Huddleston and Pullum (2002:603) define preposition as follows: "A relatively closed grammatically distinct class of words whose most central members characteristically express spatial roles or serve to mark various syntactic functions and semantic roles."
An adverbial subordinator is a morpheme such as "because", "although", "while" in English that links a subordinate clause to a main clause. In this definition, "clause" is construed broadly to mean finite clause, non-finite clause, and gerund clause. We are distinguishing complementizers from adverbial subordinators. Complementizers are used with complement clauses or subject clauses ("I said that John left."), adverbial subordinators are used with adverbial clauses ("I was sad because John left.")
An attributive adjective is a adjective that modifies the noun internal to the noun phrase by qualifying or restricting its reference. We exclude cases where the adjective is a predicate adjective in a relative clause modifying the noun (e.g., "the man who is big"). Typical adjectives include: "young", "old", "new", and "good".
Cardinal numerals are words ("three") or phrases ("thirty two") signifying the number of things referred to.
A complementizer is a word that marks the beginning or end of an embedded clause. For the purposes of this definition, an embedded clause is the complement of a higher verb. The clause "that John is nice" is an embedded clause in the sentence "I said that John is nice."
Note: There are many uses of the term complementizer in syntactic theory. In this definition, complementizer means complement clause subordinator.
Degree words are words with meanings like 'very', 'more', or 'a little' that modify the adjective to indicate the degree to which the property denoted by the adjective obtains. Degree words are traditionally referred to as adverbs, though in many languages the degree words do not belong to the same word class as adverbs.
Reference:
Dryer, Matthew S. 2008. Order of Degree Word and Adjective. In Haspelmath et al. (2008), chapter 91. Available online at http://wals.info/feature/91. Accessed on 2009-03-09.
A demonstrative is a word modifying the noun that is used to indicate the location of a referent relative to the speaker. For the purpose of this definition, we do not count "reinforcers" in the sense of Bernstein (1997). In French (and other Romance languages), it is often the case that demonstratives appear with reinforcers:
Cette table ci
This.F table CI
"this table"
The presence of the reinforcer ci in this example is ignored for the purposes of the definition for Demonstrative.
Reference:
Bernstein, Judy. 1997. Demonstratives and Reinforcers in Romance and Germanic. Lingua 102, 87-113.
One of the most controversial aspects of word order typology is the set of criteria for characterizing a language as having a particular word order. In this short essay, we clarify exactly how word order is established in our database. The only other on-line database incorporating word order properties is WALS ("World Atlas of Language Structures"). Therefore, we also give a brief comparison of our criteria for determining word order to those used in WALS.
The most basic criterion is that the word order in question must actually occur. For example, we know that English allows the order Adjective Noun, because there are phrases of the form "big ball", "the small tree", etc. Practically speaking, one can get this data from natural speech, a grammar, a book, or eliciting it from a native speaker consultant.
Furthermore, it must be the case that the pattern in question is productive. There are many examples of idiomatic or fixed expressions in English that have a word order that cannot be described as a productive word order:
"Believe me!"
"I am not kidding you."
In general, we assume that the order X Y is productive if and only if either X admits an unlimited number of instances or Y admits an unlimited number of instances (or both X and Y admit an unlimited number of instances). For example, the order in (1) is not productive, because it is only allowed for the three words "believe", "you" and "me". In other words, it is impossible to replace the verb "believe" by an unlimited number of other verbs.
Given this definition of productive, consider the phrase "tall enough", which instantiates the order Adjective Degree in English. This particular order is only allowed for the degree word "enough". The pattern Degree Adjective, on the other hand, is allowed for a wider range of degree words: "too big", "so big", "that big", "very big", etc. Nevertheless, we consider the pattern Adjective Degree productive, because "tall" can be replaced by an unlimited number of other adjectives: "tall enough", "big enough", "smart enough", etc.
Lastly, in our database, we exclude permutations in word order due to emphasis or focus (contrastive or otherwise). For example, in English the OSV word order in (3) is acceptable, but only in a very specific context;
In order to exclude word orders like (3) we require that the word order be acceptable in a "neutral context". A neutral context is one where none of the constituents is emphasized or highlighted more than the others. It can often be solicited by questions such as "What's new?" or "What is going on?"
One important property of our database is that it is possible for a language to be characterized as admitting two seemingly contradictory word orders. In this case, the property X Y will have the value "yes" and the property Y X will also have the value "yes". For example, in our database, there is no such property as "Order of Subject, Object and Verb", rather there are six separate properties, one for each order: Verb Subject Object ("yes" or "no"), Verb Object Subject ("yes" or "no"), Subject Object Verb ("yes" or "no"), etc. So it is possible for a particular language to be characterized as having the value "yes" for both the property Subject Verb Object and the property Verb Subject Object.
Another case where both of the properties X Y and Y X would have the value "yes" is when for some particular instances of X (e.g., "very" in "very big") the order X Y holds, and for other instances of X (e.g., "enough" in "big enough") the pattern Y X holds.
Compare these criteria briefly to those of WALS, whose basic criterion for determining word order is the following: "The expression 'dominant order' is used here, rather than the more common expression 'basic order', to emphasize that priority is given here to the criterion of what is more frequent in language use, as reflected in texts. The reason for assigning priority to this criterion is that for most languages, this is the only criterion for which we have any relevant information."
This definition of 'dominant order' is perfectly clear. Furthermore, the success of WALS as a useful tool for typologists supports the legitimacy of the definition. But there are a number of reasons why we prefer not to adopt this frequency based definition.
First, according to this definition, it is necessary to have a fairly large corpus of varied texts in order to determine which order is more frequent: X Y or Y X. However, for many endangered languages and less studied languages such large corpora simply do not exist, and we do not feel that they should be excluded from appearing in the database because of this limitation. Even for better studied languages, the number of actual texts can be quite limited (e.g., the New Testament).
Second, our database is planned to be language expert oriented. It will almost always be the case that the language expert is a native speaker linguist, or a linguist who is a fluent speaker of the language in question, or a linguist working in collaboration with native speakers. Because of this aspect of the design of our system, the information available to the language experts will go far beyond what is in traditional grammars, which often emphasize morphological and phonological data over syntactic data. For our database, the language expert, reading the property definitions, should be able to set the value of the property definition (either "yes" or "no") on the basis of the definition and their own extensive knowledge of the language (even if they do not have access to large corpora of texts).
Third, the "dominant order" definition naturally chooses one particular order, since by definition if one order is twice as frequent as another, it will be dominant. For example, it is surely the case that Degree Adjective ("very big") is twice as frequent as Adjective Degree ("big enough") in English, and so English would only be "yes" for Degree Adjective and not for Adjective Degree. Our database, on the other hand, naturally invites the formulation of properties where it is possible that two or more "competing" word orders are allowed for a particular language. To exclude the order Adjective Degree as a possible order of English, given our fine grained approach, would be equivalent to discarding valuable information about the word order patterns of English, which may actually be useful in typological searches.
A neutral context is one where none of the constituents is emphasized or highlighted more than the others. For sentences, a neutral context is elicited by
questions such as "What's new?" or "What is going on?" For nouns phrases, a neutral context can be elicited in new topics such as "I want to introduce you to an interesting student" or "I just bought this interesting book."
In an active sentence with an action verb (e.g. "Bill pet the dog") and two noun phrase arguments, the subject is the noun phrase argument referring to the one who performs the action (agent) and the object is the other noun phrase argument.
A predicate adjective is an adjective that is introduced by a copular verb, such as "be" (or other similar copular verbs, like "become", etc.), and that is predicated of a subject noun phrase. Typical adjectives include: "young", "old", "new", and "good".
There are many examples of idiomatic or fixed expressions in English that have a word order that cannot be described as a productive word order:
1. Believe you me! (Verb Subject Object)
"Believe me!"
2. I kid you not. (Verb Object Negation)
"I am not kidding you."
In general, we assume that the order X Y is productive if and only if either X admits an unlimited number of instances or Y admits an unlimited number of instances (or both X and Y admit an unlimited number of instances). For example, the order in (1) is not productive, because it is only allowed for the three words "believe", "you" and "me". In other words, it is impossible to replace the verb "believe" by an unlimited number of other verbs.
Given this definition of productive, consider the phrase "tall enough", which instantiates the order Adjective Degree in English. This particular order is only allowed for the degree word "enough". The pattern Degree Adjective, on the other hand, is allowed for a wider range of degree words: "too big", "so big", "that big", "very big", etc. Nevertheless, we consider the pattern Adjective Degree productive, because "tall" can be replaced by an unlimited number of other adjectives: "tall enough", "big enough", "smart enough", etc.
In an active (non-passive) sentence with an action verb (e.g. "Bill left.") and one noun phrase argument, the subject is the noun phrase argument.
In an active (non-passive) sentence with an action verb (e.g. "Bill will pet the dog") and two noun phrase arguments, the subject is the noun phrase argument referring to the one who performs the action (agent) and the object is the other noun phrase argument.