Note to readers: The post below *only* addresses Bralich’s argument that parts of speech are “discovered,” rather than artifacts of analysis (not his argument about the inertia of publishers, which is an entirely separate issue). It’s a bit long, and even I think it’s boring, but I thought it important to lay out the rationale for disagreeing with that claim. If you’re annoyed by theory-mongering, or think it’s irrelevant, skip it!

 

 

I’m going to break this into separate points, so that it will be easier to pin down disagreements:

 

(1)                 There is no single criterion that can be used to define the traditional parts of speech in English.

 

This is a point that is made in a great many extant grammar texts. Take nouns, for example – although it is true that only what we consider nouns can be made plural, it is *not* true that all nouns can be made plural. In fact, of what are traditionally viewed as the three major subcategories of nouns, only one (count nouns) can be typically pluralized. Mass nouns (luggage) and proper nouns (Abraham Lincoln) cannot be made plural under normal circumstances. If we use occurrence in a syntactic frame as a criterion (e.g. “We talked about the ____________”) a similar problem occurs – the set of nouns that can occur in one slot is always going to be slightly different from the set of nouns that can occur in another. Using meaning gets you in even more trouble – ever try to come up with an operational definition of thingness? In practice, grammarians use a mixture of all three approaches.

 

Interestingly, some of the early Greeks *did* try to use single-criterion definitions, and came up with exactly three parts of speech: nouns (which have case marking in Greek), verbs (which have tense marking) and particles (which have no marking).

 

(2)                 Although the noun/verb distinction may be an exception, there is usually no empirical basis for deciding which of multiple criteria is more important than the others.

 

I’ll start with nouns again as an example. We consider count nouns, mass nouns, and proper nouns as subcategories of nouns. That approach implicitly contains the assumption that number-marking (pluralizability) and limitations in occurrence with articles (as with proper nouns) are less important as criteria than others. If number-marking were a more important criterion, then count nouns would be their own part of speech. There’s nothing about English itself that tells us how to rank the criteria, though; the language gives us patterns to work with, but does not establish which of those patterns is to be taken more seriously than others. To the Greeks and Romans, case-marking was an obvious way to distinguish words, and in fact count, mass, and proper nouns all get case marking in those languages. It also fit the philosophical purposes to which they put grammar; if you’re making a true or false statement about something, that something will be case-marked in Greek or Latin. Once you’ve decided to use case-marking as your criterion, it makes no practical sense to split off a separate category for proper nouns. You can’t use case this way to define English nouns, though.

 

A number of linguists have claimed that the noun/verb distinction is universal in human languages, but their motivation for doing so comes not from the facts of a particular language, but from (a) patterns of child language development, (b) crosslinguistic similarities in word classes, and (c) the inability of particular formal learning theories to derive language unless a noun/verb distinction is “built in.”  I find the rationale behind this claim problematic, but have to acknowledge the fact that, since a lot of linguists who I know are very, very bright would disagree with me, I may just be stubborn. Even if we say that the noun/verb distinction is “real” and of primary importance, however, that’s about as far as you can push the claim – it doesn’t apply to the other traditional part-of-speech categories. None of that logic mandates the existence of a universal class of adjectives, or adverbs, etc. And in fact, there is enormous cross-linguistic variation in how words other than nouns and verbs pattern – what in English is expressed as an adjective, for example, patterns as a verb in some other languages and as a noun in others.

 

Within English, ignoring nouns and verbs, there are numerous alternative ways that observed differences between the behavior of sets of words Take conjunctions and prepositions, for example; both can be viewed as connecting things together. Many Greek and Latin grammars didn’t distinguish them; there was simply a class of “connector words.” Even that’s a problem, since (as my undergraduate grammar students assure me) there is a sense in which a verb connects a subject and an object. Some grammarians decided that words that connected unlike things (a noun phrase to the rest of the sentence) should be a separate part of speech from those that connected like things (two clauses, or two noun phrases). Why? Why not consider them subclasses, as with proper vs. common nouns? There is no empirical evidence which would force you toward or away from that decision; it’s more an issue of practicality, especially with students expected to memorize terms. Once some grammarians decided that words that connected noun phrases to the rest of the sentence (prepositions) were in a different major category from ones that connected like elements (conjunctions), they were faced with myriad counterexamples. Spontaneous English speech contains numerous examples of what a grammarian would call parallel structure errors, not to mention the frequency of constructions like “I bought more potatoes than Mary,” which on the surface appear to have a conjunction (‘than’) connecting unlike things. Their answer? Decide that some counterexamples don’t count because people *shouldn’t* say them, and decide others don’t count because there are words there that you can’t hear (ellipsis). That last trick renders the entire system immune from any actual data; there’s no way to disprove, for example, the claim that every English sentence contains a silent “by the way” followed by a silent (and extremely rapid) recitation of the Declaration of Independence.

 

Now, I do not mean the previous paragraph to be taken as a denigration of traditional grammarians, nor as an argument for not distinguishing prepositions from conjunctions. The point I am trying to make is that those categories themselves are not “in” the data; the language itself does not force us to consider prepositions as a unitary category separate from conjunctions, or to consider the mass/count noun distinction as being of less importance than the noun/adjective distinction. Grammarians have made those decisions, and the decisions were based on various social and pedagogical motivations that do not directly result from the patterns in the language.

 

 

(3)                 In the absence of an empirical basis for deciding which criteria are more important than others, the choice and ranking of criteria, and hence of the categories they define, is determined by the use to which the theory or pedagogical approach is put.

 

Put more plainly: The data don’t give us any one *particular* set of categories, so we use whichever set we find most convenient to our needs. For pedagogic and social purposes, that may well be eight or ten previously-established categories, like Quirk and Greenbaum’s. We don’t necessarily need to belabor our students with the logic behind those choices, nor with their “constructed” nature (again, those are pragmatic, not theoretical issues).  

 

Bill Spruiell

 

Dept. of English

Central Michigan University