# ANTH 252: Glossary of Technical Terms

Many of the readings assume some familiarity with terms and concepts from linguistics, biology, psychology, and other fields. Rather than spending time covering these in class, I will maintain a glossary of such terms, updated in time for each reading. I omit terms for which a sufficient entry can be found in Wikipedia or an online dictionary, foreign-language terms that are translated sufficiently by Google Translate, and terms that are irrelevant to the discussion or used in an obscure or confusing manner.

A   B   C   D   E   F   G   H   I   J   K   L   M   N   O   P   Q   R   S   T   U   V   W   X   Y   Z

ablauting See Indo-European Ablaut.

agent See Agent (grammar)

analysis of variance (ANOVA) Like the t-test and chi-square test, ANOVA allows a researcher to determine whether results are statistically significant or not -- i.e., how likely it is that the results are due to chance, and not to the experimental conditions. ANOVA is usually the method of choice for more complicated experimental designs. For example, a two-way ANOVA allows the researcher to examine simultaneously the effects of two different variables (like age and income) on a third (overall health). Since variables are also called factors, such multi-variable designs are often called factorial, and are described in terms of the "cross" between the two factors, e.g., Treatments X Levels. A nice feature of ANOVA is that it shows the significance both of main effects (like the effect of age on income) and interactions between variables (e.g., does income change the way that age affects health?). Like the chi-square value in the chi-square test, ANOVA uses a value, called the F ratio, to express the strength of the results. Often this value is expressed in terms of the degrees of freedom (df): F(2,61) means two degrees of freedom for the variable in question, and 61 degrees of freedom for subjects. (Because degrees of freedom is the number of distinctions made, this means that there were three different values for the variable and 62 subjects.) As with other kinds of statistical analyses, a low chance probability p, e.g., p < .05, is the standard criterion for determining whether results are meaningful.

anaphoric See Anaphora (linguistics)

anti-mentalist / -mentalism See mentalism

articulation The activity of the speech organs (mainly lips and tongue) in producing speech. Saussure uses it in this modern, standard sense initially, but then confusingly refers to "language articulation" (gegliederte Sprache) later in the same introductory chapter -- despite his crucial, fundamental distinction between language and speech.

argument See Verb argument

aspect See Grammatical_aspect

assimilation See Assimilation (linguistics)

athematic Look here.

atomic Yet another word with a strange history and confusing set of meanings. Since the time of the ancient Greek philosophers, science has sought to discover the smallest, indivisible (lit., a-tomic) units of matter. Shortly after the modern discovery of the atom, physicists realized that atoms were in fact composed of smaller, "sub-atomic" particles, but the name stuck. Logic and linguistics preserve this sense of indivisibility, so that an atomic unit is one that, like a terminal, cannot be "expanded" into further sub-units.

automata-theoretic See automata theory

auxiliaries See Auxiliary verb

case Linguists use this term to describe how nouns and pronouns get marked to reflect their role in the sentence. The English sentences "She saw him" and "He saw her" show that English does this to pronouns. Other languages (German, Russian, Latin, Japanese, Finnish, Turkish) do it to nouns as well.

case filter A constraint, perhaps one of the constraints supposed to make up Universal Grammar, that forces nouns to have an overtly marked case role; i.e., to be the subject of a verb, the object of a verb or preposition, etc. The example in Pinker and Bloom (1990) is less familiar than common expressions like "too big of a ______"; e.g., He's too big of a schmuck to thank you for a favor. The prescribed, "correct" usage would be too big a schmuck, but the the noun phrase a schmuck is "stranded" in this context -- is it a subject, an object, or what? The Case Filter disfavors such stranded noun phrases. The solution is to make a schmuck the object of the (meaningless) preposition of. Ain't language weird?

chi-square Consider the answer to a single question -- e.g., "Do you plan to buy a car next year?" -- obtained from a survey given to two different kinds of people (male / female, under 30 / over 50, Republican / Democrat, etc.) The "Null Hypothesis" is that none of these distinctions makes a difference in car buying, so the average answer obtained from both groups should be the same. The Chi-Square value tells us the extent to which the Null Hypothesis is invalid -- i.e., the extent to which a difference in the average results is meaningful. The df, or degrees of freedom, is simply the number of distinctions made in our survey: for two different groups, we have one degree of freedom. By looking up these two values (Chi-Square and df) in a statistical table, we can determine the probability p of the Null Hypothesis being true. For example, p < .01 means that there is less than a one percent chance that the Null Hypothesis is true, making it extremely unlikely that our results can be attributed to random chance. In the social sciences, a p value of .05 or less is the typical criterion for publishing your results. See also t-test and ANOVA.

C.I.E. tristimulus values Look here for a good explanation.

closed classes See Closed class

Colorless green ideas sleep furiously Look here.

competence (vs. performance) The fundamental distinction in Chomskyan linguistics is between competence and performance. Competence is the abstract "knowledge" that people have about the structure of their language, without necessarily being able to express this knowledge. For example, all English speakers know that the subject comes before the verb in simple declarative sentences ("John loves Mary"). Probably few English speakers could express this fact as such, but they clearly rely on it when trying to understand a sentence that they have never heard before ("Blair loves Bush"). Performance describes the way that people actually speak, including slips of the tongue, misunderstandings, hesitation, stuttering, etc. According to Chomsky, characterizing competence should be the true goal of linguistics.

complement(s) See Complement (linguistics)

compositional(ity) A language is said to be compositional if it uses the same word or symbol for the same concept in different contexts. For example, the word "bird" refers to the same basic thing whether it's the subject of the sentence ("The bird flew out the window."), the object ("I saw a bird in the treetop."), etc. Compositionality is such a universal feature of human languages that it's difficult for us to imagine an alternative way of speaking. It is clear, however, that in acquiring language children start from a non-compositional holophrastic stage, where a single word is used where an adult would use an entire sentence (e.g., "milk" can mean "I want milk", "you are drinking milk", "where is the milk", etc.), and that adults' language contains many expressions whose meaning cannot be understood from the meanings of their component words ("hello", "what's up?", "holy shit!", etc.) More radically, researcher Alison Wray has argued for the holistic (non-compositional) nature of human proto-language, and linguist Jim Hurford has argued that proto-thought had no names for individual things.

constituent Another word for "phrase"; e.g., noun phrase (the dog); verb phrase (bit the mailman), etc.

context-free (phrase structure grammar) A generative grammar containing phrase structure rules like S -> NP VP, which only have one symbol to the left of the arrow ->. I.e., the S symbol has no context, as opposed to a context-sensitive rule like A B -> C D, where there is more than one symbol to the left of the arrow. Although the distinction may seem trivial, context-free grammars generate a more constrained set of languages than context-sensitive grammars, the point of Chomskyan linguistics being to determine the constraints on the possible forms of human languages. Context-free grammars can also be defined as grammars that generate languages that can be processed by a nondeterministic pushdown storage automaton.

control In sentences like John promised to leave, it's clear that John is doing both the promising and (if he's honest) the leaving. Chomskyan syntactic theory describes this situation by saying that the explicit subject (John) controls an implicit, unpronounced pronoun that is the subject of the second verb (leave).

cyclic ordering A principle of generative grammar stipulating that a rule must apply to a structure before applying to any structure that contains it. The details of this rule are not important to us, but the rule is a useful example of the sorts of principles that were supposed to make up Universal Grammar.

derivation Another linguistic term with several meanings (see derivational below). Chomsky (1972) uses it to refer to a phonological derivation; e.g., the fact that the final -s in English noun plurals turns into a /z/ sound after vowels and some consonants (trees, cans), and into an /ez/ (schwa plus /z/) after others (buses, peaches). In other words, the /z/ in dogs is derived from an "underlying" /s/.

derivational (and inflectional morphology) In this sense, derivation refers to processes that change something about a word, like positive/negative or part-of-speech (verb, noun, adjective), typically by changes in morphology. In English and many other languages, this is done with affixes (prefixes and suffixes). Some derivational affixes, like un- (positive->negative adjective) and -ness (adjective->noun) are productive, whereas others, like -ity (adjective->noun) are more like fossilized relics. See also inflection.

descriptively adequate/ descriptive adequacy generative grammar is said to be descriptively adequate for a given language if it generates all the sentences of that language. Because human languages are so vastly complicated, no one has ever come up with a generative grammar for an entire language. Instead, effort has focused on writing grammars for an interesting set of phenomena in a given language. e.g., the relationship between the active ("Oswald killed Kennedy") and passive ("Kennedy was killed by Oswald") forms of declarative sentences in English. A grammar that not only describes such phenomena correctly but also makes interesting, falsifiable predictions about related phenomena, is said to have explanatory adequacy. For example, one could write a grammar for English verbs by listing the past-tense form of every single verb along with its present-tense form, but such a grammar would fail to account for English speakers' competence in forming the past tense of a novel verb ("google") by adding the suffix -ed ("I googled it and got thousands of hits.")

df See chi-square

discrete infinity Because human languages are recursive, sentences can contain an infinite number of discrete elements (words). This term is probably meant to contrast with the infinity of a single continuous object, like a line.

E(s) Used by psychologists and sociologists as an abbreviation for "Experimenter(s)". See also S.

embedded See recursive

F(2,61) See analysis of variance.

felicitous Linguistics jargon for "appropriate to the situation". See Pragmatics.

finite-state automaton An automaton without any memory; essentially, a program that goes from one step (state) to another without remembering where it has been. In an important and influential paper, Chomsky showed that such an automaton is not powerful enough to serve as a model of human language.

focus The "new information" in a sentence, typically a reply to a question. I took out the trash, spoken in reply to What did you do with the trash?, has the focus on took out; in reply to What did you take out?, the focus would be on the trash. Common strategies for indicating focus include intonation (pitch and loudness): I took out the TRASH, and movement of phrases from their default position, as in Spanish: Se fue el hombre.

gap The empty space left in a sentence that has undergone a transformation, like the transformation from a statement to a "Wh-" question:
```(1) John likes Mary
(2) Who does John like _____ ?
```
The idea of there being an invisible "something" at the end of such questions may seem odd, but consider the following examples, where (5) is something you would probably never hear in English (except perhaps as slurred/drunken speech):
```(1) I want to win the election.
(2) I wanna win the election.
(3) I want Hillary to win the election.
(4) Who do you want _____ to win the election?
(5) *Who do you wanna win the election?
```
If the wanna contraction requires that want come right before to, and there really is a gap in (4), we get a nice explanation of the problem with (5).

garden paths See Garden path sentence

generative (grammar) A grammar that can explicitly generate the sentences of a language. The kind of grammar you learn in school is mainly prescriptive: it prescribes (dictates) what is "correct" and "incorrect" English, mainly by appeal to historical tradition or logic (e.g., a double negative is really a positive, although for some reason it doesn't work that way in Spanish, French, and lots of other languages). Generative grammar, by contrast, is descriptive: it aims to characterize how people actually use language, based on the competence that they acquire during childhood. This description often takes the form of rules, like the classic S -> NP VP, which describes the fact that an English declarative sentence consists of a noun phrase and a verb phrase. Because these rules can be recursive (e.g, S -> S and S), generative grammars embody Chomsky's early observation that there is no limit to the number of sentences in a human language.

gestures See phonetic gestures.

goal If the agent is the one performing an action, and the patient is the one on whom the action is performed, then the goal is the person, place or thing toward which the action is directed: John (AGENT) gave a book (PATIENT) to Mary (GOAL).

head The head of a noun phrase (the blue cup) is a noun (cup); the head of a verb phrase (really likes books) is a verb (likes), etc.

homocentric This word can mean either "concentric" (having the same center), as in the circles of a bullseye, or "human-centered", meaning that people tend to project human qualities onto the non-human world. It is this second, human-centric sense that Zubin and Köpcke (1986) appear to be using in accounting for the fact that most names for predatory animals have masculine gender in German.

Humboldt Wilhelm von Humboldt (1767-1835)

illocutionary force For a good discussion, look here. The illocutionary force of an utterance is not always clear. Consider the story about the Russian man who, standing to inherit a small fortune from his uncle, put a classified ad in the newspaper saying "I'll bet anyone 200,000 rubles that my uncle lives another year".

immediate dominance rules Part of the set of constraints that supposedly make up Universal Grammar.

inductive process / induction The process of making generalizations based on specific cases.

inflection Another confusing term. Although most people use it to mean "tone of voice" (which linguists call intonation), linguists usually use inflection to refer to the process of adding prefixes and suffixes (or making other changes) to words, for the purpose of subject-verb agreement ("He leaves at 6PM"), tense distinctions ("Curiosity killed the cat"), gender agreement (Ô femme dangereuse), etc.

intention(al(ity)) Intentionality refers to the beliefs, desires, and other mental states that we attribute to other human and non-human beings. Although it seems reasonable for us to attribute such states to non-human animals, there is debate surrounding whether such animals can make the same kind of attributions about each other. E.g., can a monkey "understand" that another monkey wants to do something? See also Folk Psychology.

interaction See analysis of variance.

innate idea See Kantian preconditions.

Kantian precondition Immanuel Kant, possibly the most influential Western philosopher of all time, was concerned (as were Plato and many others) with the classic nature/nurture issue of how much we are shaped by experience, and how much is "innate" or built-into our human nature. For example, this discussion suggests that Kant saw basic notions like causality (something causing something else to happen) and objecthood (how we tell whether something is a single object or not) as being preconditions to experience, things so fundamental that we could not possibly learn them by an inductive process . The experiments described here suggest that the innateness of objecthood is still an open research question. Chomsky's concern, of course, is with how much of language is innate, and how much can be learned. His answer is, roughly, "all the important stuff is innate, and very little is actually learned".

!Kung San As you can see here, the exclamation point is not a typo; it represents a postalveolar click consonant. As with other common ideas about language, there is some popular confusion about who speaks the languages containing these sounds. To listen to an actual sample of someone speaking one of these languages (not necessarily !Kung), click here.

latency Response time in an experiment. E.g., how long a subject takes to say the word "blue" after being shown a picture of a blue object.

lexicon The dictionary of arbitrary sound/meaning correspondences for a given language, similar to a traditional dictionary. As in a traditional dictionary, linguists put words into various lexical categories: noun, verb, adjective, etc.

linear bounded automaton An automaton that can recognize (parse, understand) the sentences output by a context-sensitive grammar.

local maximum (plural maxima, also called a local optimum) A "good enough" solution to a particular problem, for example, the problem of vision. The idea is that there may be better solutions (the human eye is far from perfect), but modifying a locally optimal solution will likely just make things worse. This is why genetic mutation, though necessary for evolution, is almost always harmful in a particular case. See also adaptive landscape.

logical scope People who complain about the supposed misuse of the word hopefully are complaining about its logical scope. In the "correct" usage, hopefully has scope only over the verb, so He hopefully left means something like He left in a hopeful state of mind. In its more common ("incorrect") usage, hopefully has scope over the entire proposition (sentence), making this example sentence mean I hope that he left. As Pinker and Bloom (1990) note, this kind of scope correlates with a more peripheral position in the sentence; e.g., Hopefully these assholes will stop telling us how to speak correctly.

main effect See analysis of variance.

marked (word order) A separate sense from case marking, this use of "marked" means "unusual" or "atypical". For example, the usual word order for declarative sentences in English is SVO (Subject-Verb-Object), so violating this order in a declarative can be used emphasize the object: Him, I don't like!

modality See Linguistic modality

monotonic function To say that "the meaning of an expression is a monotonic function of the meaning of its parts" (Cann 1993) means that as you keep adding words to the expression, the meaning can only grow more complicated, not less.

motor program A stereotyped sequence of actions performed by an animal without conscious control; e.g., walking, chewing.

morphological See Morphology (linguistics)

nativist Someone who believes in innate ideas

neural network Look here for a good definition. A recurrent neural network is a neural network with feedback: its current output depends not just on its input, but also on its previous output. Because of the sequential and recursive nature of language, this sort of feedback is essential if one wishes to use neural networks to model language.

nondeterministic pushdown storage automaton The class of automaton (computer program) that can recognize (parse, understand) the sentences of a language generated by a context-free phrase structure grammar. Grammars and automata are two sides of the same coin. See also automata theory.

p < .01,   p < .02,   etc. See chi-square

paradigmatic Look here for a good discussion.

patient See Patient (grammar)

perfect See Perfect aspect.

performance (models) See competence.

phase change By analogy with its definition in physics, this term refers to an abrupt quantitative change in a value that suggest a new qualitative situation. An example is Figure 3 of Kirby (2000), where the number of rules in an evolving grammar makes a sudden jump.

philosophical grammar(ians) Chomsky seems to be using this term to refer to philosophers (notably, Descartes and von Humboldt, and possibly some ancient Indian and medieval European thinkers) who tried to describe universal features of human language. It is extremely important for Chomsky to distinguish his Universal Grammar from these earlier ideas about language: according to Chomsky, it is the constraints on possible grammars, and not the grammar itself, that are universal.

phonation For the modern definition, look here. Saussure's unconventional usage of "organs of phonation" to refer to all of the speech organs (as opposed to just the larynx) may be a matter of translation.

phonetic features describe the ways that people produce and perceive the sounds of speech. For example, sounds produced in the front of the mouth (like the /t/ in "top" and the vowel sound in "see") are said to have the feature "frontal"; sounds produced with the lips (like /p/, /b/, /v/, and /f/) are said to be "labial", etc. Describing sounds this way allows phonological rules (see below) to be simple and general.

phonetic gestures Another confusing term. Gesture here refers to the movement of the speech articulators in making various sounds, not to hand gestures.

phonemic analysis a technique developed by structural linguists for classifying the sounds of language. The idea is to find pairs of sounds that make a contrast in some context; e.g., the first sounds in the words "late" and "rate". The fact that these sounds make a meaningful contrast tells us that /l/ and /r/ are two different phonemes in English. Japanese (for example) does not have a phonemic contrast between /l/ and /r/, and English lacks phonemic contrasts found in some other languages. Chomsky's basic critique of structural linguistics was that this sort of classification tells us very little about how these sounds (and other elements, like words) are put together in actual language. For Chomsky, knowledge of language is much more than just knowledge of minimal contrasts. The death blow for phonemic analysis had already come from Chomsky's MIT colleague Morris Halle, whose 1959 work The Sound Pattern of Russian showed that the phonology of Russian cuts across phonemic categories, making phonemic analysis an irrelevant distraction.

phonological rules describe the ways that speech sounds change in various contexts. For example, the plurals of most English nouns end in the letter "s", but this letter sounds like a /z/ after vowels and some consonants ("trees", "cars"), and a vowel (schwa) is inserted before the /z/ after some consonants ("buses", "peaches"). Phonological rules are usually based on phonetic features.

phrase structure (component) in early versions of Chomsky's generative grammar, the phrase structure component contained rewrite rules for generating the basic, or deep structure form of a sentence. E.g.,
```S -> NP VP
VP -> V NP
```
are rewrite rules specifying standard subject/verb/object order for simple declaratives. This deep structure would then undergo various transformations to convert it into a surface structure having more or less the same meaning. The classic transformation was the conversion from the active form of a sentence (subject/verb/object, "John loves Mary") into its passive form ("Mary is loved by John"), although this transformation, and the whole notion of specific transformations in specific languages, fell out of favor as the field developed.

Platonistic Look here for a good discussion. Chomsky's competence/performance distinction seems very much like Plato's distinction between the ideal forms and their earthly manifestation in physical objects.

pragmatic See Pragmatics.

predicate See Predicate (grammar)

preterminal category In a phrase structure rule like N -> John, the N (noun) is a preterminal category; i.e., one that goes directly to a terminal (in this case, John). Because terminals are also known as lexical items, preterminal categories are often called lexical categories.

prisoner's dilemma You and your friend have both been arrested and are being interrogated by the police in separate rooms. If you rat each other out, you both get two years in jail. If you both keep quiet out of loyalty to each other, you both get six months. If one of you keeps quiet and the other rats, the rat goes free and the loyal sucker gets 10 years. So what should you do? Answering this question is the job of game theory, a discipline which turns out to have a lot to do with evolutionary biology, international diplomacy, and other fields where the value of pursuing a given strategy depends on the strategies pursued by others.

proposition(al logic) In philosophy, a proposition is a statement that is either true or false (e.g., "All birds fly"). As Chomsky points out, propositions constitute only one part of language, so efforts to analyze language solely in terms of propositional logic are unlikely to shed much light on the nature of language.

prosodic rules Phonological rules that change the intonation of entire phrases or sentences; for example, the rule that yes/no questions in English must end in rising pitch (Did you take out the trash?). Compare segmental rules.

rank order correlation Correlation refers to the degree to which two measured values agree with each other, whether or not one necessarily causes the other. E.g., smoking is highly correlated with respiratory ailments, and likely causes them, but although height and weight are highly correlated, one does not cause the other. Rank order is a way of arranging the elements in a set (like students in a class), according to some measurable criterion (like grade point average). So, rank order correlation is a way of correlating two values according their rank order, instead of their actual number. An example might be future income bracket as a function of class rank.

real-time deterministic automaton Obviously some sort of automaton, presumably more constrained and biologically plausible than a nondeterministic pushdown storage automaton. Chomsky's point here seems to be that the phrase-structure component of the grammar is less important than the transformations.

recurrent neural network See neural network.

recursive/recursion In linguistics, this term refers to the fact that a sentence or phrase can contain (embed) another sentence or phrase -- much like a box within a box, or a picture of someone holding a picture. Common recursive structures include (1) subordinate clauses; e.g., He said that she left, where she left is itself a sentence; (2) relative clauses; e.g., She's the one who took the book; (3) coordinated clauses; e.g., She took the book and left, which contains two verb phrases connected by the coordinating conjunction and.

reduplicated See Reduplication, especially the Indo-European material in the Function and Meaning section.

rewrite rules Also called phrase structure rules. Kirby (2000) uses a special kind of rewrite rule that specifies not just syntactic (structure) but also semantic (meaning) relations.

Rogerian analyst See Person-centered psychotherapy

S(s) Used by psychologists and sociologists as an abbreviation for "Subject(s)". See also E.

segmental phonology The study of phonological rules that affect individual sounds, like the rule for English plurals -- as contrasted with rules that change the properties (pitch, intonation) of larger units (syllables, phrases); or the set of such rules in a particular language. Compare prosodic rules.

serial interface like Saussure's linearity property, this term expresses the fact that we communicate in a series of discrete units: phonemes, syllables, words, etc.

spectral (changes) The pitch of a sound (like the human voice) can be distinguished from the additional spectral components of the sound, sometimes called overtones or resonances. Spectral information is what allows you to distinguish between, e.g., a note played on a flute and the same note played on a guitar, the difference between the vowels in "hit" and "hat", etc. In human speech, much of the information is contained in the changes , or transitions in the spectrum of the speech signal, as opposed to the absolute spectral values themselves. See also formant.

stop See stop consonant.

subjacency A constraint on grammatical transformations that prohibits, e.g., questions like (4) below:
```(1) Grover stole a car.
(2) Which car did Grover steal?
(3) Ernie saw a monster who stole a car.
(4) *Which car did Ernie see a monster who stole?
```
successor function A function that allows the natural (counting) numbers to be defined recursively, providing the basis for arithmetic; essentially, the function that adds one to a number. If we accept that zero is a number, and that the successor of any number is a number (note the embedding of number in that second part!), we can define all the numbers recursively, without having to list them. Hauser, Chomsky and Fitch (2002) suggest that the ability of humans to think this way may be related to, or even the basis of, linguistic recursion, i.e., the "Faculty of Language -- Narrow" that we do not seem to share with other species.

synthetic Synthetic languages like Russian and Turkish use morphological features like noun suffixes to indicate the case role of nouns. Such language tend to have a freer word order than analytic (a.k.a. isolating) languages, like English and Chinese, that don't mark nouns in this way.

terminal symbols The symbols in a phrase structure grammar that can only appear on the right side of the arrow: essentially, the symbols that actually appear in a sentence (John, loves, and Mary in John loves Mary), as opposed to the abstract symbols (NP, VP). See also preterminal category.

thematized See athematic.

Trubetzkoy Nikolai Trubetzkoy (1890-1938)

(two-tailed) t-test A statistical test used in cases where the variable of interest isn't a categorical (yes/no, tu/vous) distinction, but rather a measurement (IQ, lifespan, etc.) The t-test deals with the same issues (degrees of freedom, probability of results due to random chance) as the Chi-Square test, and is often considered a simple form of analysis of variance.

(two-way) analysis of variance See Analysis of Variance.

[+voice], [-voice] See Voiced.

VOS language Languages are often classified in terms of the way that they order the subject (S), verb (V), and object (O) in the sentence. So a VOS language is one in which the order is verb-object-subject. This order, evolved in one of the simulations reported in Kirby (2000), is actually rare or unattested in actual human languages, which strongly favor putting the subject before the object. Hence, the only widely-attested orders are SVO (e.g., English), SOV (e.g., Japanese), and VSO (e.g., Irish). The emergence of an unattested word order poses no embarrassment in Kirby's model, however, because he makes no attempt to control for word order, and ends up with other orders as well.

Whitney William Dwight Whitney (1827-1894)