Compiled by: Owen Rambow
Note: this section needs work!
Each node in the dependency tree can be thought of as an attribute-value matrix, i.e., a bundle of features with values. All of the following features must be set for each node in the tree. This will require checking each node before finishing the analysis.
In addition, there are other features that are speicifc to certain word classes; they are mentioned elsewhere.
Here is a list of features:
- Position (wpos). The linear position of the word in the sentence. This should not be modified or annotated, except for new empty nodes created by the annotator, which should always be given a wpos feature which inserts the new node in the place where it would belong if it were not empty.
- Word. This is the inflected word form associated with the node. It is almost always correctly displayed already. Example: went.
- Part-of-Speech (POS). This is the lexical class, taken from a short list. Example: verb. Specific options:
o V -- verbs, but not auxiliary verbs (=Aux)
o N -- common nouns
o PN -- proper nouns
o Adj -- adjectives
o Adv -- adverbs
o P -- prepositions and subordinating conjunctions
o Conj -- coordinating conjunctions, but not subordinating conjunctions; also includes the comma used in enumerations instead of repeated and
o Det -- determiners; only used for demonstratives and so on, since the and a do not appear in IL0
o Aux -- auxiliary verbs; at IL0, only modal auxiliaries are included, not the auxiliaries for passive, progressive, etc.
o Pun -- punctuation marks, but not the comma used in conjunctions
o Sym -- various symbols (dollar signs and the like)
o Uh -- speech-specifc sounds, even if meaningful (such as /UH HUH/)
o Misc -- everything else, including greetings (Hi, Hello) and interjections (Okay)
o Misc -- everything else, including greetings (Hi, Hello) and interjections (Okay)
Note that pronouns are given the POS that they correspond to (for example, N for he or him, et for his); the fact that they are pronouns is recorded as a feature.
- Lexeme. This is the base form of the inflected word from. Example: go for surface word form went.
- Morphological Features (Morph). A complete specification of the morphological features needed to derive the inflected word form from the base form. The options are grouped by part-of-speech. Possibilities are:
- NOUNS (including proper nouns and determiners)
- sg -- singular
- pl -- plural
- VERBS (including auxiliaries)
- 3sg_PRES -- 3rd person singular present sings
- PRES -- present tense, but NOT 3rd person singular sing; also use for all subjunctives (lest he sing)
- PAST -- past tense sang
- PAPRT -- past participle sung
- PRESPRT -- present participle singing
- INF -- base form when used in inifinitive sing
- ADJECTIVES and ADVERBS
In comparatives and superlatives formed with more and most, label the adjective or adverb as None.
- COMP -- morphological comparative longer
- SUPER -- morphological superlative longest
- MISC
- None -- use this when the word is not inflected, other than infinitive verbs (e.g. adjectives in base form)
- --- -- use this when the word does not ever inflect (prepositions, particles, etc).
- Functional Role Reassignment features. This is a series of binary features which are only used on verbs, nouns, adjectives, or prepositions. They reflect ways in which the usual distribution of roles has been changed. There are three binary features (in all cases, the default value is `-'); only one of them can have value `+'.
o Pass -- passive (only verbs). See discussion on passives.
o There -- there-insertion (only verbs). See discussion on there-insertion.
o Pred -- predicative (only nouns, adjectives, prepositions). Use this to indicate that the noun, adjective, or preposition is used as head of a predicative construction with a dependent form of be which is analyzed as an auxiliary. See discussion of copula constructions.
- Surface role (SRole). This is the role of the node with respect to its mother, as the node appears in the surface string. All nodes have a surface role.
o Subj -- surface subject. An argument that agrees in person and number with the tensed verb is always a surface subject in English; other surface subjects are the empty nodes of non-finite verbs. Surface subjects can also be verbs (sentential subjects, as in That John came bothered Mercedes.), or adjectives (Earlier is better than later) or prepositions (In the morning suits me best).
o Obj -- surface object. This will never have a preposition as its head. This is also the surface role of a complement of a preposition. This is also the role of sentential complements (the verb which heads the complement has the role Obj).
o Obj2 -- surface indirect object.
o Mod -- all adjuncts, including modifiers, appositions, and the like. Also, all function words (determiners, auxiliaries, subordinating conjunctions) will be labeled "M".
- Deep role (DRole). This is the role of the node with respect to its mother, in a deeper representation. We use strictly syntactic criteria. Specifically, DRole reflects the argument patterns of the verb if it were in its active voice form.
o Subj -- deep subject. The surface subject, except for passives verbs, in which case there is no deep subject or it is expressed by the by phrase. The deep subject may, but need not, agree in person and number with the tensed verb; empty surface subjects are like overt surface subjects in that they may or may not be deep subjects.
o Obj -- deep object. This will never have a preposition associated with it in the underlying form (on the surface it might). This is in surface subject position for passives. This is also the deep role of a complement of a preposition.
o Obj2 -- deep indirect object. This is any second object.
o Mod -- all adjuncts, including modifiers, auxiliaries, appositions, and the like.
Verbs are heads of sentences and clauses.
Back to verbs
The head of any complete clausal utterance is the main verb, or, in the case of predicative constructions (see section on copula constructions), the main prdicate. Incomplete utterances (NPs, PPs, Greetings) should have as their head the usual head for that type of phrase.
Auxiliary verbs (do, have, had, auxiliary-be) are deleted. Their meaning is represented as features on the main verb (for example, tense:fut). Modals (can, must, etc) are syntactically very much like auxiliaries, but they are included in IL0 for semantic reasons as dependents of the main verb. In all cases, when the main verb is missing, as in VP ellipsis, an empty verb node should be created and used as the head of the entire clause.
Sequences of auxiliary verbs (had been Ven, are to be Ving, could have been being Ven) should be annotated with the main verb as the head, and all auxiliaries removed (except modal auxiliaries).
When a form of the copula is present in a sentence, the head of the clause will vary depending on the type of copular sentence. Predicative copular constructions will have the predicate as their head. Existential copular constructions will have the copula as their head.
In an infinitive construction, to should be treated as an auxiliary, i.e., it should be removed. This includes instances of want to and have to (in which the to, if it were present, would depend on the embedded verb).
There are several features which describe the verb and its auxiliaries. The verb and its auxiliaries will be referred to as the `verbal complex'. Note that these features describe the entire verbal complex, not just the main verb! They are on the main verb, even if modal auxiliaries are present (other auxiliaries have of course been eliminated).
Back to verbs
Arguments vs. Adjuncts
In distinguishing between arguments and adjuncts, consistency is the most important thing. This distinction will matter most for annotating empty categories. In addition, each argument will be annotated with a feature encoding its grammatical role. All non-arguments will be annotated as adjuncts with role `MOD', including function words.
The only NPs that will be considered arguments for annotation purposes are
A list of argument patterns of common verbs can be consulted for questionable cases.
Back to verbs
The role of each argument (subject, object, indirect object) must be annotated as a feature of its node. See the features page for a more detailed description.
Both deep and surface grammatical relations should be annotated. These will be different when there is a functional role reversal, i.e. a mismatch between surface subject and deep subject. In English, there is only possible case when the surface and deep roles can differ:
See the general discussion in the section on Empty Nodes. ``````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````````.
Back to verbs
Raising verbs will not have a missing category. Instead, annotate them with the surface subject as the direct dependent of the lower verb. In other words, in a raising construction, it is really the lower verb that is imposing the selectional restrictions on the subject of the whole clause.
Verbs (and adjectives) that will be regarded as raising predicates here include seem, appear, need, tend, start, turn out, be supposed to, be going to (gonna), have to, continue, be certain, be likely.
Tests for raising (vs. control structures), include using expletive there (as in (1)), expletive-it (as in (2), weather-it (as in (3)) and a non-thematic subject from a sentential idiom as the subject of the verb in question (as in (4)). Raising structures occur with all of these types of subjects (exception discussed below.) Control structures do not occur with these.
Raising
Control
Finally, don't get confused by these tests. Raising verbs can occur with ordinary NPs as subjects too.
Back to verbs
Control structures should have an empty node included as the subject of their lower verb.
Subject control structures are easy to confuse with raising structures because they appear similar in some contexts.
Raising
Object control verbs include: tell, tempt, force, persuade, appeal to. As with subject control verbs, object control constructions cannot be used with expletives or non-thematic subjects of sentential idioms. Here too an empty node must be included as the dependent of the lower verb. Just like subject control verbs can be confused with raising, object control verbs can be confused with ECM verbs. Using an expletive object is generally a good test to distinguish between the two, as shown here with the control verb decide and the ECM verb believe.
Here are some more examples to motivate the different treatment of the two constructions.
In English, we cannot use that as a deictic pronoun to refer to people without a derogatory effect (since the designated person becomes an object, that being used only for objects): ??that (= George) likes apples or ??I work with that (= Hardy). The pronoun that can, however, be used to refer to something deictically in order to predicate of it that it is a (particular) person: that is my husband or that is my co-worker. Here, that does not refer to a person, but to unformed sense data, which is then identified as being a person. The data above shows exactly the same pattern: that can be used felicitously (without derogatory effect) as a subject of a predication (1, 3), even if that subject has raised to surface subject (1) or surface object (3) position of another verb. This is because in raising (1) and ECM verbs (3), the argument is not an argument of the higher verb. However, that cannot be used to refer to a person (without derogatory effect) in any other argument position -- in (2), that is not only subject of the lower predication, but also of the higher verb (subject control verb), and in (4), it is not only subject of the lower predication, but also object of the higher verb (object control verb). Thus the odd effect comes from the use of control verbs and, as a consequence, the that participating in the higher verb's argument structure.
Back to verbs
In an exceptional case marking (ECM, also known as AcI "Akkusativ cum Infinitiv" or "raising-to-object") construction, the NP that morphologically appears to be a direct object is really the subject of the lower verb. That is, it will have as its head not the ECM verb, but the lower verb.
Common ECM verbs include expect, assume, believe, forbid, know, let, need.
As with raising verbs, the best tests are to use expletive there and non-thematic subject idioms.
ECM constructions may be confused with object control. See Control for a discussion of this matter.
Exceptional case marking constructions with for as in (1-2) below should be analyzed as a subordinate clause with for as a complementizer dependent on the subordinate clause's main verb:
Some ECM verbs (need) subcategorize for either an NP and an infinitive or an NP and a past participle. In the case of the latter, the analysis will be the same as that of the small clause complement analysis. The past participle will be tagged as an adjective.
Back to verbs
Non-finite (gerundive and infinitive) verb phrases (as present participles or infinitives) can appear with or without subjects. Past participles can only appear without subjects.
When they appear without subjects, an empty noun node should be included as a dependent of the verb. If a subject noun phrase is present and part of the VP, as in (1) and (3) above, an empty node should not be included. Instead, that head noun (and its dependents if any) should be a dependent of the non-finite verb.
In general, non-finite clauses will be dependents of main verbs. Exceptions are reduced relative clauses, (such as 5 above), if they modify nouns. In cases that are not clear, the default choice of a head should be the verb.
Small clause complements will be analyzed with the predication as the head of the small clause and dependent on the head verb. The predication may be nominal, prepositional, or adjectival. In the following, the small clauses are bracketed:
The analysis of small clauses is identical to predicative copular constructions, since the overt copula is omitted anyway at IL0.
In the case of a past participle-headed predication,like the following, the participle should be tagged as an verb as well. The missing arguments (the deep subject) needs to be added.
Back to verbs
When the subject of a clause is an expletive it or there, the expletive is removed. In addition, the head node in a there-construction will have an FRR value of "there" or "it".
In sentences like (3) above, the dependency tree will have to incorporate the analysis of predicative copular constructions. In this example the head of the sentence will be the head of the predicate in the garden, i.e., the preposition in, which will have feature FRR:there.
Often, however, there-constructions with the copula will be missing a clear predication, as in the following example:
In fact, this is an existential be. In such a case, include a node for be as the head of the sentence, with the noun phrase three nonstops as its dependents. The NP will be the underlying subject as above. (The lack of agreement in this example is another issue.)
Deciding whether post-NP material is part of the noun phrase or in fact the main predication of the sentence can be tricky. The best test to use is simply to attempt to paraphrase the sentence as a copular sentence, then parse it on that basis. For example, (1) and (3) can be treated as having a main predication. The sentence in (5), however, seems better analyzed as missing a predication and being existential, according to this test.
Back to verbs
As with other full clauses, the head of a wh-question will be its main/lexical verb. The wh-word will be a dependent of the main verb like any other argument.
When the wh-word is part of a long-distance dependency, it will not be a dependent of the highest main verb, but of the embedded main verb heading the clause in which the wh-word originated. The linear order will allow a reconstruction of the wh-word's surface position. In cases of long-distance dependencies, there may be "crossing arcs". This is ok.
If an overt subject is not present, as in (1), include an empty noun; otherwise an imperative will have the same analysis as a declarative sentence. A feature records the fact that it is an imperative.
Back to verbs
A relative clause is the dependent of whatever it modifies, in most cases a noun. The arc is labeled MOD. As with other clauses, its main verb will be its own head. The relativizer will be a dependent of the main verb like any other argument (or adjunct, in cases such as the place where he saw the fish).
Wh-word relativizers and that should be analyzed the same (except of course for part-of-speech). Empty relative pronoun nodes should be inserted if and only if neither a wh-word nor a that-complementizer is present. The arc label should reflect the grammatical function of the relativized argument, independently of the type of complementizer (wh, that, or empty).
In long-distance dependencies, the relativizer will not be a dependent of the highest main verb, but of the embedded main verb heading the clause in which it originated. The linear order will allow a reconstruction of its surface position.
Reduced relative clauses (the flight chosen by you or the airline flying to Wausau) are analyzed like regular relative clauses without overt relative pronoun. They have only an empty subject inserted, but not an empty complementizer, nor an empty auxiliary.
Reduced relative clauses appear similar to non-finite past or present participial clauses and may be difficult to distinguish from these. However, they will always depend on a nominal rather than a verbal head. Although most reduced relative clauses are postnominal, it seems that they can be preposed as in (1) below. When sentence initial, it may be difficult to decide what they depend on. If it is clear that they modify a noun phrase (as in (1) below), choose the noun; otherwise choose the verb as their default head, have in (2) and (3), sang in (4). Note that world knowledge needs to be used when making these decisions.
Two tests to use to decide whether the clause is modifying the verb or a noun:
Back to verbs
The surface vs. deep subject of a passive construction can be indicated through the use of the features. The grammatical subject (usually the patient) will be indicated as the surface subject but underlying object.
The underlying subject (usually the agent), if expressed, will be a surface oblique argument, but the deep subject. If it is not expressed, an empty node should be included (see section on empty nodes).
Passive morphology (i.e. the auxiliary be or got) will be a dependent of the main verb.
Back to verbs
VP-ellipsis should be annotated with an empty verbal head as the root node. Any auxiliaries and the subject will be dependents of this node. No missing arguments should be added. Also see section on empty nodes.
The head of a noun phrase is the head noun. Any determiner is a dependent. Adjectives are separate dependents from determiners. If there are multiple adjectives, the default structure will simply have each adjective as a direct dependent of the noun. This is the case for multiple determiners also.
Adverbial noun modifiers can be dependents of the determiner or the noun in the phrase they modify. For example, approximately, nearly, practically, almost, about, at most, only can depend on cardinals or some quantifiers; at least, only, just, even can depend on nouns (i.e. modify entire noun phrases). These classes have some overlap; the default head choice in cases of ambiguity should be the noun.
Compound noun phrases, when clear, can have multiple noun phrases as dependents. For example, child safety seatwill have seatas the head and childand safetyas its direct dependents. A good test for this is to remove each noun in turn, to see if the phrase still retains part of its original sense. Because a child safety seatis a seat for children and a safety seat, this analysis is the one we want.
In contrast, a phrase like seven-day advance purchaseshould be annotated with purchase as the head, advanceas its dependent and seven-dayas the dependent of advance--> advance purchase vs. *seven-day purchase
In cases where it's not clear whether or which nouns modify each other, the default compound structure will have all modifying nouns as direct dependents on the rightmost noun.
http://www.cis.upenn.edu/~creswell/dependency/compound.gif
Proper nouns should have the value PN for feature POS. They are treated largely like nouns, except that compound proper nouns are not analyzed syntactically as if they were common nouns, but rather given right-branching structures. (The intuition is that they are really fixed phrases.) So in British Airways, British is the head, has POS PN, and carries the other features of this proper noun (in American English, singular number). Airways is a dependent on British (with SRole Adj), and also has POS PN.
InLondon Heathrow airport, London Heathrow is interpreted as a compound proper noun as described above, and airport as a common noun, which has London Heathorw as its dependent (SRole Adj).
http://www.cis.upenn.edu/~creswell/dependency/pnouns.gif
In a noun phrase consisting of only a quantifier, the quantifier should be the head of the NP. Any modifying phrases are directly dependent on it.
In a sentence with the copula as the main verb, a post-copular NP headed by a quantifier and modified by a relative clause (all I needed) should be treated as an equative construction.
Adjectives and adverbs will be coded in much the same way that nouns and verbs are coded. The same procedure is followed.
Adverbs and adjectives point to modifying concepts -- adjectives for nouns, adverbs for verbs. For example, in the phrase, "the blue book" the adjective "blue" modifies the concept book by identifying the color of the book. In "she danced gracefully." the adverb "gracefully" modifies the verb by specifying the manner in which the action was performed.
The degree of the modification can be specified by other modifiers, such as "very" or "light", as in "the light blue book" or "she danced very gracefully." These degree modifiers are also adverbs.
In addition, there are two kinds of degree specification that have special forms in English. You probably know them as the comparative and superlative forms. In the first, the degree of modification is specified by comparing the case in question to one other case: "This book is bluer than that book." In the second, the degree of modification is specified by comparing to all other cases. "She is the prettiest girl in New Mexico."
In English the comparative and superlative degrees are represented in two ways, depending on the word involved. If it is a short word (one or two syllables), there are special forms of the word, created by adding "-er" or "-est" to the word: pretty, prettier, prettiest. For longer words, the comparative and superlative degrees are represented by phrases, prefixing "more" or "the most" to the adjective or adverb: beautiful, more beautiful, the most beautiful". (This method of representation is called "compositional" because the meaning is expressed through the "composition" of several words.) Other languages usually use one or the other.
In order to simplify the lookup procedure in Omega, and to allow for a common interlingual representation of degree, adjectives and adverbs will be shown in their base form (called their "positive degree"). If they are in the text as comparatives or superlatives, that will be indicated as a feature of their node.
There are a few words in English that have irregular forms of comparison. These will also be represented in the parse tree in their base forms. Below is a short list, with the positive form in capitals, followed by the irregular comparative and superlative forms.
GOOD/WELL -- better -- best
BAD/BADLY -- worse -- worst
FAR -- further/farther -- furthest/farthest
LITTLE -- less -- least
MUCH -- more -- most
Quite often participial forms of a verb will show up in syntactic positions also occupied by adjectives. Some adjectives also have the form of participles. The present participle of a verb ends in "-ing," e.g., eating, buying; the past-participle ends in "-ed," e.g., loved, believed.
These participles and participial adjectives can show up
(a) in pre-nominal position:
(b) post-nominal position:
(c) copulative position:
The semantic distinction between participles and adjectives is that participles refer directly to the event denoted by the verb and cast the referent of the modified noun into one of the roles of that event. Adjectives, on the other hand, refer to a state that characterizes the referent of the modified noun.
It is not always easy to tell the difference. Here are some clues / tests to tell the difference:
(1) If there is no corresponding verb, it must be an adjective. E.g., unexpected, talented, down-hearted, diseased.
(2) If you can add the adverb "very" in front of the participial form, then it is probably an adjective. For this test to work, however, the adjective must be scalar or gradable. For example, the adjective blue is scalar and thus intensifiers like very can be added easily, and comparative and superlative forms exist: very blue, bluer, bluest. The adjective, triangular, however, is not gradable. The intensified and comparative forms sound funny: very triangular, the most triangular, etc.
The very smiling man. (bad, and thus a verb) The very frost-bitten man. (good, and thus an adjective) The very heart-breaking results. (good, and thus an adjective) The very quail-hunting vice-president. (bad, but maybe because hunting is not gradable?)
(3) If there are dependents on the participial form (a direct object, or an agent), then it is likely that it is a verb. Thus most postnominal modifiers will be verbs, since their position almost guarantees the presence of additional dependents.
(4) If the word is not listed in an on-line dictionary like Merriam Webster as an adjective, it is likely to be a verb.
(5) When in doubt, make your best guess and discuss the issue with Owen. Participles, which are sometimes coded as adjectives, are generally coded here as verbs. Participles are the -ing and -ed form of the verb, and are not main verbs. For example, in the sentence "The man eating the eggplant is old." the word "eating" is a present participle and modifies or specifies the "man". Similarly, in the sentence, "The man killed yesterday by police was buried today." "killed" is a past participle and again modifies or specifies "man". Since these are coded as verbs, they will also assign semantic roles.
See the manual section on copular constructions for how to handle such sentences as The book is blue.
IN PROGRESS.....
Note that Penn TreeBank did something arbitrary, but consistent, across verbs. What we have decided to do for this project is mainly for the sake of consistency, not out of any strong theoretical bias.
For now, there will be separate nodes for V and Prep. Annotators will
annotate each with the correct concept, and if that concept conflates meaning of
the preposition in the verb, then mark the preposition as "EMPTY".
At IL2,
the preposition will disappear.
This will have the effect of causing "John
gave a book to Mary" and "John gave Mary a book" to look different at IL0 and
IL1, but the same at IL2.
See the manual section on copular constructions for how to handle such sentences as The book is in the tub.
To verbs
Sentences whose main verb is a form of to be fall into several types, mainly existential, equative, and predicative. Existential use of to be asserts existence (there is a man in the moon), equative use of to be equates two entities (the culprit is John), while the predicative use asserts that the post-verbal predicate holds of the deep subject (John is a doctor). These three constructions are treated in two different ways: existential in one way, and equative and predicative in another way.
In the case of existential to be, the head of the sentence is the verb to be, with any prepositional phrase as an adjunct. There can be no adjective or noun after the form of to be. The meaning of the existential construction is that the existence of the subject is asserted. Any PP is modifying the existence assertion.
In the case of equative and predicative to be, the predicate (noun, adjective, or preposition) following the verb to be as the sentence head with one deep syntactic of Subj. The verb to be is treated as an auxiliary, and thus deleted. See verbs and auxiliaries: choosing a head. This analyses makes predicative and equative copula constructions look just like small clauses. Note that the grammatical role of the predicate reflects the role of the predicative construction in the sentence. In Mary is all red, red is the root of the sentence, while in Being a doctor, John often saw blood, doctor depends on see and is a MOD.
The meaning of a predicative construction is the assertion that the predicate holds of the subject. The meaning of the existential construction is that the identity of the two arguments is asserted.
This analysis of predicative constructions also makes English look like other languages in which the verb "to be" is routinely deleted (such as Arabic present tense verb "to be").
If you are having trouble determining whether the use of to be is existential or predicative, use the following rules (also see Expletive subjects and there-insertion):
This rose is very red. (predicative)
This flower is a rose. (prediactive)
This flower is the flower that I was picking when I first saw Pat. (equative)
The flower was in the vase near the jar. (predicative)
The flower I had searched for all my life, and whose existence was doubted by all, was on the fields of that beautiful country. (existential)
Universal consciousness is. (existential)
There is universal consciousness. (existential)
Conjunction has its own part-of-speech (Conj). The conjunction (and, or, but, etc) is palced as a dependent of the first conjunct with role Mod, and the second conjunct is a dependent of the conjunction with role Obj.
If a comma acts as a conjunction, it is treated as such (given part-of-speech Conj and analyzed as in the above paragraph). However, note that in "chicken, ducks, and geese", the second (last) comma does not serve as a conjunction (since there is an explicit "and"), and it is removed at IL0. The first comma does serve as a conjunction.
To verbs
This section discusses cases in which the annotator must add an empty node to the tree. An empty node is a node which does no correspond to a word (or other graphical manifestation such as a punctuation mark) in the input string.
New empty nodes are created using the "new" option under "Node" in the TrEd tree editing tool. The new node should have feature POS to N (most cases) or V (if VP ellipsis). Give the new node a wpos feature so that it ends up in a position that roughly corresponds to its grammatical function (i.e., if it is a subject, to the left of its governing verb, and so on). When the fs files come out of the parser, the nodes have wpos features in increments of 10, so there are enough unused positions to place new nodes where they belong. Never reuse an already used position.
For the lex feature, first identify which node this node is coreferential with. This is usually straightforward. Then copy the co-referential node's word and lexeme values to the empty node, but add brackets around the value, for example "<Dominic>". Alternatively, use lex feature "<pro>" if it is hard to identify the correct coreferential node.
Sometimes, it is not clear what the reference of the empty node is ("arbitrary pro"). Arbitrary empty subjects can usually be found in adjunct clauses. For example, in Eating jicama is good for children's brain growth, the subject of "eating" is not specified. In these cases, we label both the lexeme and the word feature of the new node "<pro>". In case of doubt ("<pro>" or "<child>"), ask yourself: can I tell from syntax alone what this node means? If no, "<pro>". If yes, fill in the lexeme. In the example sentence Eating jicama is good for children's brain growth, the syntax does not determine the identity of the missing subject (compare Buying Microsoft products is good for Bill Gate's wealth growth).
We now list cases of missing nodes.
Missing adjuncts will not be replaced by empty nodes. In some cases, deciding whether the missing constituent is just an adjunct or a seemingly optional argument is quite difficult. Consistency is the important thing in such a situation. See the discussion of arguments and adjuncts.
Do not replace missing semantic arguments which are not syntactic arguments. For example, a non-agentive (ergative) construction (the window opened) and other verbs with missing participants should not have their missing semantic arguments included as an empty NP. Only missing syntactic arguments should be added as empty nodes.
Also, in cases of verbs that can take an optional direct object, do not fill in missing direct objects. For example, in John ate, do not add an empty node as object just because John ate an apple is also possible. We consider the intransitive use a syntactic option, not an elliptical construction (after all, eat does not mean the same as eat something).
There are several standard cases of missing arguments, we list them here but the list is not necessarily exhaustive.
Note that in all of these cases, in addition to labeling the surface grammatical role, you must also label the deep grammatical role, which as usual may or may not be the same as the surface role. The principal reason this may not be the case would be passive voice in the lower verb.
John likes beans, and so does Mary
Henry thought he could jump over that wall, but Jules knew he couldn't
VP ellipsis requires an empty verbal head; the auxiliary is deletd in the usual manner and replaced as needed by features. In addition, add all missing arguments (but not any adjuncts!), as described above. The lexeme and word of the empty head should be filled in from the antecedent between brackets, e.g. "<play>" forMary plays with cats and so does Tony.
John bought two dogs, and Mary bought three
Here, put in an empty noun head (in this case, for dog).
Remove all punctuation, except meaningful punctuation. Examples:
Do remove:
