that my ability to handle the complications below has improved noticeably since April, when the first version was written. This can be seen by how many of the formulations below stumble on exactly the difficulties the text describes. (I was, however, aware of my own short-comings even then, as can be seen in some self-referencing examples.) While my current style of writing is far from flawless, I suspect that most critique that can be raised today indicates different taste and priorities—not the objective weaknesses of old.
Revisiting in 2023, I am not certain. I am, for natural reasons, far ahead of where I was in 2009, but I still often find language a clumsy tool (or myself a clumsy wielder of that tool). I also have the suspicion that I might make greater assumptions about the ability of the reader today than back then.
A notable issue is that my ability to “find fault” has increased with my greater experiences with writing. Certainly, I am less happy with the quality of my 2009 texts today than I was in 2009—and might well be less happy with my 2023 texts in 2037 than I am in 2023. (This even aside from changes in my own taste and aside from various language errors, e.g. the original use of “viewers” over “viewer’s” above. Going through old texts, I often see some particular issues, including a weird blindness for when that apostrophe is needed. While I am by no means perfect in this regard today, I am far better than back then.)
One of the largest problems is that language is essentially linear: A written text starts with one clause, which is followed by another clause, which is followed by another clause, and so on. Thoughts, OTOH, are non-linear and tend to form a network of interconnecting associations, dependencies, causalities, etc. To cover one relatively small subject matter in a way that corresponds to the original thought(s) implies mapping this network onto a linear structure—which is impossible for any non-trivial set of thoughts. Even when making ample use of parentheses, footnotes, compound sentences, and similar, the task cannot realistically be done—and the attempt can severely reduce the legibility of the text.
There seems to be only two work-arounds for this: Firstly, to simply cut out anything that is not essential to the text, with the hope that what remains will fit a linear structure sufficiently well. (Incidentally, this will ideally lead to writings that adhere to common principles of good reader friendliness, e.g. “keep to the point”. As far as my own writings are concerned, I am torn about such principles, because I am often more interested in catching and developing my own thoughts for me than being an enjoyable writer for others.) Secondly, compromise on the network aspect and bring the thoughts down in a linear structure of paragraphs, sections, chapters, “fact boxes”, etc., while relying on the imagination of the reader to fill in the network (with some help of phrasings like “cf. chapter 8”, “s. above”, “further, ...”). Ultimately, the result will be at most satisfactory, and often well short of that, for all but the greatest writers.
What about professional (non-fiction) writers of various types that fall short of “greatest writers”? In my opinion, many of them are quite poor. Journalists, especially, tend to be very poor writers—and, often, the more so the more renowned they are. (Possibly, for reasons like those discussed in another text.)
Many, journalists or otherwise, fail through giving a too simplistic view, “lying to students”, or similar. Their texts might then seem pleasing on a casual glance—but prove highly unsatisfactory on closer inspection. In an exaggerated example: To describe an ocean as “lots and lots of water” results in something concise, easy to read, and, as far as it goes, correct. The description is also almost useless and misses the point of much of what an ocean is. Consider, among many objections, that someone might conclude that a certain river or lake is an ocean, that no mention is made of salinity, that life in the oceans is ignored, and that the interactions of the oceans with weather and climate see not one word.
There are many authors, especially in the harder sciences, who are capable of giving an excellent and well-structured treatment of a field or topic in book form. However, they still do so through forcing something non-linear into a linear or near-linear form, and (outside the trivial) readers cannot be expected to have a full appreciation until having read the entire book—possibly, repeatedly.
Another issue is the rendering of interconnections: Not only is the number of ways that text can be connected in a paragraph limited, but using the same means of connection several times in a short space leads to texts that are, for want of a better word, silly. To take a specific example: I like to use a colon (“:”) to set off a statement from an explanation (as exemplified by the first sentence of this paragraph)—and often find myself using this means a second time while still being in the first explanation, trying to elaborate on a part of the original explanation. The result is that I have to re-write the text. Similar problems occur with “—”s and parentheses—even to the point that I find myself writing nested parentheses. (Typically, I take this as a reason to restructure the text or cut something out.)
The example with the colon leads to another problem: Many constructs and characters can be used in somewhat differing roles, which makes language unnecessarily ambiguous: It would more or less impossible to write an algorithm to convert between English (or any other natural language I am familiar with) to pure idea (even assuming that “pure idea” could be represented sufficiently well in another form). (By coincidence, the preceding sentences illustrate the point of the preceding paragraph rather well. For this reason, I let them stand uncorrected. (Unfortunately, they fail to include a nested parenthesis, but one cannot win all the time...)) The colon is one example of this, having a general meaning of a “lead in” not limited to statement–explanation, but also including e.g. listings of items (“There are five types of birds in X-land: Imaginary shrike, fictitious thrush, ...”)
Obviously, one of the largest problems is finding the individual words that correspond to a certain concept—not to mention, words that the readers will both understand and interpret in the manner intended by the author. Relatedly, I sometimes have problems finding an adequate translation of a German or Swedish word into English (or the other way around). Consider e.g. the German word “werbewirksam” and the corresponding translations and discussions available from Leoe: The only suggestion available at the time of writing (2009-04-15, “effective in advertising”) lacks the common connotations of sarcasm and deliberate cliche of the original (“politically correct” is an English expression with similar connotations), and is too unwieldy to be an easy replacement. A reasonably accurate translation of what many Germans would mean by “[company] hat eine werbewirksamme Kampagne gestartet” would be “[company] has started a campaign using means by which it hopes to influence the public, but which I consider of dubious efficiency and indicative of a very low opinion of the customers.” (The use in the advertising industry and among executives is usually without these connotations—just like many use “politically correct” with a straight face.) There are simply many cases where a particular concept/idea cannot be expressed by one single word in a particular language.
Another interesting example: How to bring across that two persons carry the same name without using a clumsy formulation like the one in this sentence? (In German, the handy “gleichnamig” is present.) “Namesake” would be the most ready solution (and the one that I have so far used); however, in its stricter sense, it actually refers to one person being named for another (as opposed to e.g. a co-incidental naming): George W. Bush is a namesake of his father, but a random John Smith is not a namesake of another random John Smith. (Other variations in semantics include allowing “namesake” for the elder Bush with respect to his son; and whether a non-Bush George can be considered a namesake, irrespective of whether named for the one of the ex-presidents. Typically, a pragmatic use, where the meaning is given by context, is needed.) No other word that I am aware of (including “eponymous” and “homonymous”) match entirely in meaning either.
As mentioned above, there is also the sub-issue of whether a good match in meaning will be understood by the readers. A specific example is “m.m.”: This is short for the Latin expression “mutatis mutandis”, roughly “when that which should be changed has been changed”—and a handy way of saying “Replace A with B, bearing in mind the natural differences between A and B.”. Unfortunately, while a valid expression in English, it is also an expression that most will not understand. Notably, even the much more common “Latin” abbreviations “i.e.”, “e.g.”, and “cf.” often cause problems for non-native speakers and native speakers without a tertiary education—while those new to the Internet might have problems with expressions like “POV” or “OTOH”.
Then we have the sub-issue of whether the reader and writer actually mean the same thing by an expression that they both know. Words relating to ideology or religion like “conservative”, “liberal”, “feminist”, “communist”, “christian”, ... can vary considerably both in literal meaning and connotation when looking at different countries (even country-internal circles), eras, and contexts. Combine them with other words and the difference can be gigantic (consider e.g. “good communist”, which could imply anything from a dedicated follower of Leninism to a corpse.) Even more everyday and neutral words often have large differences in interpretation.
The single problem that I have heard mentioned most often by others is expressing emotions and nuances from spoken language or body language in writing: Intonation, facial expression, hand gestures, ..., are all missing. This is not something that I personally have ever been very bothered by; possibly because I am more information oriented, possibly because my relatively large experience with writing indirectly colors how I naturally tend to express myself. (A phenomenon that is easy to observe in those used to programming in one language and switching to another, e.g. Smalltalk to Java; and, to a lesser degree, those using a natural language that is not their main language.) Nevertheless, the issue is worthy of mention in this context.
A solution to some problems could be replacing traditional natural languages with an artificial construct. Notably, statements in programming languages typically have a unique meaning and are much better suited to express thoughts—alas, only in a very limited area. The AI concept “semantic network”w provides a good way of writing thoughts down in an at least two-dimensional manner (which allows a greater, if imperfect, focus on the interconnections)—and, after some minor training, it is possible to read such networks just as easily as normal text. Unfortunately, any attempt at developing a new mode of writing would fail on lacking acceptance; in particular, because it would likely not be compatible with spoken language, and only a very small minority of the population have the same needs and wishes as I do in this regard. (It is also notable that comparatively unambitious and uncontroversial new languages, e.g. Volapuk and Esperanto, have had only minor success.)
Various forms of “hypertext” (e.g. on the WWW) can help by connecting different texts and parts of texts in a more complex manner than a steady stream of text. Unfortunately, it is hard to take hypertext beyond being a more user-friendly way to make textual references. In effect, the user is still stuck with a linear text connected to other texts, with the main advantage being that he can manage his own access to these texts in a more convenient manner. It is not suitable for reducing the linearity constraint of one individual text—indeed, trying to do so can easily lead to the text being chopped up into user-unfriendly mini-pages.
The following is an automatically generated list of other pages linking to this one. These may or may not contain further content relevant to this topic.