<?xml version="1.0" standalone="yes"?>
<Paper uid="E95-1043">
  <Title>petence and Performance in the Human Sentence</Title>
  <Section position="5" start_page="292" end_page="292" type="metho">
    <SectionTitle>
DEFINITION Left Attachment:
</SectionTitle>
    <Paragraph position="0"> Let D be the current tree description, with root node R. Let S be the subtree projection of the new word, whose left-most attachment site, A is of identical syntactic category as R. The updated tree description is S LJ D, where A is unified with R.</Paragraph>
  </Section>
  <Section position="6" start_page="292" end_page="292" type="metho">
    <SectionTitle>
DEFINITION Right Attachment:
</SectionTitle>
    <Paragraph position="0"> Let D be the current tree description, with the first right attachment site A. Let S be the sub-tree projection of the new word, whose root R is of identical syntactic category as A. The updated tree description is S tA D, where A is unified with R.</Paragraph>
    <Section position="1" start_page="292" end_page="292" type="sub_section">
      <SectionTitle>
3.3 Tree Lowering
</SectionTitle>
      <Paragraph position="0"> It should be clear that, while simple left and right attachment will suffice for attaching arguments without reanalysis, it will not allow us to derive the reanalysis required in example (2). For this, we intuitively require some means of inserting one tree description inside another. Schemat~ ically, what we require is illustrated below, where \[1\] is intended to represent the current tree description built up after John knows the truth has been parsed, and \[2\] is intended to represent the subtree description of the new word hurts.</Paragraph>
      <Paragraph position="1">  We will call this operation &amp;quot;tree-lowering&amp;quot;. Intuitively, the operation finds a node on the current trec description which matches the left attachment site of the projection of the new word, and attaches it, while inserting the root of the new projection in its place. The result is that the node chosen is &amp;quot;lowered&amp;quot; or &amp;quot;subordinated&amp;quot;. In order to maintain structural coherence, the new word attached via tree-lowering must be preceded by all other words previously attached into the description. We can guarantee this by requiring the lowered node to dominate the last word to be attached. We also need to ensure that, to avoid crossing branches, the lowered node does not dominate any unsaturated attachment sites (or &amp;quot;dangling nodes&amp;quot;) We therefore define accessibility for tree-lowering as follows:</Paragraph>
    </Section>
  </Section>
  <Section position="7" start_page="292" end_page="292" type="metho">
    <SectionTitle>
DEFINITION Accessibility:
</SectionTitle>
    <Paragraph position="0"> Let N be a node in the current tree description.</Paragraph>
    <Paragraph position="1"> Let W be the last word to be attached into the tree.</Paragraph>
    <Paragraph position="2"> N is accessible iff N dominates W, and N does not dominate any unsaturated attachment sites.</Paragraph>
  </Section>
  <Section position="8" start_page="292" end_page="293" type="metho">
    <SectionTitle>
DEFINITION Tree-lowering:
</SectionTitle>
    <Paragraph position="0"> Let D be the current tree description. Let S be the subtree projection of the new word. The left attachment site A of Smust match a node N accessible in D. The root node R of S must be licensed by the grammar in the position occupied by N.</Paragraph>
    <Paragraph position="1"> Let L be the set of local relations in which N participates. Let M be the result of substituting all instances of N in L with R. The attachment node A is unified with AT.</Paragraph>
    <Paragraph position="2"> The updated tree-description is D U S U M s It will be noticed that tree-lowering is similar in spirit to the adjunction operation of Tree Adjoining Grammars (Joshi et al, 1975). The difference is that the foot and root nodes of an auxiliary tree in TAG, (corresponding to the &amp;quot;lowered&amp;quot; node and the node that replaces it respectively) must be of the same syntactic category, whereas, as we have seen in this example, in the model proposed here, the two nodes may be of different categories, so long as the resulting structure is licensed by the grammar.</Paragraph>
    <Paragraph position="3"> In the case of example (2), at the point where the truth has been processed, the parser must find an accessible node which matches the category of the left attachment site of hurts (i.e. an NP). The only choice is NP~: (3) \[s \[NP1 John\] \[vv \[v knows\] \[NP~ the truth\]\] NOw, all the local relations in which NP2 participates are found: {dora(VP,NP2), prec(V,NP~) } and NP2 is substituted with the root of the new projection, $2 to derive two new relations: {dom(VP, $2), prec(V, $2)} These relations are found to be licensed, because the verb which V dominates (&amp;quot;knows&amp;quot;) may subcategorise for a clause, so these new relations are added to the set 6. Now, adding the subtree projection of hurts to the set, and unifying its left SNote that Abney's STEAL operation (1987, 1989) is more powerful than tree-lowering, since it may change domination relations, and thus will allow sentences such as (1), though it excludes reduced relative garden paths, such as The horse raced past the barn fell. The original D-theory model (Marcus et al (1983)) is also more powerful, because it allows the right-most daughter of a node to be lowered under a sibling node.</Paragraph>
    <Paragraph position="4"> eNote that the relations defining the original position of NP2, (i.e. dom(l/P,NP2) and prec(V,NP2)) are not subtracted from the set.</Paragraph>
    <Paragraph position="5">  attachment site with NP2 results in the derived structure with NP~ &amp;quot;subordinated&amp;quot; into the lower clause.</Paragraph>
    <Paragraph position="6"> Is \[NP~ John\] \[vp Iv knows\] Is2 \[gP2 the truth\] \[vP~ hurts\]\]\]\] With the tree-lowering operation so defined, the problem of finding which relations to add to the set at a disambiguating point reduces to a search for an accessible node at which to apply this operation. However, this implies that, if more than one such node exists, the parser must be given a preference for making the requisite decision. Consider the following sentence fragment, for example: (5) I know \[NP1 the man who believes \[NP2 the countess\]\]...</Paragraph>
    <Paragraph position="7"> If the input subsequently continues with a verb, then we have a choice of two nodes for lowering, i.e. NP1 and l~P2. Though no experimental work has been done on this type of sentence, there seems to be an intuitive preference for the lower attachment site, NP2. In (6), binding constraints force lowering to be applied at NP2, while in (7), it must be applied at NP1. Of the two, most native English speakers report (6) to be easier.</Paragraph>
    <Paragraph position="8">  (6) I know the man who believes the countess killed herself.</Paragraph>
    <Paragraph position="9"> (7) I know the man who believes the countess  killed himself.</Paragraph>
    <Paragraph position="10"> Note also, that, on standard X-bar assumptions, the attachment of post-modifiers may be derived via lowering at an X I node. In this case, the lowered node and its replacement will be of the same syntactic category (like the root and foot node of a TAG auxiliary tree). Researchers have noted a general preference for low attachment of post-modifiers (this is accounted for by the principle of late closure (Frazier and Rayner, 1982)). This would suggest that a reasonable search strategy for English would be to search the set of accessible node in a bottom-up direction for English. The algorithm is constructed in such a way that lowering is only attempted in cases where simple attachment fails. This means that arguments (which are incorporated via simple attachment) will be attached preferentially to adjuncts (which are incorporated via lowering). This captures the general preference for argument over adjunct attachment, which is accounted for by the principle of Minimal attachment in Frazier and Rayner (1982), and by the principle of simplicity in Gottell (in press).</Paragraph>
  </Section>
  <Section position="9" start_page="293" end_page="294" type="metho">
    <SectionTitle>
4 Processing Japanese
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="293" end_page="293" type="sub_section">
      <SectionTitle>
4.1 Main/subordlnate clause ambiguity
</SectionTitle>
      <Paragraph position="0"> Japanese presents a challenge for any incremental parsing model because, typically, it is not possible to determine where an embedded clause begins.</Paragraph>
      <Paragraph position="1"> Consider the following example: (8) John ga \[Oi ronbun wo kaita\] seitoi wo hometa. John NOM essay ACC wrote student ACC praised &amp;quot;John praised the student who wrote the essay&amp;quot; Up to the first verb kaiia (&amp;quot;wrote&amp;quot;), the string is interpretable as a full clause (without a gap), meaning '~John wrote an essay&amp;quot;, and the incremental parser builds the requisite structure. However, the appearance of the head noun seito (student) means that at least part of the preceding clause must be reinterpreted as a relative clause including a gap (note that there is no overt relative pronoun in Japanese). One way of looking at what is happening here is to see the subject NP John ga as being dissociated from the clause in which it is originally attached, and reattached into the main clause. But looking at it from a different perspective, as Gorrell has noted (in press), one can see the subject NP as remaining in the main clause, and the constituent bracketed in (8), (tonbun wo kaita (&amp;quot;wrote an essay&amp;quot;)) as being lowered into the relative clause. If this is possible, then we would expect examples like (8) to be unconscious garden paths, and this does indeed seem to be reflected in the intuitive data (see Mazuka and Itoh (in press)). However, if we are to allow our parser to handle such examples, we must expand the definition of tree-lowering, since, in order to build a relative clause, we have to assert extra material (including the empty subject and the new S node), which is not justified solely by the lexical requirements of the disambiguating word, the head noun seito. This involves reconstructing all the clausal structure dominating the lowering site (including asserting empty argument positions), with reference to the verb's case frame, and attempting to attach the result as a relative clause to the head noun.</Paragraph>
    </Section>
    <Section position="2" start_page="293" end_page="294" type="sub_section">
      <SectionTitle>
4.2 Minimal Expulsion
</SectionTitle>
      <Paragraph position="0"> Inoue (1991), describes a &amp;quot;minimal expulsion strategy&amp;quot;, which predicts a preference, on reanalysis, towards expelling the minimum amount of material from the clause. In our terms, this means that (assuming a binary right-branching clause structure, with the verb in its right corner) the node selected for lowering must be as high as possible. This means that the bottom-up search which we use for English will wrongly predict a Maximal expulsion strategy. In cases such as (8), assuming the bottom-up search, when a postclausal noun has been reached in the input, the  parser starts its search from the node immediately dominating the last word to be incorporated, (i.e. the verb of what will become the relative clause). This means that, in cases such as (8), the first preference will be to lower the verb (and therefore &amp;quot;expel&amp;quot; both subject and object), whereas the human preference, (to lower the object and verb, and therefore expel only the subject) is the parser's second choice on the bottom-up search strategy.</Paragraph>
      <Paragraph position="1"> Mazuka and Itoh (in press) note that examples where both subject and object must be expelled from the relative clause, as would be the first choice in a bottom-up search, often cause a conscious garden path effect. An example, adapted  from Mazuka and Itoh is the following: (9) Yamasita ga yuuzin wo \[O Oi houmonsita\]  kaisyai de mikaketa.</Paragraph>
      <Paragraph position="2"> Yamasita NOM friend ACC visited company LOC saw &amp;quot;Yamasita saw his friend at the company he visited.&amp;quot; null In order to capture the minimal expulsion strategy in this class of Japanese examples, therefore, search for the lowering node should be conducted top-down. We are currently investigating the consequences of changing the search strategy in this way.</Paragraph>
    </Section>
  </Section>
  <Section position="10" start_page="294" end_page="295" type="metho">
    <SectionTitle>
5 The Problem of Retrospective
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="294" end_page="295" type="sub_section">
      <SectionTitle>
Reanalysis
</SectionTitle>
      <Paragraph position="0"> Having formulated the constraints of Gorrell's model in terms of the accessibility of a node for ~ree-lowering, we can see that the model can be falsified if we can find a case where the relevant disambiguating information comes at a point in processing where the node which is required to be lowered is no longer accessible. Consider the following pair of sentences: (10) I saw the man with the moustache.</Paragraph>
      <Paragraph position="1"> (11) I saw the man with the telescope.</Paragraph>
      <Paragraph position="2"> It is familiar from the psycholinguistic literature that there is a preference for attaching the with phrase as an instrumental argument of the verb (as in (11), on the reading where the telescope is the instrument of seeing). On the assumption that saw selects for a PP instrumental argument, we can derive this preference in the present model via the preference to attach as an argument as opposed to an adjunct. However, since we are constrained by incrementality, we will have to make an attachment decision for the PP as soon as the preposition with is encountered, and it will be attached in the preferred reading as a sister of the verb. This means that, in cases such as (10), where, on the globally acceptable reading, the PP is an adjunct of the NP the man, this attachment will have to be revised, and the PP retrospectively adjoined into the relevant N t node. However, once the preposition with has been attached, the required N' node will no longer be accessible, and a conscious garden path effect will be predicted, which, intuitively, does not occur. Note that there is no garden path effect even if the preposition is separated from the disambiguating head noun by a series of adjectives: (&amp;quot;I saw the man with the neat, quaint, old-fashioned moustache/telescope&amp;quot;).</Paragraph>
      <Paragraph position="3"> The same result obtains if we abstract away from the particular implementational details of treelowering, and return to the abstract level at which Gorrell states his model. Once the PP has been attached as an argument of the verb, it can never be reanalysed as the adjunct of the preceding NP, because the NP will precede the PP before reanalysis, and dominate it after reanalysis, which is against the &amp;quot;exclusivity condition&amp;quot; on trees (i.e. no two nodes may stand in both a dominance and a precedence relation). 7 A similar problem concerns examples such as the following, from Gibson et al (1993): (12) the lamps near the \[that was damaged in the (13) the lamps near the paintings of the house flood\].</Paragraph>
      <Paragraph position="4"> painting of the houses \[that was damaged in the flood\].</Paragraph>
      <Paragraph position="5"> (14) the lamp near the paintings of the houses \[that was damaged in the flood\].</Paragraph>
      <Paragraph position="6"> in the above, Gibson et al have manipulated number agreement to force low (12), middle (13) and high (14) attachment of the bracketed relative clause. The results of their on- and off-line experiments show clearly that the low attachment (corresponding to 12) is easiest, but the middle attachment (corresponding to (13)) is most difficult. This behaviour cannot be captured whether we adopt a bottom-up or a top-down search for tree-lowering. However, even if we can incorporate the required preferences into the parser, the constraint of incrementality will force us to make the decision on encountering that. This means that, assuming we decide initially to attach low, but number agreement on was subsequently forces high attachment, as in (14), then a conscious garden path effect will be predicted, as lowering cannot derive the reanalysis. This is true on the abstract level as well, since there will be nodes in the description which precede the original low position of the relative clause, but are dominated by the subsequent high position of the relative clause. ZNote that in Marcus et al (1983), since precedence relations were not computed for non-terminals, lowering into a predecessor was possible, thus (11) would cause no processing difficulty. However, presumably, their parser would overgenerate on examples such as the horse raced past the barn fell.</Paragraph>
      <Paragraph position="7">  However, intuitively, of the above sentences, it is only (13) which causes the conscious garden path effect.S</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>