American Journal of Computational Linguistics 
~i crof i che 45 
SYNTAX 
i N 
AUTOMATI c SPEECH UNDERSTANDLNG 
Boston University 
and 
Bolt Beranek and Newman Inc. 
50 Moulton Street 
Cambridge, Massachusetts 02138 
This research was principally supported by the Advanced Reeearch 
Projeclts Agency of the Department of Defense (ARPA Order No. 
2904) and was monitored by ONR under Contract No. N0001.4-75-c- 
0533. Partial support of the author by NSF grant GS-39834 to 
Harvard University is gratefully acknowledged. 
Copyright Q 1976 
Association for Computational Linguistics 
Table of Contents 
Sect ion 
1 Introduction 
2 Thz BBN Speech Undarvtandinp System 
3 The Grammar 
4 Overview of SPARSER 
PraYninaries 
Be~innin~ tn Parse an Island 
Parsinp Through an Xslaqd 
Ending an Island 
Endinr a Theorv 
Prooessins Multiple Theorids 
Processins Events 
5 &ore Details of the Parsln~ Process 
Depth vs. Breadth 
Scorinp Paths 
~Eorinp Predictions 
6 Examples and Results 
Example 1 
Examole 2 
Example 3 
7 Conclus~ons 
Strengths anb 'ikaknesses of SPARSER 
Prasodics 
Extensions and Further Research 
Conclusion 
Appendix I MINIGRAMMAR 
Ap~andix I1 The Vocabulary and Svntax Classes 
Section 1 
Introduction 
Understanding speech is an extremely complex process which 
requires tho use of many types of kn~wledea, one of which is 
syntax. This report presents a system called SPARSER which is 
designed to provide and use the syntactic knowledge necessary to 
support an artificial speech understanding system. (We will 
assume for the remainder of this paper that unless explicitly 
stated otherwise "speechw means grammatical speech spoken at a 
mo4erate rate with natural inflections and pauses, spontaneously 
produced but similar to the type of speech produced by reading 
text. ) 
We will make the following assumptions about the 
characteristics of speech and a speech processor: 
1. There is not enough infopmation in the spech signal to 
uniquely identify the phonemes or words in a normally spoken 
utterance. 
2. The acoustic processing component of any artificial speech 
understanding system will introduce additional errors and 
ambiguity as it attempts to identify the phonemes-or words in the 
utterance. 
3. As a consequence of 1 and 2, when an utterance is scanned to 
try to identify the words, it is reasonable to suppose that a 
number of (perhaps overlapping) candidates will be found. 
Thls is illustrate in Fipure 1.1 by a structure called a 
word lattice which shows schematically that many Words may 
initially appear to be present. In this representation, the 
numbers along the horizontal scale are se~ment boundary points in 
the utterance which rouphly correspond to points in time This 
word lattice was prodused by the lexical retrieval component of 
the BBN speech understanding system from an utterance which had 
been begmented and labeled by hand under conditions desipned to 
simulate the performance of an automatic sepmrntar and labcler. 
ten I people lard Qlass som~le I s l 
I and I d 
Figure 1 .I 
mognetite 1 
A Word Lattice 
Sentence: Give me all glass samples 'with magnetite. 
lead 
not 
In the system described here, such a word lattice can be 
been . 
did 
represented by a collection of word matches, each of which is 
composed of a word, the boundary points at the left and right 
, ond I 
ends of the portion of the utterance where it appears to match 
y@ 
well, and a score indicating how well it matches the ideal 
I done 
did 
done 
phonemic representation of the word, 
We also make a number of assumptidns about the nature of the 
speech unde,rstanding process and the charactelristic,s of a system 
to carry out that process: 
1. People can understand ua speaker even when the speech .is 
fairly ung~amm~tical, so a syntax-driven system which would 
accept only input marbing rigid syntactic requirements would not 
be adequate for natural, converstional speech. 
2. Since a number oY word candidates are likely to be found 
throu~hout the uttarkne, it may be fruitful to be able to select 
a subset of them on semantic, pragmatic, or prosodic grounds as 
well as syntactic, depending on which cues seem most robust. 
3. Syntax must interact with semantics in order to cut down the 
combinatorial explosion of syntactically correct but meaningless 
subsets of the utterance. Even in the small word lattice of 
Figure 1.1 It can be seen that there are numerous short sequences 
which are syntactically but not semantically valid (tr .g. "Ten 
people are glass samples with magnetiteff, "glas-s samples give 
magnetitew, tllunar samples give magnetitetf, "samples give leadn, 
"people are percentif, etc. ) . 
4. The input to a speech parser will be similar to the word 
lattice described above, thus the parser will have to face not 
only the problem that one or more words in its input mipht be 
incorrect, but that gaps may appear in the input as well. 
5. The parser will have to have the ability to predict words and 
syntactic classes which are consistent with partial hypotheses 
about the content of the sentence in order to help fill gaps in 
the lattice. 
6. Because of the combinatorial explosion of syntactic 
Page 6 
alternatives which occurs when all syntactic possibilities are 
explored for small sections of an utterance, the syntactic 
component must limit the number of such alternatives which are 
actually generated, or at least factor them or treat them 
implicitly rather than explicitly. One way of partially solving 
this problem is to order the alternatives in such a way that only 
the best alternatives ape extended. 
Section 2 
The BBN Speech Understandin3 System 
In the past few years there has been a flurry of activity in 
the field of automatic speech understanding, resulting in a 
number of different systems. For surveys of a number of these 
systems the reader is recommended to wolf 1311, Bates[4], and 
Hyde [lo]. Fgr more specific details on soma of the individual 
systems, sea [I, 2, 7, 8, 16, 19, 20, 21, 22, 28, 29, 33, 351. 
Since SPARSER was implemented as part of a speech understanding 
system called SPEECHLIS which is under development at Bolt 
beranek and Newman Inc., that system is briefly described hare 
and is further documeatad in [3, 4, 5, 6, 15, 23, 24, 26, 33 9 
351. SPEECHLIS has used two task domains; that of the LUNAR 
text question-answering system 1361 which deals with chemical 
analyses of Apollo 11 moon rocks and one dealing with travel 
budget management. 
The overall design of the systefn is illustrated in Figure 
2.4. The acoustics component anaJyzes the acoustic signal to 
extract features and segment the utterance into a iattice of 
alternative possible sequences of phonemes t~chwartz and Makhoul 
[26]), phonological rules augment the output of the acoustic 
component to include sequences of phonemes which could have 
reaul ted in the observed phonemes; the lexical retrieval 
componant retrieve8 words from the lexicon on tho basis of this 
information (Rovner, et.al. [24]); the word matcher determines 
the dr~rea to which the ideal phonetic spelling of a given word 
matches the acoustic analysis at a particular looation [24]. All 
of these components structure their output in such a way bs to 
represent the ambiguity whi;*l is inherent in their analyses. For 
example, they can be used to produce word lattices such as that 
which was shown in Figure 1.1. 
The syntactic component is SPARSER, the system comprising 
the body of this paper (see also Bates [3, 41). Acceptable 
utterances are not restricted to context-free syntax, since the 
grammar which SPARSER uses is a rnoqifiad ATN grammar, capable of 
handling a large, natural subset of English. The remaining 
sections of this thesis detail the structure and operation of 
SPARSER. 
The semantic component uses a semantic network to associate 
semantically related words and to judge the meaningfulness of a 
hypothesized interpretation (See Nash-Wabber [15]). This 
semantic formalism is very-geqoral although a new network must be 
constructed for each new task domain. 
MATCH 
I 
Page 8 
SYNTAX 
(SPARSER) 
J 
LEXICAL 
RETRIEVAL CONTROL 
SEMANTICS 
ACOUSTICS 
u 
Figure 2.4 
Design of BBN SPEECHLIS 
PRAGMATICS 
n 
The pragmatics component is current19 being implemented, but 
is projected to contain information about the past dialogue, a 
model of the user, and other pragmatic data (sat? Bruce [6]). 
control component contains an overall strategy for 
employing the other components in order to obtain an 
interpretation of an utterance (see Rovnsr, et a1 D31) . It 
decides which csmponent is to be called, what input it is to be 
given, and what is to be done with the output. It sets 
thresholds on word match quality. It combines the scores 
produced by the other components in order to rank competing 
hypothesies, and is the primary interface to all other 
components. 
Section 3 
The Grammar 
We have chosen the Augmented Transition Network formalism 
[32] for the grammar which drives SPARSER because it is a 
representation which allows merging of common portions of the 
analysis, it is amenable to both bottom up and top down parsina 
techniques, it fairly clearly separates the use of local 
information from infoflmation which was obtained from a distant 
portion of the utterance and, the author s previous experience 
with a large ATW .grammar for parsing text laid the grouudwgrk for 
the development of a similar grammar for speech. 
We have tried as much as possible to keep the formalism 
which was developed by Woods intact, but some chances have been 
necessary or desirable to make the grammar more amenable to the 
speech parser. We call the formalism a Modifi& Au~mentdd 
Transitiom Network (MATN), and assert that it has the same power 
as the original ATN formalism. The changes, are briefly indicated 
here. For a fuller discussion, see Bates [4]. 
Every arc of an ordinary ATN has a test component, which may 
be any predicate. It is usually a boolean combination of tests 
on the current input word (its features, etc.) and the contents 
of registers which have been sat by actions on previous arcs. In 
the MATN formalism, the test component of each arc is, on all but 
the PUSH arc, a list of two teats. The first is a test on the 
Page 10 
current word and its features, i.e. a local, context-free test. 
The second is a test on the register contents, i.e. a 
context-sensitive teat. Both tests must succeed for the arc to 
be taken. 
The reason for splitting up the testa in this wav is that 
register checking tests cannot be made unless the repisters are 
sat, apd in mally situations in the speech environment there may 
not be enou~h left context to puarantee that the prcper r*egisttbrs 
would be set. Thua it is useful to be a to evaluate the 
context-free test on an arc at a different time in the parsing 
Rrocess from the context-sensitive one. 
On PUSH arcs, there are the types of tests which a used. 
It is useful and efficient to test the next word of input before 
actually doing the PUSH, to sea, for example, if the next word 
can begin a constituent of the ty~e being PUSHed for. This test 
is called a look-ahead test, and takes the place of the normal 
context-free test in the test component of the arc. Them is 
also the usual context-sensitive test on repisters ;rhich were set 
before the PUSH arc was el-]countered. And finally, when the PIISH 
arc returns with a constituent, another context-free tedt may be 
done on the structure of the entire constituent. Therefore, the 
test component of a PUSH arc is a list o.f the three tests iust 
desaribed. 
SENDR s were an efficient mechanism for text parsing because 
they allowed tests to be made on a lower level which involved 
information obtained somewhere (possibly fqr) to the left in the 
input string -- information which would normally be inaccessible 
Page 11 
beoause it would be hidden on the stack during the parsing of 
sub-constituents. 
There are apveral reasons for not allowing this mechanism in 
the speech parser Suppose, in the input that 1-oaks like If.. . 
" the word "personff is not the word 
the person who. travels . . . , 
which was really uttered. If it were allowed to be passed down 
it would become an integral part of the analysis at the lower 
level, and if another word were to be hypothesized in its place, 
the lower level the analysis would have to be redone even if none 
of the words in the relative clause had been ch'anged. Thfs is a 
process which would be extremely wasteful, especially in tho 
speech environment where one wants to be able to take as much 
advantage a3 possible of information which was gained at one 
point and slightly altered at another. In particular, it is 
advantageous to consider as constituents such constructions as 
relative clauses so that they can be placed in a 
well-formed-substring table for use by other probesses. 
Another reason i that some typdv of verifications 
(semantic, prosodic, and pragmatic, at least) can be done most 
conveniently on portions of an utterance which have been assigned 
a syntactic structure, i.e. on constituents. If a portion of an 
utterance is parsed (e.g! "that I gave youff from the complete 
utterance "The book that I gave youf1) but doas not form a 
complete constituent because it is missing a piece of information 
from a higher constituent to the left which would have bean sent 
down had it been available, then these verifications may not be 
made until the missing word or words are identified. Yet it may 
Page 12 
be important to build and verify the constituent in order to 
predict the missing word to tho left. Therefore, it is better to 
allow constituents to be built without information which would 
normally have been passed down When parsing possibly inoorrect 
fragments with little or no left context, it is brat to keep 
constituents a3 small and as independent as possible. 
The conversion process from an ATN eramrnar to a MATN prammar 
with rapard to SENDR s is straiffhtforwardand infalvrs the use of 
a dummy symbol which is used in the construction of the lower 
Level constituent. When the structure is popped, the PUSH arc 
examines it for agreement and may replace the dummy node by the 
appropriate item which would have baeb sent down. The structure 
returned by the PUSH for a relative clause on the fragment "that 
I gave youw might look like Figure 3.1 (where thd structure is 
shown in both the usual tree diagram form and a corresponding 
form more amenable to computer output). 
S REL 
S NP PRO I 
FEATS NU SG 
AUX TNS PAST 
VP V GIVE 
NP **NP** 
PP PREP TO 
NP PRO YOU 
FEATS 
S+NP** YOU NU 
I 
Figure 3.1 SG 
Two Representations of a Parse Tree 
Page 13 
The fourth element of every arc in a MATN is a small integer 
whlch is called the wei~ht of the arc. This weight was 
originally conceived of as a rough measure of either (a) how 
likely the arc is to be taken when the parser is in that state or 
(b) how much information is likely to be gained from taking this 
arc, i.e. whether the parse path will block quickly if the arc 
is wronn. That these two schemes are not equivalent can be seen 
by the following example. In a given state, say just after the 
maln verb of the sentence has bean found, the arc which accepts a 
particle may be much less likely than the arc whdch Jumps to 
another state to look for complements. However if a particle 
which agrees with the verb is found in the input stream at this 
point, then the particle arc is more likely to be correct. Since 
it is not at all clear how to measure or even inbuit how much 
inforrnatiop is likely to be gained from taking an arc, it was 
decided that the weights would reflect relative likelihoods. The 
actual weights which have been used in the speech grammar reflect 
an Tntuitive, though experienced guess as to how likely the arc 
is to be correct if it is taken, assuming the state itself is on 
the corr2ct path. 
Two grammars which will figure predominantly in the 
remainder of this paper have been written in the MATN formalism. 
One is an extensive grammar which can handle ,many questions, 
declaratives, noun phrase utterances, imperatives, active and 
passive forms, relative clauses (reduced and unreducad) , 
complements, simple quantifiers, noun-noun modifiers, 
varb-particle constructions, numbers, and dates (but not 
conjunctions). It began as a modification of the grammar for the 
Page 14 
LUNAR system [361 but has been considerably adapted and expanded. 
This grammar is. called SPEECHGRAMMAR, and is listed ina[4]. 
Exampled are given below which ware produced using this pramnar. 
For some illustrative purposes, SPEECHGRAMMAR ip too hir nnJ 
complex, so we have produced a UINIGHAEIMAR which 1 be ~IF~V to 
show the basic operation of the speech parser. A detailed 
3 
listing is given in Appendix I, but the diagram In Flpu~e ?... 
probably shows the structure mor-c clearlv. The serious rrodrr is 
encouraged to sketch n ccpy of this grammar for r-rfrrencc later 
on. 
CAT AD3 CAT N PUSH PP/ 
PUSH NP/ POP 
PP/PREP PP/ NP 
Since the work reported here was finished, the author has 
written another grammar, called SMALLGRAM which uses the 1IATN 
formalism but which embodies a great deal of semantic and 
pragmatic, information specific to the domain of discourse 
currently baing used by the BBN speech understand in^ project. 
Page 15 
In or for the parser der to move from right to left (to 
predict what could precede that first given word), it must be 
able to determine for any state which arcs can enter it, and for 
any arc which state it comes from. Since the  ramm mar is 
organized for normal parsing in just the oppoeite fashion, i.e. 
for any state one can determine what arcs leave it and for any 
arc (except POP) one can determine which state it terminates on, 
it was necessary to build an index into the granlmar. This index 
consists of a number of tables centaining pre-computed 
informationwhich in effect inverts the grammar. 
Section 4 
Uverview of SPARSER 
The input to SPARSER is assumed to be a set of words 
together with their boundary points (which may or may not be 
related to points in time). A word together with its boundaries 
Is termed a word match. A word match also includes a score which 
indicates how well the ideal phonemic representation of the word 
matched the acoustic analysib of the utterance (but as we shall 
see the parser has little need of this information). Since the 
same word may match at several sets of boundary points or may 
match in deveral ways between the same boundary points, each word 
~t~~ is also given a unique number to help identify it. Thus 
the structure for a basic word match is: 
(number word leftboundary rightboundary lexicalscore) 
e.8. (4 TRAVEL 5 11 94), or (4 TRAVEL 5 11 (94 110)) where the 
score is given as a pair of numbers representing the actual and 
maximum scores, or (4 TRAVEL 5 11) where the score is omitted. 
How is the input to the parser to be constructed? We assume 
that acoustic processing and lexical scanning components can 
operate on a digitized waveform to produce a number of word 
matches such as prev.iously shown in the word lattice of Figure 
1.1. (That this is possible has bean demonstrated by Woods 
[33]). Allowing the parser to operate unrestricted on the entire 
word lattice would probably not be fruitful because of the large 
numbe~ of locally syntactically correct combinations of words, 
but one possibility for input to the parser would be to take a 
set of the best-matching, non-overlapping word matches in the 
lattice, such as those in Figure 4.1. 
A set of non-overlapping word matches is a hypothesis about 
the content of the utterance. In order to avoid creating large 
numbers of such sets which are put together combinatorially with 
no basis except local acoustic match, semantic or pragmatic 
processes can be used to group word matches based on what is 
meaningful or likely to be heard. For example, if a dialogue has 
been about various nickel compounds, the combination "nickel 
analysesw may be more likely than "chemical analysesff even though 
the word match for 'tchemicalff has a higher score than that for 
mnickelfff'. We will not attempt to detail here how this semantic 
grouping could.be done and how the sets could be scored, since it 
has been described elsewhere [15]. 
DO MANY PEOPLE DONE CHEMICAL ANALYSES ROCK 
0 2 6 11 14 22 30 35 38 
GIVE EIGHTY PEOPLE DONE TEN MODAL DETERMINATION ROCK 
0 3 6 11 14 15 18 21 26 35 38 
WERE ANY PEOPLE METAL SEVEN 
0 3 6 11 17 21 27 32 
Fijiure 4.1 
Sample Word Match Sets 
Using more terminology from the BBN speech system, the word 
theorv to denotes a set of word matches such as we have just 
described together with (possibly empty) slots for information 
from each of the possible knowledge sources in the system. From 
the point of view.of SPARSER, usually only the word match portion 
of a theory is of fnterest, hence we shall fall into the habit of 
using the word "theoryv to refer to the word match set it 
contains. When speaking of the syntactic component of a theory, 
however, we are refering to the information slot for syntax whicn 
accompanies each word match set. 
Theories have the fallowing characteristics: 
1) They contain a set of basic, nondverlapping word 
matches. 
2) They tend at first to contain long content words and not 
many shdrt function words. This is because long words are more 
reliably acoustically verified and content words are easier to 
Page 18 
relate semantically and pragmatically. Since small words such as 
tam r~d~v , rrtherl, rr~nall , "have", rr Of 11 11 in rr 
I T , etc. may be 
reprssented by very little acaQstic information, they would tend 
to match at many places in the utterance where they do not really 
occur. Consequently they ase not searched for? by the initial 
word match scan, nor are they proposed in the semantic stages of 
hypothesis formation. 
3) They need not (and generally do not) completely span the 
utterance, but have numerous gaps of va~ving sizes (a.p. for the 
function words). 
4) They tend to contain some sequences of contiguous word 
matches. Such a sequence is called an island. 
That such a set of theories can be created has been 
demonstrated by the BBN SPEECHLIS system. ?he syntactic 
component, SPARSER, is expected to process these theories one at 
a time. In certain circumstances which will be detailed later, 
the input to SPARSER will be a theory together with one or more 
word matches which are to be added in order to create a new 
larger theory which is then to be syntactically analyzed. 
We will assume that there exists a cantrol component which 
presents SPARSER with theories to process and to which SPARSER 
can communicate predictions and results. 
Preliminaries 
Given a theory, what is to be done with it? We begin by 
considering a subset of the question: Given an island of word 
matches, what is to be done with it? The answer is to create one 
Page 19 
or more parse patbn through tho island and to predict what words 
or syntactic classes could surround the island. A parse path is 
tho Sequehce of arcs in the grammar which would be usad by a 
conventional ATN parser to process the words in the island, if 
the island were embedded in a complete sentence. 
For example, consider the way a parser might process an 
island of word match'es such as (1 CHEMICAL 14 22) 
(2 ANALYSES 22 30) using the MINIGRAMMAR of the previous section. 
Beginning in state NP/ of the grammar (omitting for the moment 
the problem of how it is known that NP/ is the rieht place to 
begin) the sequence of s arcs which would bh taken to parse 
"chemical analysesw as a noun phrase is that shown below in 
Figure 4.2. 
JUMP JUMP POP - 
Figure 4.2 
Portion of MINIGRAMMAR needed to parse %hemica1 analysesm 
Let us define a confiatmation to be a representation of the 
parser being in a given state (say NP/QUANT) at a given point in 
the utterance (say 14). We will write configurations as 
STATE:POSITION in text (e.g. NP/QUANT:14) and schematically as a 
box within yhich are written the state and the position. If a 
configuration represents a state which is either the initial 
state of the grammar or a stake which can be PUSHed to (i.e a 
4 
state which can begin the parsing of a constituent), 
it is called 
Page 20 
8rl initiah configuration, and is indicated schematically by a 
filled-id semi-circle attached to the left edge of the box. Note 
that a confi~uration N~/QUANT:I~ is quite distinct from a 
configuration NP/QUANT:22 since they are at different positions 
in the input. In SPARSER, each configuration .is also assigned a 
unique number which is a convenient inte~nal pointer. 
The process of traversin~ an arc of the grammar using a 
particular word is represented by a transition from one 
config.uration to another. A transition ean be made only if the 
arc type is compatible with the current item of input and if the 
context-free test on the arc is satisfied. (The 
context-sensitive tests are evaluated later.) A transition 
carries with it information about the arc which it represents and 
the item of input it uses. The item of input is usually the word 
match which the arc uses, but it is NIL in cases such as JUMF 
arcs which do not use input, and it is a complete constituent fop 
PUSH ams. A unique identifying number and the list of features, 
if any, which is associated with the input word or constituent 
are el30 recorded on the transition in SPARSER, but they are not 
shown schematically. A transition is represented schematically 
by an arrow from one configuration to a~~other with an abbreviated 
form of the arc written above the arrow and the item of input 
under it. 
The syntactic part of any theory which SPARSER processes 
contains, among other things, lists of the transitions and 
configurations which are created or used by the theory. Thus 
wheh we talk about creating a configuration or transition it is 
implicitly understood that SPARSER also adds it to the 
appropriate list, and when we talk of adding an ax is tin^ 
configuration or transition to a theory we mean adding it to the 
npprbpriate list. Therefore, removing a confi~uration or 
transition from a theory means removing it from the syntactic 
part of the thebry, not removing it entirely from SPARSER s data 
base. 
Like confi~urations, transitions are unique, so only one 
transition is ever constructed from point A to point B for arc X 
and input Y. We will frequently speak of creating a transition 
or a configuration, but the reader must bear in mind that if such 
a confi~uration or transition already exists, this fact will be 
recognized and the pre-existing configuration or transition will 
be used. (Timing, measurements indicate that it takes about ,052 
seconds to create a configuration and only .01 seconds to test if 
a particular configuration already exists. For transitions, 
creation takes about .54 seconds and recognition .012 seconds. 
The sequence of configurations and transitions which would 
parse the above example is displayed in Figure 4.3. 
A conpected sequence of transitions and configurations is 
called a at. If the sequence begins with an initial 
configuration and ends with a transition representing a POP arc, 
POP 
NIL 
Figure 4.3 
Path for parsing lfchemical analysesv 
I 
NP/ADJ 
14 
NP/ JUMP NP/ART JUMP 
14 NIL 
CAT N 
CHEMICAL 
CAT N 
ANALYSES' 
NP/ADJ 
22 
NP/QUANT 
14 
1 
NP/N 
30 
JUMP , 
NIL 
Page 22 
it is a complete path, otherwise it is a partiaL path. Paths are 
assumed to be partial unless otherwise specified. 
darrinnina Paraa a Island 
SPARSER processes an island of words by beginning with the 
leftmost word and determining its possible parts of speech. Then 
the arcs of the grammar which can process the word arc fpund (by 
looking in the previoRdly constructed Rrammar index). For each 
arc, two confi~urations are constructed one for the state at the 
tail of the arc and one for the state at the head, using th$ left 
and right boundary positions of the word match, respectively, and 
a transition for that arc using the current word match is also 
built. Schematically, we have for our example a situation which 
looks like that of Figure 4.4 (such a display of all or some of 
the transitions and co~fi.qul*at ions which the parser has 
constructed is called a map). Notice that a configuration may 
have any number of transitions entering or leaving it. 
Figure 4.4 
Initial map for parsing llchemical analysestr 
Page 23 
The idea of this process is to begin t6 set up paths which 
my be used to parse the island. However it is not necessarily 
the case that the only donfigurations which could start paths 
throu~h the island are those which have just been obtained, since 
it may be possible to oreatr transitions which enter them via 
JUMP arcs or TST ahca. For each state, the sequence of arcs 
which can reach it without using the previous word of input have 
bean be pre-calculated by the grammar indexing package so the 
appropriate configurations and transitions may be constructed. 
These transitions ape cplled lead-in transitions. Thus the map 
becomes that in Figure 4.5 
Note that any of the configurations (except for NP/ADJ:22 
22 
J J* 
and NP/N:22) could actually be the correct leftmost configuration 
for this island, depending upon what the (currently unknown) left 
NP/ JUMP NPIART JUMP NWQUANT 
14 NIL* 14 
14 
b 
context of the island is. 
'NP/N 
Figure 4.5 22 b 
Lead-in transitions for parsing llchemical analysesn 
JUMP 
NIL 
NP/ADJ 
14 
By looking in the grammar index, SPARSER can determine, for 
t~c 
each configuration which could start the island, just what sort 
of left context could be appropriate. For example, th'e CAT ADJ 
arc in MINIGRAMMAR which enters state NP/QUANT implies that an 
adjective could precede the island and, if it did, the tra~sition 
which would proceas it would terminate on configuration 
NP/ADJ:14, 
Because the initial configuration NP/:14 could start the 
isiand, anything which could precede n noun phrase could occur to 
the left; again the grammar index provides the information that 
the CAT PREP arc could lead to a configuration which could accept 
a noun phrase (via the PUSH NP/ arc), so a preposition could also 
prefix the island. If the index functions indicate that a 
constituent could be picked up by a PUSH arc which could 
terminate on the configuration under consideration, an indication 
is made in tho WFST so that any time a constituent of the desired 
type is built which ends at the proper location, it may be tried 
here. 
Because of the highly recursive nature of ATN grammars, it 
is vary likely that as we chain back through the possible 
sequences of PUSHas which could lead to tho beginning of tho 
current constituent (or the seauence of POPS which could be 
initiated by the completion of the current constituent) a large 
number of predictions will be made. Rather than make all these 
predictions automatically, beford we are even sure that there is 
in fact a constituent at the current level, the possible 
configurations which could make predictions on other levels are 
saved to be activated later if the predictions from the current 
set of active configurations are not sufficient. 
Page 25 
The predictions which are made (not saved) are not acted 
upon at this time, but ard kept internally by SPARSER until all 
the islands of the theory have been prooessed. We shall see 
bdew what then becomes of the predictions. 
Island 
Once proceasing has proceeded this far, we can go back and 
consider the set of configurations which represent states the 
parser could be in just after processing the first word of the 
island. In our example, these are configwatiqns WP/ADJ:22 and 
NP/N:22. Configurations such as those whlch are waiting to be 
extended to the right are called active configurations. SPARSER 
selects a subset of the set of active co'nfi~urations (how this 
subset is selected will be discussed in the next section) and for 
each configuration tries to extend it by tryin8 to parse the rest 
of the island beginning in that confipuration. When the parser 
is considerin~ a configuration at some position, the input 
pointer is set to the word match of the island, if any, which 
begins ah the same position in the input. 
The grammar associates with the state of the configuration a 
list of arcs which may be tested (using the arc type, the context 
free test on the arc, and the Current input) to determine whether 
a transition can be made to extend the path. We will consider 
each type of arc in turn, since the effects of taking various 
types of arca are different, and explain for each case what 
happens if the arc is taken. Whether just one transition, or 
several, or all possible transitions are made from an active 
Page 26 
conf,iguration is a matter to be discussed in Section Five. 
Soma JUMP arcs do not look at the current item, so they may 
be taken whether the input pointer is set to a word match or to 
NIL. The transition which results from taking an arc of this 
type has a null item associated with it, even if there is a word 
match in the theory at this point. The positions of the 
confipurations at each end of the transition are the same; this 
corresponds to the fact that an ATN parser would not move the 
input pointer as a ,consequence of taking this arc. 
Rarely, a JUMP arc may test the current iten in some way, 
for example, to make a feature check. If there is no word match 
for input, an arc of this type cannot be taken. If there is a 
word match, it is noted on the trahsition wh ch is created, but 
the configurations at each end of the transition have the same 
posit ion. (It is than the case that thainext input-using or 
input-consuming transition on the path including this transition 
must use the same word match.) 
These are TST, CAT, and WRD arcs which end in a 
(TO nextstate) action. The operation is exactly tho same as that 
above except that the configuration on which the transition 
terminates has the position of the right boundary of the current 
word match. 
Taking a POP arc results in the creation of a transition 
which has a null final configuration and a null item, because POP 
arcs are not permited to look at input. 
Page 27 
When a PUSH arc is encountered, a monitor is placed in the 
Well-Formed Substring Table (WFST) at the current Dosition to 
await the occurrence of a constituent of the required type. If 
one or nore such constituents are already in the table, then for 
each one there are three possibilities: it may be composed of 
word matches which are in the current theory, it may be composed 
of word matches some of which are not in the current theory but 
which could be added without violating the non-overlapping 
constraint, or it may be composed of word matches some of which 
are incompatibld with the current theory. 
In the first caae a transition is set up using the 
constibuent as the current word. The transition terminates on a 
confleuration whose state is determined from the termination of 
--. 
the PUSH arc and whose position is that of the right boundary of 
the rightmost word match in the constituent. 
In the second case, a notice is created and sent to the 
control component. A notice is a request that SPARSER be called 
to enlarge a theory by addinp some new information, in this case, 
some additional word matches which form a constituent that the 
theory can use. SPARSER does not try to determine when (or even 
whether) the theory should be so enlarged. That is an issue for 
the main cqntroller to decide (see Rovner, et.al. C231). We 
will discuss below how SPARSER enlarges a theory if called upon 
to do so. 
In the final case, if there are no usable cohstituents in 
the WFST, a new configuration is set up to start looking for one 
and is added to the list of active configurations. Its state is 
Page 28 
the state specified by the PUSH arc and its position is the same 
as the current configuration. 
There is a considerable amount of processinc that can happen 
any time one of the transitions lust discussed is rnndr. Whenever% 
an initial configuration is constructed, this fact is r*t)c@rded in 
the configuration. Whenever a transition is nade from such a 
confipuration, the information that there is a path I sene 
initial conf igur-at ion is recorded on the subsoquenf 
configuration. Similarly, whenever a POP tr-ansition is made, the 
c~n~figuration it emanates from and all previotls configuratims on 
any path which can terminate with the POF transition are marked 
to indicate that they can reach a POP transition. Whenever a 
transition is made which completes a path from an initial 
configuration to a POP transition, the path is executed, one 
transition at a time, and the rqister setting actions and 
context sensitive tests are executed. If a test fails or an arc 
aborts, the transitions and configurations f the path are 
removed from the list of configurati~ns and transitions which are 
in the syntactic part of the currant thaory (unless they are used 
by another path in the theory) but not removed from the nap. If 
the execution is successful, a deep structure tree is produced. 
That structure together with its features is given a score, whioh 
may include evaluations by other cornpodants such as semantics and 
prosodies, and is entered in the WFST. 
It is quite important that sources of knowledge other than 
syntax be called upon to verify and to rank syntactic 
constituents. This is because there are likely to be many 
Page 29 
ombinations of plausible words from the word lattice which form 
ayntact%rally reasonable constituents but which may be ruled out 
om @%her grounds. To allow immediate use of this information 
which syntax cannot provide alone, SPARSER has an interface to 
he semantic component so that constituents can be vrrif ied 
dinctlg without going through the control component. It will be 
%trivial modification to insert verification aalls to pragmatics 
aad prosodies when they become available. In the meantime, even 
semantic knowledge can be turned off; if the parser gets no 
IngosnratIon from the call to semantics, it proceeds without it. 
Plaoement of a constituent in the WFST causes a number of 
Lbiags to happen. First, any monitors which have been set by the 
mmnt theory at that position aFe activated. That is, for each 
ebnfigaration which was waiting for this constituent, a PUSH 
tmaaitioa 3s made which uses the constituent as its input item. 
If ao rmonltors have been set which can use this constituent, it 
as treated exactly as if it were the first word of an island: 
a11 the PUSH arcs which can use it are found in the grammar index 
md appropriate configurations and transitions (including lead-in 
tnositions, if appropriate) are set up. Next, if there are any 
monitors for other theories which can use the constituent, 
patfees are created and output to Control as was described above 
is the section on PUSH transitiohs. 
Figure 4.6 shows SPARSER s map after our example island has 
ken completely processed. The parsing results-in the creation 
oi a CAT II transitio~ to configuration NP/N:30 using the word 
*~~alyses~ The PUSH PP/ arc at state NP/N would oause 
configuration PP/:30 to be created. Similarly, PP/:22 would be 
created when tho configuration NP/N:22 is picked up to be 
extended. The POP arc transitions from each of the 
configurations for state NP/N result in the formation of complete 
paths, resulting in the creation of two noun phrases ("chemical 
analysesw and ttchamicalft). Since there were no monitors for 
them, they result in the creation of configuration PP/PREP:14 and 
$8 subsequent paths. 
Figure 4.6 
Map after processing island 
7 w 
NPIADJ CATN NP/N POP 
22 ANALYSES. 30 
NP/ JUMP 
14 NIL 
N P/AD J 
30 
- 
NP/ART 
14 
JUMP 
MIL 
NPIQUANT 
14 
JUMP NP/ADJ 
NIL 14 
Page 31 
Endinq an Island 
It may be the case that no path can be found from one end of 
an island to the other, (This would occur when all active 
configurations block.) In this case, there is no possible way 
that the island could form part of a grammatical string, so 
SPARSER can inform the control component that the theory is 
wrong, 
When an active configuration is picked up to be extended and 
there is no word match at that point, the end of the island has 
been reached. That does not mean that no more transitions can be 
made, since arcs which do not test the input word can be taken as 
usual. Arcs which do use input cannot be taken, but they can be 
used to predict what sort of input would be acceptable at that 
position. For example, a GAT V arc which has a test requiring 
the verb to be untansed would allow SPARSER to predict an 
untensed verb beginning at the position of the ourrent 
configuration. CAT and WRD arcs cause the prediction of 
syntactic categories and specific words, respectively, modified 
by the Context-free test on the arc. TST arcs provide only the 
test which must be satisfied, and PUSH arcs cause a monitor to be 
set in the WFST as well as a TST monitor for the the look-ahead 
test (if any) on the arc. 
Bndinq 3 Theorv 
When k11 the islands of a theory have been processed 
in the 
manner just described, it is time to deal with the gaps between 
the islands. As we have seen, arcs in the grammar which can 
Page 32 
enter configurations at the left end of an island or which can 
leave configurations at the right and of an island can be used to 
make predictions about warda that may be adjacent to the island. 
The prediction is a list of the arc, the confi~uration it would 
connect to, and an indication of whether the transition caused by 
the arc will enter the configuration from the left or leave it to 
the right. 
If a gap between two island8 is small enou~h that it may 
contain just one word, than it is likely that tho arc which would 
process that word may have caused a prediction from both tho left 
and right sides of the gap. If this is the case, and if the 
predictions intersect in a single possibility, it is highly 
probable that the word (or syntactic class) so predicted is 
correct. If the predictions do not intersdct, parsing is 
continued from the active cbnfigurations which were not triad 
earlier because of their scores and from the configurations which 
could begin constituents at the right and of an island. This 
continued parsing is an attempt to find a path which results in a 
common prediction acg#ss the gap. If that too fails, then the 
configurations which were saved because they could lead up a 
chain of PUSHes or POPS to new configurations are triad. If no 
possibilities are left to try and there is still no prediction to 
fill the gap, this information is noted, but it does not 
definitely mean that the islands are incompatible, since in some 
cases the gap could actually be filled by two words instead of 
one. 
SPARSER has two kind of predictions - those bhioh seem 
highly likely and those which seemaless likely. A highly likely 
prediction, such as one which is made from both side3 of a small 
gap, is output in the form of a prooosal, which is a request to 
the rest of the system to find a word meeting the requirements of 
the proposal. A proposal contains: 
1) the item being proposed, which is either a particular word 
or list of words (from a WAD arc), or a syntactic olass (from a 
CAT arc), @r NIL, meaning any word (from a TSI a~c) 
2) tho loft and/or right boundary pointb) of the item 
3) a test which the item must satisfy (the context free test 
from bhe arc) 
4) the context of the proposal, i.e. the word match(es) on 
the left apd/or right side of the item baing proposed. (This is 
to help the lexical retrieval component take into acceunt 
phonological phenomena which may occur across word boundaries.) 
All predictions whether or not they are confident enough to 
become proppsals are output oas monitoys. A monitor is a 
notification to the control component that if a word meeting the 
requirements of the monitor is somehow found (perhaps by the 
action of a proposal) , it may be added to the theory. Thus a 
monitor acts like a demon which sits at a particular point in the 
word lattice and watches for the appearance of a word match which 
it can use. A monitor contains: 
1) the item being monitored for (generally a syntactic 
categcry, but may be a word or a test) 
2) the left or right boundary position of the item baing 
monitored far 
3) a. test which the item must satisfy (same as for proposals) 
4) the thaory which generated the sonitor 
5) the arc in the grammar which will process the item if 
found 
6) the configuration from which the prediction was made 
7) a score, indicating roughly how important the monitor is, 
i.e. how much information is likely to be gained by processing 
an event for that monitor. 
(Notice that monitors which are sent to the control component are 
very much like monitors which are set in the WFST by the 
occurrence of PUSH arcs.) 
Once the proposals have been made and the monitors have been 
set., SPARSER bundles up the information it knows about the 
current theory, such as the configurations and transitions in the 
theory, any configurations which a still candidates for 
expansion, the constituents in the theory, the notices, 
proposals, and monitors which have been created, etc. and 
associates the bundle with the thaory number. This insures that 
SPARSER will be able to pickup where it left off if it is later 
given the thmry to process further. 
Processinq Mult i~le Theories 
Thus far we have seen only the opSrations which SPARSER 
performs on a single theory, but we made the assumption that 
SPARSER would be given a number of theories to process in 
sequence. Let us now examine what will happen when the second 
(or nth) tkeo~y is processed. 
Page 35 
SPARSER will no longer have a blank map and WFST; instead 
it will have all the configurations, transitions, and 
constituents which have been constructed by all previous 
theories. For concreteness, let us imagine that the theory (1 
CHEMICAL 14 22) (2 ANALYSES 22 30) has been processed, result in^ 
in the map shown in Figure 4.6. Now we are going to process a 
theory containin8 the island (4 NICKEL 16 22) (2 ANALYSES 22 30), 
which results in the map of Figure 4.7 where the configurations 
and transitions added by this theory are shown in dotted lines. 
The process bagina as usual with the creation of 
conf,iguration #P/ADJ:16 and three possible lead-in tranbitions. 
Tho transitions for the two CAT N arcs, however terminate on 
configurations which already existed in the map, so the complete 
paths from configuration NP/:16 to configurations NP/N:30 and 
NP/N:22 will be discovered and processed, resulting in the 
construction of two new noun phrases. Those new constituents 
would then result in the creation of configuration PP/PREP:16 
and two new transitions. Thus we have constructed only five new 
configurations and seven new transitions and have been able to 
take advantage of six old configurations and six old transitions. 
In this fash-ion any information which has once been 
discovered about a possible parse path is made available to any 
other path which can use it. - No reparsinq - is ever done 
SPARSER merely realizes the existewe of relevant configurations 
and transitions and incorporates them into the current theory. 
NP/ADJ CATN NP/Npgp 
22 ANALYSES? 30 
NPI ~UMP* NPART JUMP. NPIQUANT JUMP NPIADJ 
#' 
NIL ' 
0 
14 14 NIL 14 
> 
NP/ADJ 
30 
.. 
- 
I---- 1 ' 
NP/N ' POP- 
I NWADJ t 
I CAJJ?*--- 
16 - 'N\C(ICKEL 
+!J 
L--,,J 
Figure 4.7 
Map after processing islaad for "nickel analysesif 
If the new word (or wokds) in a theory are at the and (or in 
the middle) of an i$land, when SPARSER begins to parse the island 
it will discover the existing configurations and transitions from 
the previous theory. Whenever a transition which can be used in 
the current theory is discovered in the map, it and its 
'terminating configuration are added to the syntactic part of the 
current theory. This is callad tracinq the transition. In 
addition, all paths beginning with that transition which do not 
require the next word of input are also included in the syntactic 
part of the theory. This is accomplished by tracing from the 
terminating configuration all transitions which use either tho 
same word of input as the previous transition or no input word at 
all. (A similar process 19 used to trace backwards, i.e. right 
to left, when neessary.) Uhen a configuration is reached which 
has no traceable transitions emanating from it, the tracing 
process, stops. Since both transitions and configurations are 
stored in such a way as to facilitate tracin~ (for example, each 
transition has a code attached to indicate whether or not it 
consumes or tests input), this process is considerably faster 
than creating that portion of the map in the first place. (To 
illustrate this, a theory was processed twice, onca with an empty 
map and onca start in^ with the map previously created; the time 
required for processing the theory fall from 47.5 seconds td 
16.5.) 
Configurations which can end trvaced paths are put on the 
active conCigurations list. If, when one oP them is picked up 
for extension, it is discovered that the next word of input was 
used on a transition already in the map, the tracin~ process is 
repeated. If the next word of in~ut is new (or at least has not 
caused any transitions from thi? con fipurat ion beinp considered ) 
then para in^ continues in' the normal manner. 
Processih~ Events 
As-was mentioned earlier, SPARSER can be called upon to add 
some new word matches to a theory it has previously processed. 
In this case, SPARSER is said to process an event. An avant may 
be thought of rather abstractly as the discovery of a piece of 
information that has been syntactically proposed, monitored for, 
Page 38 
or noticed. Concretely, an event is a piece of data consistina 
of: 
1) the old theory that proposed or set a monitor far the 
event 
2) something to be added to the theory (a new ward natch or 
constituent) 
4) the arc in the Rramnar which will process the new 
information 
4) the corifipuration ih the old tbrory which will be at one 
end of the transition created by the above arc 
When SPARSER is ~iven an event, it retrieves from its tables 
the bundle of configurations, transit ion, etc. in the old 
theory. Then using the arc and the new word or constituent in 
the event, it creates the appropriate transitioncs). Then 
processing continues as usual, that is, any complete paths are 
noticed and processed, and any new active confi~urations are 
exbended, if possible. 
New predictions may be made as a result of this increased 
information. (A record is kept of previous predictions so none 
are remade unless with a more liberal score.) Finally SPARSER 
returns the nGw, larger theory. This new thsory may be processed 
as part of another event at some later time, thus gradually 
reducing the number and size of the caps in the theory. 
If an event results in filling the final gap in a theory, 
and if the resultinp complete sequence of words can be parsed, 
SPARSER notifies the control component of this fact, since the 
entire utterance may have been discovered. Of course, this may 
not be the correct solution -- it is up to the control component 
to look atm the acoustic ~oodness, semantic meaningfulness, 
pragmatic likelihood, etc. of the result as well as the 
syntactic structure before daclarlng the utterance to have been 
understood. If for reasons other tham syntactic, the utterance 
appears to be bad, the control component of the system could ~o 
on to try to find anothar, more suitable, possibility. 
Section 5 
More Details of the Parsinfi Process 
5.1 DEPTH vs BREADTH 
The parsing strategy just outlined works bottom up when 
beg inn in^ to parse an island and when a constituent is created 
which was not monitored for by the current theory. It works top 
dawn after an island has been started and to make syntactic 
predictions at the ends of islands. Both top down and bottom up 
techniques can be either depth or breadth first. Depth first 
processing takes at every step the ffl-st piece of information 
available and pursues its consequences. Breadth first processing 
considers at every step every possible next step of every 
alternative and pursues all paths in parallel. Breadth first 
processing generally takes much more space than the depth first 
many paths would have to be remembered at once 
Rags 40 
instead of having just one stack which could be popped and reused 
when necessary. 
The breadth first process mi~ht save some computation steps 
and might produce several ambiguous parsin~s simultaneously while 
tba depth first process would find one before the others (the 
latter is a'small difference, since both processes would have to 
be run to exhaustion to insure that all possible parsings had 
been found). In parsing speech, some mixture of breadth first 
and depth first proc~ssing can be extremely useful. 
To illustrate an advanta~e of breadth first processing in 
the speech environment, consider what might happen if, durinp the 
processin8 of an island the parser picks up a confirtiration to 
extend which has several possible arcs emanating from it. If one 
arc is chosen and all the others are held as alternativss (imem 
depth first), but the chosen arc is wronp, all subsequent paths 
beg inn in^ with that arc would have to block before the 
alternatives would be tried. However, if the end of the island 
were reached before. the success or failure of the first choice 
were confirmsd, the only way that backup would ever take place 
would be to have one or more events add words to the thsory so 
that the path could be extended until it failed. Since the pap 
wou.ld be likely to be filled by (incorrect) words predicted by 
the erroneous path, or by no words at all if the (incorrect) 
predictions were not satisfied, it is not at all clear how the 
process would aver know to back up. 
This problem cannot be eliminated completely without 
pursuing all alternatives to thair Fulleat extent (a 
combinatorially unacceptable solution) but it can be modified to 
a praat extant by a judioious combination of depth and breadth 
first processing to find the best path, not just the first one, 
through the island. This "bast pathw is not ~uaranteed to be the 
correct one, so it is possible to continue process in^ by 
extending paths with were suspended earlier. 
SPARSER handles the problem by assigning a score &o every 
configuration which reflects the likelihood of the path which 
terminates on that confi~uration to be correct. The score can 
also be thought of as a measure of how good that eonfiguration 
looks in relation to others as a candidate for extension. One 
question which was previously left unanswered, how a subset of 
the active confisurations is chosen for extension, can now be 
answered : the subset of maximally scoring configurations is 
chosen at each step until the maximal score of act ivt! 
configurations begins to fall. (The score on a configuration and 
the score of a path terminatinr on that configuration are the 
same thing -- we will use which ever terminology seems most 
natural at the time.) 
The result of this process is a sort of modified breadth 
first approach, where at one step all the alternatives are tried 
but at the next step only th& best ones are chosen for further 
extension. This is similar to the best-first parser describad by 
Paxton in [I81 but it can be applied to the sort of partial paths 
which SPARSER generates rather then requiring the perfect 
Page 42 
information resulting from a strictly left to ripht approach. 
The auccesg of this method is directly dapandent on the ralativz 
accuracy of the scores which are assigned to the paths. 
5.2 SCORING PATHS 
Sevdral attempts have b~en made to d*velop rigorous systems 
for parsing arrorful or spedch-like input baaed on probabilities 
[I, 14, 271. These attempts have all simplified the problem to 
such an axtent that it is no lonper realistic or extendible, e.p. 
by assuming the input is a sequence (rather than a lattice) of 
probability distributions, by assuming that all the neclsbary 
information is present in the searhh space to begin with so the 
only problem is to find an optimal path throuph ths spacz, by 
requirine a small vocabula~y, and/or by limiting the Rranmar to 
be context free. 
The ideal scoring mechanism for SPARSER would be one which 
accurately reflected at every step the probability that the path 
8 correct. Bayas rule could be Used, but it would ba 
necessary to know, at any point in ths parsing process, what the 
probability is that th~ next arc under consideration is correct, 
given that the entire path up to the current step is correct. In 
order to use this application of Bayas rule it would be 
necessary to pra-calculate the probabilitiss for evary possible 
path and partial path which could be generated -- a clearly 
impossible task sincs there are an inftnite number of such paths. 
Givzn that we cannot calculate the probabilities we need 
exactly, what is the next best option? If we ignore the effect 
of tila path traversed up to the current point, but can say for 
ray ~iven state how likely each arc em an at in^ from that state is 
to ba correct, we would have a model which uses only local 
Paforaation rather than one which takes into account accurately 
all1 tbe Left context which is available. 
Since it was not practical to run large amounts of data 
tbm~h a parser in order to obtaid accurate measurements even 
Sor the limited model, the author re'lied on considerable 
experfeace with ATN grammars to assign a weight to each arc of 
tba grqmmar representing tho intutive likelihood that the arc 
fii it can be taken) is the correct cane to choose from that 
state. These weights are small integers (0 throu~h 5) -- the 
jarger the weight the more likely the arc. 
The question might arise as to why the score oG the word 
sate4 used by an arc should not be used to influence the score of 
tbu path using it. SPARSER tries to treat each theory as 
ladrependently as possible and tq assign scores based only on the 
syntactk information which is available. The one exception to 
this rule is the semantic information which is used to score 
constituents. If lexical Qord match scores were used, the 
cbmtrol component would not be able to separate the lexical 
goudwss from the syntactic goodness of the theory and make 
Judgments aa to their relative importancz. In a syntax-driven 
apeecb understanding system, however, it would probably bz usaful 
to combine lexical scores with syntactic information. 
Page 44 
As was described in the previous section, when SPARSER 
begins to parse an island each possible partial path is begun by 
creating a configuration at the head of a transition for an arc 
which can use th@ current word. Rather arbitrarily, it was 
decided to giva this confi~uratioq a score of one. This starts 
all partial paths out equally, a technique which is not quite 
accuratb, since some contexts are more likely than othms. For 
example, the words tttom and "forw are more likely to occur In 
prepositional phrases than in sentantial complements. If this 
simplification appaarv to harm the overall performance of 
SPARSER, it coula be remadikcf by giving eaah state an a priori 
score similar to the weights on arcs. Configurations on lead-id 
paths are also given a score of one. 
After the initial step, whenaver a transition (othar than a 
PUSH or POP) is made, the score of the subsequent configuration 
is influenced by the score of the conf'i~uration being extended 
and the weight on the arc beins us$d. If the scores were actual 
probabilities; they would be multiplied; since they are not, it 
was arbitrarily decided to add them. 
When attempting to create a configuration which already 
exists (a situation encountered whenever two or more parse paths 
for the same theory merge), ths configuration is given the 
maximum of the sxisting score and the score which would haw been 
assigned had the configuration been created anew. 
Whsn a PUSH arc is encountered and a configuration created 
to begin the search for the required constituznt, the score of 
that configuration is set to be the sum of the scora of the 
configuration causing the PUSH, and the value (If any) of the 
look-ahead test on the PUSH arc. For example, upon encountarinr 
an arc such as (PUSH NP/ ((NPSTART) T T) ... ) the look-ahead 
function NPSTART returns a high integer valua ir the next word is 
a noun and a lowen valua if it is a verb (e.g. ltaccounting 
coststt). Of course, if tha look-ahead funcrion fails altogether, 
the c~nfipuration is not set up, althou~h hhe monitor in the WFST 
remains. 
When a constituent is completed (or found in the WFST) end a 
PUSH tranaitiqn is about to be made, the score of the 
confi~uration on Ghich the transition terminates is a function ~f 
the score of the confi~uration heinc extended the weipht on the 
arc, an@ the score of the constituent itself. The score of tht* 
constituent is currently very ad hoc, bdng a function of the 
number of words in th* constituent (lass a function of the number 
of sub-constituents subsumed by this constituent, boosted if the 
constituent is a mador one) and the score which is determined by 
semantic verification. Thus semantically ''roodIf constituents 
will st the scores of the paths which use them more than 
semantically "badw ones. 
Due to the level of effort required to gather accurate 
statistics on the relative frequencies of arcs, the current 
scores are admittedly ad hoc. It is not clear whether different 
scoring mechanisms would be better, however it is clear that the 
current scoring strategy is better than no scoring at all, as 
praiiminary measurements indcate that the number of transitions 
created (as well as the number of confi~urations and predicions) 
is reduced about 25% by thz current strategy. 
(It is rzasonable to ask why semantic scores are used to 
influance parse paths, sincd it was tust argued that lexical 
Ycores should not be ushd in this way Semantic scores may be 
more reliable than lexical ones because we are assuming that the 
utteran& is semantically maanibaful. Under this ausumption, a 
constitudnt like "range remainder" as a noun-noun modifier 
analogous to ltaurplus moneyff should be ruled out as early as 
possible. Since such con8tituents cannot be ruled out on 
syntactic  rounds alone, since prosodic information (which might 
help to rule them out) is not available (see discussion in 
Section 7.2), and since they would seriouslv overrun the parser 
with a plethora of false paths if they wwe not reJected, it 
seems reasonable to permit semantics to influence the parser.) 
5.3 SCORING PREDICTIONS 
The previous section discussed three ways in which SPARSER 
can make predictions about what could fill in gaps between 
islands. Monitors wait for the occurrence of a word in the word 
lattice (or a constituent in the WFST), proposals request a 
search for a particular set of words, and notices indicate the 
presence of a usable word in the word lattice (or a constituent 
in the WFST). Since the processing of a typical theory is likely 
to result in a number of predictions it is necessary t'o be able 
to order them so that predictions most likely to bz correct or 
most likely to yield important information will be acted upon 
first. For example, it is more important to fill a Rap between 
two islands than to extend a sin~le islahd, since by filling the 
Rap one can chzck the consistency of information which was 
locally good in aach island individually but may not be 
consistent when they are joined. Since two words can occur 
to~ethar in (usually) many contexts but lon~er szquances arl 
generally more restrictiv~, addine a word to a one word island is 
likely to be leas profitable in terms of the number a,f possible 
paths which are sliminatod by the addition than add in^ a word to 
a multi-word island. 
It is up to the syntactic component to indicate to the 
control component the relative importance attachsd to each notica 
and monitor; the hipher the score, the stron~er the prediction. 
Several factors influence the score attached to predictions. 
One is the length of the island to which the prediction is 
attached. One word islands, if they are processed at all, yield 
very little information anc many pradictions, bznca the 
predictions are not scored high. Proposals are lass important if 
there is already a noticsable word in the word lattice (since 
that word is acoustically better than the word to be proposed, 
else it would have bean found earlier. Howevsr, if a proposal 
fills a gap between two islands, it is given a higher score. 
Notices are boosted in importance if an entire constituent may be 
added and penalized if they will add onto a one word island. 
Scores range from 0 to 95 for proposals, 0 to 40 for notices, and 
0 to 15 for monitors. 
Page 48 
Theas scores appear to work fairly well with the rest of the 
BEN SPEECHLIS system, but have been dcvalopod by a process of 
interaction with tha other components (in order to make the 
scores of syntactic prediction8 commensurate with those of 
semantic predictions) and may be chanazd considerably au the 
atire sy~tem evolves. 
Small syntactic classes em. detePml'naru and prepositions) 
are proposed in their entirety (that is, their elurnants are to be 
dnumrratad and ~ivdn to the lexical matchinc componlnt for 
verification) if the island which monitored for them is more than 
one word lon~. If a gap batwsen two island3 is small enouch for 
iust one word and if a syntactic class has been monitored for 
L 
from both sides of the gap, it is propossd ir its entirety also. 
Sect ion 6 
Examples and Results 
SPAR3ER is written in INTERLISP and runs on a PDP-10 und~r 
the TENEX operating system . The program and initial data 
structures occupy approximately 90000 words of virtual memory. 
(The other componants of the BBN speech undzrstandicp system 
occupy separate forks from the syntactic component.) 
Page 49 
At the time the 'examples in this section were run, the 
al~orithm controlling tha dacision-making process in the control 
component was under~oing reviaion and was not solidified into a 
function which could operate automatically. Rathar, there ware a 
number of primitlvd operations such as scanning an utterance (or 
some specified partion of it), creating thoorieu, call in^ SPARSER 
with a theory or event, calling f"or the processing of proposals, 
etc., which could be invoked by a human simulator The follow in^ 
examples were produced in this mode, with the user act in^ as the 
control component in a way which could be modelled by later 
imple,mentation. 
Several convention8 have been used in tracin~ the operation 
of SPARSER. Conf ieurations are rspresented as 
NUMBER : STATE : POSITION (SCORE). For example, the 
configuration written as 30:NP/HEAD:23(39) is the confi~uration 
for state NP/HEAD at position 23 which has been given the 
(unique) numbe~ 30 and which currently has a score of 39. The 
creation of a transition is indicated by naming the type of arc 
causing the transition, the (unique) number of the transition, 
and the configurations at each and of the transition. For 
example, 
CAT N TRANS #9 FROM 14:~PhET:6(1) TO 15:NP/DET:19(4). 
Annotations havc been inserted within brackets { 1; typeout 
in upper case was produced by the program. 
Page 50 
EXAMPLE 1 
This example parallels that given in Section Four. A word 
lattice was artificially created which contained only the 
following three word matches: 
(1 SUMMER 12 16 100) 
(2 WINTER 12 16 100) 
(3 TRIP 16 21 100 -S) 
(In this version or the system, ra~ular inflectional endings are 
included in word matches after the element representing the 
score, hence the somewhat peculiar word match for the word 
"trips".) Two theories were constructed, one for word matches 2 
and 3, the other for 1 and 3. What follows is an annotated (but 
otharwisc! unedit,ed except for considerations of spacing) 
transcript of SPARSER processing these two theories in sequence, 
using the MINIGRAMMAR of Figura 3.3 and Appendix I. 
SPARSER PROCESSING THEORY'I: 
0 12 WINTER 16 TRIP -S 21 30 
(This is a linear representation of the thepry baing 
processed. The endpoints are 0 and 30, but the words 
occupy-only the middle part of the utterance.) 
STARTING AN ISLAND 
"WINTER" TRYING CAT N ARC FROM NP/ADJ TO NP/ADJ 
{This is the first of two arcs retrieved from ths index 
tables ) 
CAT N TRANS 81 FROM 1:NP/ADJ:12(1) TO 2:NP/ADJ:16(3) 
(The first transition is created, and since th~re is ,a 
CAT N arc which enters state NP/ADJ, a monitor is set up 
to monitor for nouns which end at position 12.) 
ENDING AT 12: 
MONITORING [ N 1 
JUMP TRANS #2 FROM 3:NP/QUANT:12(1) TO 1:NP/ADJ:12(1) 
(Now the lead-in transitions are being created, along 
with the monitors far syntactic categories which may 
precede the newly constructed configurations. 
Configurations along the lead-in path are all assigned a 
score of 1.3 
MONITORING [ ADJ 1 
JUMP TRANS #3 FROM 4:NP/ART: 12( 1 ) TO 3:NP/QUANT: 12( ?) 
ENDING AT 12: 
MONITORING [ ART 1 
JUMP TRANS #4 FROM 5:NP/:12(1) TO 4:NP/ART:12(1) 
{The laad-in transitions are all made. Now the second 
arc which can uae the noun ia about to be processed.) 
"WINTER" TRYING CAT N ARC FROM NP/ADJ TO NP/N 
CAT N TRANS #5 FROM 1:NP/APJ:12(1) TO 6:NP/N:16(6) 
{This is tha second of thd two arcs obtained from the 
index table for "winterw. Th* lead-ln transitions to 
configuration 1 havd already been Constructed, so they 
are not remade. Now we are ready to choose 
configurations to extend. The pool of candidates for 
extension contains confi~urations 2 and 6.) 
SELECTED CONFIGS (6) FOR EXTENSION 
[Only this one is chosen because it has a hi~har score 
than configQration 2, since the use of a noun as a head 
noun of a noun phraac! is more likely than its use au a 
modifier.) 
PICKING UP CONFIG 6:NP/N':16(6) WITH WORD TRIP 
TRYING PUSH PP/ ARC 
{No.action is taken about starting a configuration for 
state PP/ because the look-ahead test which checks that 
the next word can begin a prepositional phrase fails on 
the word trip.) 
TRYING POP ARC 
POP TRANS 86 FROM 6:NP/N:16(6) 
{Creating ths POP transition completes a gath from 
configuration 5. Thz path is expressed as a list of 
transition numbers. We are about to-execute the path, 
that is, check the context-'sensitlvs tests and db tha 
re~ister build in^ actions along it.) 
EXECUTING PATH (4 3 2 5 6) 
BEGINNING AT TRANS 4, CONFIC 5 
{We must ba~ln axecutinp the path at the first 
transition, because no part of it has been executed 
before. Later we will see that it is possible to begin 
execution of a path in the middle, since th* repister 
contents are stored at each step.] 
DOING JUMP ARC FROM 5:NP/ TO 4:NP/ART 
DOING JUMP ARC FWM 4:NP/ART TO 3:NP/QUANT 
DOING JUMP ARC FROM 3:NP/QUANT TO I:NP/ADJ 
DOING CAT ARC WITH WINTER FROM 1 :NP/ADJ TO 6: NP/N 
DOING POP ARC FROM 6:NP/N 
TEST FAILED 
{Tha test failed because thzre is no determiner, and 
MINIGRAMMAR requires that singular, undetermined nouns 
can be complete noun phrases only if they are mass 
nouns. "Winterv is not marked as a mass noun in our 
dictionary, hence it will not parse as a complete noun 
phrase. ) 
Page 52 
SELECTED CONFIGS (2) FOR EXTENSION 
{Since oxtanding confi~uration 6 did not do much for us, 
we go back to try tha lower scoring confi~uration 2.) 
PICKING UP CONFIG 2:NP/ADJ:16(3) WITH WORD TRIP 
TRYING CAT N ARC 
CAT N TRANS 17 FROM 2:NP/ADJ:16(3) TO 7:NP/N:21(8) 
TRYING CAT N ARC 
CAT N TRANS 18 FROM 2:NP/ADJ:16(3) TO q:NP/hDJ:21(5) 
SELECTED CONFIGS (7) FOR EXTENSION 
{Again, the hi~her  coring of the two active 
configurations, 7 and 8, is chosen.) 
PICKING UP CONFIG 7:NP/N: 21 (8) 
STARTING AT 21: 
MONITORING [ PP/ 1 
SETTING UP CONFIG ?:PP/r21(8) 
MONITORLNG [ PREP ] 
(Since there is no next word to test, a confipuration is 
set up to begin processing a prepositional phrase, and 
the? syntactic catogories'which @an be~in such a phrase 
-- in this case, only one -- are monitorzd for.) 
TRYING POP ARC 
POP TRANS #g FROM 7:NP/N:21(8) 
EXECUTING PATH (4 3 2 1 7 9) 
B'EGINNING AT TRANS 1, CONFIG 1 
{Creation of the POP trans completed a path, the first 
part of which has already -bean exicutid. We can 
therefore pick up in the midads. of' the path and execute 
only the last three transitions.) 
DiIItlG CAT AliC I ! E'HL'E! 1 : flP/A?J TO 2: NP/ADJ 
DOING CAT ARC WIT[; '.'RIP FROt-1 2:?!E3/ADJ TO 7:NP/N 
DOING POP ARC FROM 7:NP/N 
**#* 
MADE #1 FROIi 12 TO 21: 
NP ADJ NP N WINTER 
NU SG 
N TRIP 
NU PL 
+**+ 
{Thz path succeeds -- no determiner is needed since the 
head noun is plural -- and a constituent is constructed. 
The semantic camponent has been turned eff for this 
example, so it adds nothin~ to the acore r:hich SPAzSZS 
wssiun$ -- 5 poifits for each word in the constit~lent.~ 
SYIJ WEIGHT + SEM WT = 10 + 0 = 10 
{No monitor exists in the WFST for a NP/ at this placz, 
so the arcs (in NINIGRAMMAR there is only one) which 
could push for a NP are processed bottom up in exactly 
the same manner as the two arcs which couqd use a noun 
at the beginninp of the island.) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #I0 FROM 10:PP/PREP:12(1) 
Page 53 
MONITORING [ PREP ] 
1 
SELECTED CONFIGS (11) FOR EXTENSION 
PICKING UP CONFIG 11:PP/NP:21(7) 
TRYING POP ARC 
POP TRANS #11 FROM 11:PP/NP:21(7) 
SELECTED CONPIGS (8) FOR EXTENSION 
PICKING UP .CONFIG 8:NP/ADJ:21(5) 
STARTING AT 21: 
MONITORING [ N ] 
MONITORING [ N 1 
ALL ARCS TRIED AT THIS CONFIG 
(Now the thewy has been procdssed. There followa a 
summary of the proposals, monitors, and notlces 
constructed. The syntactic wore assigned to the theory 
is ~ivon -- here just the acore of the conutituent 
constructed. Then there is a summary of ~tatistics.) 
PREDICTIONS: 
MONITORENG [ PREP 1 STARTING AT 21, SCORE 10 
MONITORING [ N ] STARTING AT 21, SCORE 10 
MONITORING [ N ] ENDING AT 12, SCORE 10 
MONITORING [ QUANT 1 ENDING AT 12, SCORE 10 
MONITORI~G [ ADJ 1 ENDING AT 12, SCORE 10 
MONITORING [ ART j ENDING AT 12, SCORE 10 
MC?~ITORING [ PREP ] ENDING AT 12, SCORE 10 
PROPOSING (QUANT ART PREP) ENDING AT 12 
FINISHED THEORY 1 WITH SYN SCORE 10 
{Exclusive of tracin~ and fork interactions, this 
processing took 5.5 seconds.) 
{Now we are ready to process the second theory 
syntactically.) 
SPARSER PROCESSING THEORY 2: 
0 12 SUMMER 16 TRIP -S 21 30 
STARTING AN ISLAND 
nSUMMER" TRYING CAT N ARC FROM NP/ADJ TO NP/ADJ 
CAT N TRANS 112 FROM 1:NP/ADJ:12(1) TO 2:NP/ADJ:16(3) 
{This transition completes a path which includes 
transitions and configurations constructed auring the 
pravious theory.) 
EXECUTING PATH (4 3 2 12 7 9) 
BEGINNING AT TRANS 12, CONFIG 1 
DOING CAT ARC WITH SUMMER FROM 1:NP/APJ TO 2:NP/ADJ 
DOING CAT ARC WITH TRIP FROM 2:NP/ADJ TO 7:NP/N 
+*+u -DOING POP ARC FROM 7:NP/N 
MADE #2 FROM 12 TO 21 : 
NP ADJ NP N SUMMER 
NU) SG 
N TRIP 
NU PL 
**** 
SYN WEIGHT + SEM WT = 10 + 0 = 10 
NP/ WAS PUSHED FOR AT CONFIG 10 
{This time there are monitory in the WFST, one which is 
looking for a NP start in^ at position 12 and one which 
is looking for a NP ending at position 21. One 
transition is sufficient to satisfy both of these, ant! 
the preposition needed to complete a PP/ is monitored 
for.) 
PUSH NP/ TRANS- 61 3 FROM TO: PP/PREP: 12( 1 ) 
TO tl:PP/NP:21(8) 
NPI MAY LEAD TO CONFIC 11 
{This is caused by the fact that there was a monitor Tor 
a , noun phrase ending at confipurat ion 11 -- the one 
craated when constituent 1 was made. The transit ion 
which would bd 8at up is the transition Just created, so 
it is not remade. 
All of the processinp which resulted from the 
completion of a constituent is finished; however there 
are honitors still to be set for configurations alone 
the path.) 
ENDING AT 12: 
MONITORING [ PREP ] 
ENDING AT 12: 
MONITORING [ N ] 
ENDING AT 12: 
MONITORING [ QUANT ] 
MONITORING [ ADJ 1 
ENDING AT 12: 
MONITORING [ ART ] 
(Since each monitor consist8 of thz item beinc monitored 
for, its associated test (if any) th~ theory which is 
to be notified when the monitor is satisfied, and the 
configuration and arc causing the monitor, monitors must 
be mad* anew each time one of the elements changes, 
although some of the list structure can be shared, hence 
thz seeming proliferation ofmonitors.) 
{N80w SPARSER procassss the othsr arc which could uss the 
word "summerw. ) 
"SUMMER" TRYING CAT N ARC FROM NP/ADJ TO NP/N 
CAT N TRANS #14 FROM l:NP/ADJ:12(1) TO 6:NP/N:16(6) 
EXECUTING PATH (4 3 2 14 6) 
BEGINNING AT TRANS 14, CONFIG 1 
DOING CAT ARC WITH SUMMER FROM 1:NP/ADJ TO 6:NP/N 
DOING FOP ARC FROM 6:NP/N 
TES? FAILED 
i&cauae, "sum apn cannot be a complete noun phrase in 
s grammar. Y 
Page 55 
SELECTED CONFIGS (11) FOR EXTENSION 
PICKING UP CONFIG 11:~~/NP:21(8) 
TRACING POP TRANS 11 FROM 11:PP/NP:21(8) 
[This transition was created before, but is now mads 
part of the currant theory. It doas not compllte a pqth 
or cause any further action. If it had a trrminatinff 
configuration, i.e. if a transition other than a POP 
tranai tion hao been traced , the terminatinp 
configuratdon would have been placed on the list of 
possible confipurations to extend.} 
SELECTED CONFIGS (6) FOR EXTENSION 
PICKING UP CONFIO 6:NP/N:e16(6) WITH WORD TRIP 
TRACING POP TRANS 6 FPOM 6:~~/~:16(6) 
SELECTED CONFIGS (2) FOR EXTENSION 
PICKING UP CONFIG 2:Np/A~J:16(3) WITH WORD TRIP 
TRACING CAT N TRANS 8 USING "TRIP" FROM 
2:NP/ADJ:16(3) TO 8:NP/ADJ:21(5) 
STARTING AT 21 : 
MONITORING [ N ] 
MONITORING [ N ] 
{Tharz are two noun arcs Leaving state NP/ADJ, hence two 
monitors.) 
SELECTED CONFIGS (8) FOR EXTENSION 
PICKING UP CONFIG 8:NP/ADJ:21(5) 
ALL ARCS TRIED AT THIS CONFIG 
PREDICTIONS: 
MONITORING [ N ] STARTING AT 21, SCORE 10 
MONITORING r PREP ] ENDING AT 12, SCORE 10 
MONITORING [ N ] ENDING AT 12, SCORE 10 
MONITORING [ QUANT ] ENDING AT 12, SCORE 10 
MONITORING [ ADJ ] ENDING AT 12, SCORE 10 
MONITORING [ ART 1 ENDING AT 12,' SCORE 10 
PROPOSING (PREP QUANT ART) ENDING AT 12 
FINISHED THEORY 2 WITH SYN SCORE 10 
{The processing of this theory tbok approximately 4.5 
seconds. ) 
This example has shown the trace produced by runninp SPARSER 
on input which is analogous to the example presented with 
illustrations of the map in Section Four. The Interested reader 
is urged to draw his own maps while reading the follow in^ 
Pact! 56 
examples in order to best understand the dynanic operation of 
SPARSER. 
EXAMPLE 2 
I-- - 
This example is more realist ic t ban the previous cntb -- it 
shows the operation of SPARSER in the context of an pt terance 
which has been 2utornatically segmented and labelee, ~ith the 
lexical rr;trieval and match component in operat ion. It 
demonstrates how SPARSER can help to select the best set vf words 
from a or complex word lattict. This example uses the 
SPEECHGRAMMAR described in [4]. 
The utterance "What is the registration fee?" was spoken by 
an adult male speaker in a quite room and was record on tape. 
Tha ut teranca was automatically diqit ized rind passed through the 
warnentat ion and labelinr routines of the BBN speech 
understandins ~ystem. The initial scan of the utterance, usinp 
the lexical retrieval component, produced a w~rd lattice of 
fifteen entries, includin~ several for inflectional endings. (In 
this version of thl system, they were not combined with the root 
form into a single word hatch, and hsncr? could match evan without 
a root word.) Tha format for a word match is: 
(NUMBER WORD LEFT-END RIGHT-END LEXICAL-SCORE). 
(2 WHAT 0 3 191) 
(3 ONE 0 3 189) 
(11 WHEN 0 3 102) 
(9 THE 4 6) 
( 1 REGISTRATION 6 19 237) 
(10 REGISTRATION 7 19 103) 
(5 HAS 9 12 121) 
Page 57 
The two best matches, for "what1! and ftrzgistrationll, appear 
to be good candidates for a theory, so we begin by build in^ and 
procasring that theory. 
ARSER PROCESSING THEORY 1: 
%PWHL~ 3 ' 6 REGISTRATION 19 
23 
STARTING AN ISLAND 
STARTIHG AT LEFT END OF SENTENCE 
(Itno~ing that it is not necsssary to go through the 
usual startup procedure for islands when beginning an 
island at position 0, SPARSER starts with a 
configuration for state 3/ at position 0.) 
SELECTED CONFICS (1) FOR EXTENSION 
PICKING UP CONFIG 1:S/:0(1) WITH WORD WHAT 
TRYIIG JUMP S/Q ARC 
JUMP TRANS #I FROM 1:S/:0(1) TO 2:S/Q:0(6) 
SELECTED CONFIGS (2) FOR EXTENSION 
PICKXICG UP CONFIG 2:S/O:0(6) WITH WORD WHAT 
TRYIMG PUSH NP/ ARC 
HONrTORING 1 NP/ ] 
SETTING UP CONFIG 3:NP/:0(11) 
TBYIHG CAT QWORD ARC 
CAT QWORD TRANS #2 FROM 2:S/Q:0(6) TO 4:S/NP:3(11) 
SELECTED COHFIGS (3 4) FOR EXTENSION 
{This time two active configurations have the same 
maximal score, so they are both processed.) 
PIafffi UP COIMG 3:NP/:0(11) WITH WORD WHAT 
TPTIYC GAT QDET ARC a 
CAT QOET TRANS #3 FROM 3:NP/:0(11) TO 5:NP/ORD:3(16) 
ma= UP CONFIG ~:s/NP:~(II) 
STAPIIMG At 3: 
HMITO~lrG [ MODAL 1 
WI~~~ING I V 1 
Page 58 
TRYING POP ARC 
POP TRANS #4 FROM 4:S/NP:3(11) 
EXECUTING PATH (1 2 4) 
BEGINNING AT TRANS 1, CONFIG 1 
DOING JUMP ARC WITH WHAT FRdlY l:S/ TO 2:S/Q 
DOING CAT ARC WITH WHAT FROM 21S/Q TO 4:SINP 
DOING POP ARC FROM Q:S/WP 
TEST FAILED 
{This test failed because the crammar does not allow 
'lwhatw to be a complete sante.nca.) 
SELECTED CONFIGS (5) FOR EXTENSION 
PICKING UP CONFIG 5:NP/ORU:3(16) 
STARTING AT 3: 
MONITORING [ QUANT/ ] 
SETTING UP CONFIG 6: QUANT/ : 3 ( 16 ) 
(Hare all the words which can start quantifiers, like "a 
hundredw or "point fivam, ard proposed. The grammar 
does not preclude a quantifier following a 
que$tion-determihzr, e-g. "What three men traveled to 
Spain'llI. ) 
{MONITORING [ INTEGER ZERO NO POINT A] 
For considerations of space, long listings of monitors 
and proposals in this example will be compacted as shown 
here. Such alterations to the actual trace produced 
will be surrounded by brackets.) 
TRYING JUMP NP/QUANT ARC 
JUMP TRANS #5 FROM 5:NP/ORD:3(16) TO 7:NP/QUANT:3(21) 
SELECTED CONFIGS (7) FOR EXTENSION 
PICKING UP CONFIG 7 : NP/QUANT :,3 (2 1 ) 
TRYING JUMP NP/DET ARC 
JUMP TRANS 86 FROM 7:NP/QUANT:3(21) TO 8:NPIDET:3(26) 
SELECTED CONFIGS (8) FOR EXTENSION 
PICKING UP CONFIG 8:NP/DET:3(26) 
STARTING AT 3: 
MONITORING [ NPR/ NPP/ 1 
{There are two PUSH NPR/ arcs from this state so two 
monitors are craated, but only one configuration is set 
up- 1 
SETTING UP CONFIG g:NPR/:3(26) 
{MONITORING [NPR NPR N ADJ N V ADV]) 
TRYING JUMP NP/HEAD ARC 
JUMP TRANS #7 FROM 8:NP/DET:3(26) TO 10:NP/HEAD:3(29) 
SELECTED CONFIGS (10) FOR EXTENSION 
PICKING UP CONFIG 10:NP/HEAD:3(29) 
{This is an ex~mple of  he fallibility of using only 
context free tests on partial paths, The parser thinks 
it has succassfully reached state NP/HEAD, while in fact 
tbis cannot be the case because no head nouh has bean 
dlscovarad for the noun phrase. Thus it is incorrect to 
predict relativz clauses at this point. This issue will 
be discussed in more detail be1ow.J 
STARTING AT 3: 
MONITORING R/ PP/ R/NIL 1 
SETTING UP CONFIG 11:R/:3(29) 
{MONITORING lPREP WHOSE WHO WHICH THAT WHOM]} 
(PROPOSING LYHOSE1l lfWHOw 'WHICH" "THAT" "WHOM"] 
SETTING UP CONFIG 12:PP/:3(29) 
MONITORING [ PREP 1 
SETTING UP CONFIG 13: R/NIL: 3(29) 
MONITORING [ THERE ] 
PROPOSING "THERE" 
TRYING POP ARC 
POP TRANS 88 FROM 10:NP/HEAD:3(29) 
EXECUTING PATH (3 5 6 7 8) 
BEGINNING AT TRANS 3, CONFIG 3 
DOING CAT ARC WITH WHQ FROM 3:NP/ TO 5:NP/ORD 
DOING JUMP ARC FROM 5:NP/ORD TO 7:NP/QUANT 
DOING JUMP ARC FROM 7:NP/QUANT TO 8:NP/DET 
DOING JUMP ARC FROM 8:NP/DET TO 10:NP/HEAD 
TEST FAILED 
{A question-determiner alone cannot bs a complete noun 
phrase; although this is permitted by considering 
"whatw as a QWORD as in tranaition #2.) 
STARTING AN ISLAND 
"REGISTRATION" TRYING CAT N ARC FROM NP/DET TO NP/DET 
CAT N TRANS #g FROM 14:NP/DET:6(1) TO 15:N~/DET:19(4) 
{This arc is using wrsgiatsationv as a noun modifier for 
Qome future head noun.) 
ENDING AT 6: 
{MONITORING [ NPR/ ADJ N V ]} 
JUMP TRANS 110 FROM 16:NP/QUANT:6(1) TO 14:NP/DET:6(1) 
ENDING AT 6: 
MONITORING [ QUANT/ ] 
JUMP TRANS 811 FROM 17:NP/ORD:6(1) TO 16:NP/QUANT:6(1) 
ENDING AT 6: 
(MONITORING [ ORD QDET ONLY I) 
PROPOSING "ONLY" 
JUMP TRANS #I2 FROM 18:NP/ART:6( 1 J TO 17:NP/ORD: 6{ 1 ) 
ENDING AT 6: 
(MONITORING [ ART QUANT POSS k1HOSE I} 
NOTICING IITHE" 
PROPOSIllG "WHOSE1' 
JUMP TRANS W13 FROM 19:MP/ONLY:6(1) TO 18:NP/AHT:6(1) 
El.IDIIJG AT 6: 
llOMITORING [ ONLY 1 
PROPOSING 'lfObJLY" 
JUMP TRP!IS #14 FROM 20:NP/:6(1) TO 19:NP/ONLY:6(1) 
HEGISTRATION" TRYING CAT N ARC FROM NP/DET TO NP/HEAD 
CAT N TRANS #J5 FROM 14:NP/DET:6(1) TO 21:NP/HEAD:19(6) 
{This arc is using ltregistrationtl as the head noun of a 
noun phrase. ) 
SELECTED CONFIGS (21) FOR EXTENSION 
PICKING UP CONFIG 21 : NP/HEAD: 19( 6) 
STAHTING AT 10: 
FIONIT nxwc [ R/ PP/ W/NIL I 
SETT 9 NG UP CONFIG 22:R/:lQ(6) 
{MONITORING [PREP WHOSE \,!I10 WHICH THAT WHOM] ) 
SETTING UP CONFIG 23:PP/:19(6) 
MONITORING [ PREP ] 
SETTING UP CONFIG 24:R/NIL:19(6) 
MONITORING [ THERE ] 
NOTICING "FEEtt 
{This notice is in response to the look-ahead test on 
the push arc to atate R/NIL. Since 'vfee'l can start a 
reduced re&ative clause, it is noticed, but there is not 
a specific monitor set up becauue the arc within the 
relativd clause network which will actually process the 
word Itfed" is not known.) 
TRYING POP ARC 
POP TRANS #16 FROM 21:NP/HEAD:19(6) 
EXECUTING PATH (14 13 12 11 10 15 16) 
BEGINNING AT TRANS 14, CONFIG 20 
TEST FAILED 
{The path failed because thare is no determiner for 
tlregistrat ion. "1 
PREDICTIONS: 
NOTICING (4 FEE 19 23 155 Q), SCORE -5 
NOTICING (9 THE 4 6 103 01, SCORE 0 
PROPOSING (ONLY WHOSE) ENDING AT 6 
PROPOSING (ZERO NO POINT A WHOSE WHO WHICH THAT WHOM THERE) 
STARTING AT 3 
{MONITORING [ PREP ] STARTING AT 19, SCORE 0 
MONITORING [ WHOSE WHO WHICH THAT WHOM THERE ] 
STARTING AT 19, SCORE 5 
MONITORING [ ADJ N V ORD QDET ART QUANT POSS ] 
ENDING AT 6, SCORE 0 
MONITORINC [ WHOSE ONLY ] ENDING AT 6, SCORE 5 
MONITORINC [ MODAL V INTEGER NPR N ADJ V ADV PREP ] 
STARTING AT 3, SCORE 0 
MONITORING [ WHOSE WHO WHICH THAT WHO11 THERE ZERO NO POINT A ] 
STARTING AT 3, SCORE 5) 
PROPOSING (V N ADJ) FROM 3 TO 6 
{Proposals were made to fill the gap because there were 
monitors from both sides of a pap small enoush to 
contain one word.) 
FINISHED THEORY 1 WITH SYN SCORE 0 
{It took 11.9 seconds to process this theory.) 
{Processing the proposal-s just made results, notably, in 
the detection of the word lvotherv between vv~hatvl and 
wre~istrationlv, but the word match score is very low. 
Word matches for *isw and Itare" from position 3 (next to 
"whatvv) to position 4 are also found, 
but since they do 
Page 61 
not fill the pap, the event scores are low. The bzst 
event is that for the word "faen. Procassin~ it is 
fairly uninteresting, since it completes no constituent, 
so we will omit the trace of that event. After it has 
bean processed, however, the best event is that for the 
word "thevf and the theory just created. ] 
SYNTAX PROCESSING EVENT FOR THEORY#2 
WITH NEW WORD (4 THE 6) 
TO GET NEW THEORY#3. 
0 WHAT 3 4 THE 6 R,EGISTRATION 19 FEE 23 
"THE" TRYING (CAT ART --) FROM STATE NP/ONLY TO CONFIG 18 
CAT ART TRANS #26 FROM ~~:NP/ONLY:~(~) TO 18:NP/ART:6(6) 
ENDING AT 4: 
MONITORING [ ONLY ] 
PROPOSING "ONLY" 
JUMP TRANS #27 FROM 33:NP/:4(1) TO 32:NP/ONLY:4(3) 
EXECUTING PATH (27 26 12 11 10 9 22 25) 
BEGINNING AT TRANS 22, CONFIG 15 
+*++ 
MADE #l FROM 4 TO 23: 
NP DET ART THE 
ADJ NP N RECISTRATION 
NU SC 
N FEE 
FEATS NU. SG 
**++ 
 ha format of this noun p5rase is slightly different 
from that in the previous example because tpe ~tructure 
building action for noun phrases in SPEECHGRAMMAR is 
different from that in MINIGRAMMAR. 
There ara many places in the SPEECHGRAFIMAR which push 
for noun phrases, and since there were no monitors in 
the WFST which can us& thiv constitusnt, all of them 
must be tried, resulting in a numbdr of predictions and 
notices. ) 
SYN WEIGHT + SEM WT = 15 + 0 = 15 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #28 FROM 34:FOR/FOR:4(1) TO 35:T0/:23(10) 
ENDING AT 4: 
MONITORING [ FOR ] 
PROPOSING "FO' 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 129 FROM 36:PP/PREP:4(1) TO 37:PP/NP:23(10) 
ENDING AT 4: 
MONITORING [ PREP ] 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 130 FROM 38:R/NIL:4(1) TO 39:S/NP:23(9) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 131 FROM 40:R/WH:4(1) TO 39:S/NP:23(9) 
ENDING AT 4: 
MONITORING [ R/WHOSE 1 
MONITORING [ WHICH THAT WHO WHOM WHICH WHOM ] 
{PROPOSING "THATf1 llWHO1l llWHOMu "WHICH" "WHOM") 
{There arc two arcs entwine state R/WH which use the 
words l'whichn and llwhomv. There is a check made to see 
that duplicate proposal8 alre not actually communicated 
to the control component, although they appear to be 
duplicated in the trace. } 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #32 FROM 41:S/DCL:4(1) TO 39:S/NP:23(9) 
JUMP TRANS #33 FROM 42:S/:4(1) TO 41:S/DCL:4(1) 
ENDING AT 4: 
MONITORING [ PP/ I 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #34 FROM 43:S/NO=SUBJ:4(1) TO 
44:VP/V:23(9) 
JUMP TRANS #35 FROM 45:S/AUX:4(1) TO 43:S/NO-SUBJ:4(1) 
ENDING AT 4: 
{MONITORING [ MODAL NEG V 1) 
(NOTICING I~IS~' tl~~~v} 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #36 FROM 46:S/Q:4(1) TO 39:S/NP:23(9) 
ENDING AT 4: 
MONITORING [ QADV 1 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #37 FROM 47:VP/MEAD:4(1) TO 48:VP/NP:23(?) 
ENDING AT 4: 
MONITORING [ PARTICLE 1 
MONITORING [ V ] 
JUMP TRANS #38 FROM 49:VP/V:4(1) TO 47:VP/HEAD:4(1) 
ENDING AT 4: 
{MONITORING [ NP/ NP/ V V ADV V ]} 
{NOTICING "IS" "ARE" "IS" "AREr1} 
{The words "isV1 and failed the? context fres test 
on the arc causing tha last monitor, hsnce they are not 
noticed. } 
JUMP TRANS 839 FROM 45:S/AUX:4(1) TO 49:VP/V:4(1) 
JUMP TRANS /I40 FROM 43:S/NO-SUBJ:4(1) TO 49:VP/V:4(1) 
JUMP TRANS #41 FROM 43:S/NO=SUBJ:4(1) TQ 49:VP/V:4(1) 
JUMP TRANS 842 FROM 50:S/THERE:4(1) TO 49:VP/V:4(1) 
ENDING AT 4: 
!lONITORING THERE J 
PROPOSING "THEREvf 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #43 FROM 51:VP/NP:J+(1) TO ~~:VP/VP:Z~(?) 
ENDING AT 4: 
MONITORING [ NP/ 1 
JUMP TRANS #44 FROM 47:VP/HEAD:4(1) TO ~I:VP/NP:~(I) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #45 FROM 49:V~/V:4(1) TO 44:VP/V:23(9) 
{The creation of transition 1/27 completed 3 paths. The 
first two have bean executed, resulting respaatively jn 
failure and the completion of a constituent with all the 
processing that entails. Now the third path is still 
pending and is about to be executed.] 
EXECUTING PATH (27 26 12 11 10 15 16) 
BEGINNING A? TRANS 15, CONFIG 14 
DOING CAT ARC WITH REGISTRATION FROM 14:NP/DET TO 
2l:NWHEAD 
DOING POP ARC FROM 21tNP/HEAD 
**+* 
MADE #2 FROM 4 TO 13: 
NP DET ART THE 
N REGISTRATION 
FEATS NU SG 
*#** 
{This constituent can now satisfy the monitors set by 
the discovery of the largar one, result in^ in the 
creation of many new transitions but no new 
predictions.) 
SYN WEIGHT + SEM WT = 10 + 0 = 10 
NP/ WAS PUSHED FOR AT CONFIG 34 
PUSH NP/ TRANS C46 FROM 34:FOR/FOR:4(7) TO 53:T0/:19(8) 
{Similar tIP/ transitions are ast up at configurations 
36,38,40,41,43,46,47,49, and 51 because of the monitors 
set when the first constituent was found.) 
SELECTED CONFIGS (55 54 39 37) FOR EXTENSION 
{Because these ara the maximally scoring configurations 
from ths large pool of possibilities.) 
PICKING UP CONFIG 55:S/NP:19(8) WITH WORD FEE 
TRYING POP ARC 
POP TRANS 856 FROM 55:S/NP:19(8) 
EXECUTING PATH (48 56) 
BEGINNING AT TRANS 40, CONFIG 38 
TEST FAILED 
EXECUTING PATH (33 50 56) 
BEGINNING AT TRANS 33, CONFIG 42 
++** 
MADE #3 FROM 4 TO 19: 
S NPU 
NP DET ART THE 
ADJ NP N REGISTRATION 
NU SG 
N FEE 
FEATS NU. SG 
WITH FEATURES (NPU) 
**** 
{Here is an example of a constituent which has features 
attached to it. The feature NPU can be tested by tne 
semantic component to determine that the constituent is 
a noun phrasa utterance. If nscessary, it could also be 
tasted on a PUSH S/ arc in the prammar; since there 
are 
aome times , e.~. durin~ the construction of a 
sentantial complement, when an embedded sentence must 
contain a verb.) 
SYN WEIGHT + SEM WT = 10 + 0 = 10 
{No arcs in this  ramm mar push for noun phrase 
utterances, so this constituent is not used furthar.) 
PICKING UP CONFLG 54:PP/NP:19(8) WITH WORD FEE 
TRYING POP ARC 
POP TRANS #57 FROM 54:PP/N~:1?(8) 
PICKING UP CONFIG 39:S/NP:23(9) 
TRYING POP ARC 
POP TRANS #58 FROM ~~:s/NP:?~(!J) 
EXECUTING PATH (30 58) 
BEGINNING AT TRANS 30, CONFIG 38 
TEST FAILED 
EXECUTING PATH (33 32 58) 
REGINNING AT TRANS 32, CONFIG 41 
**** 
MADE #4 FROM 4 TO 23: 
S NPU 
NP ADJ NP N REGISTRATION 
NU SG 
DET ART THE 
N FEE 
FEATS NU SG 
WITH FEATURES (NPU) 
**** 
SYN WEIGHT + SEM WT = 15 + 0 = 15 
PICKING UP CONFIG 37:PP/NP:23(10) 
TRYING POP ARC 
POP TRANS 659 FROM 37:~~/~P:23(10) 
PREDICTIONS: 
NOTICING (19 IS 3 4 -79 0), SCORE 10 
NOTICING (20 ARE 3 4 -128 O), SCORE 10 
PROPOSING (ONLY FOR WHICH THAT WHO WHOM THERE) ENDING AT 4 
{MONITORING [ ONLY FOR WHICH THAT WHO WHOM THERE 1 
ENDING AT 4, SCORE 15 
MONITORING [ MODAL NEC V QADV PARTICLE V V V ADV PREP V 1 
ENDING AT 4, SCORE 10 
MONITORING [ MODAL V INTEGER NPR N ADJ V ADV PREP 1 
STARTING AT 3, SCORE 0 
MONITORING [ 
ZERO NO POINT A WHOSE WKO WHICH THAT WHOM THERE] 
STARTING At 3, SCORE 5) 
PROPOSIHG (MODAL) FROM 3 TO 4 
PROPOSING (MODAL PREP) STARTING AT 3 
PROPOSING (PREP MODAL NEG QADV) ENDING AT 4 
CREATING THEORY 3: 
0 WHAT 3 4 THE 6 REGISTRATION 19 FEE 23 
WITH SYN SCORE 15 
{Tkis $vent took 34.5 saconds, largely because of the 
ex enslvz bottom up procdssinq necessitated by the 
Page 65 
discovery of the noun phrases which were not monitored 
for. 1 
(Processing the proposals from this theory results in 
the bavt event being the one for "isw in the last gap. 
The word "arevt also fills the gap, but the lower lexical 
acore prevents ths event for it from surfacin~. If it 
were syntactically procasued, however, no new theory 
would be created since tha completed string would be 
ungrammatical.) 
SYNTAX PROCESSING EVENT FOR THEORY#3 
WITH NEW WORD (3 IS 4) 
TO GET NEW THEORY#4: 
0 WHAT 3 IS 4 THk 6 REGISTRATION 19 FEE 23 
"IS" TRYING (CAT V --) FROM CONFIG 4 
CAT V TRANS #60 FROM 4:S/NP:3(11) TO 45:S/AUX:4(16) 
{This transition does not immediately complete any 
paths, so the best scoring configurations of the theory 
are triad.) 
SELECTED CONFIGS (31) FOR EXTENSION 
PICKING UP CONFIG 31:NP/DET:23(36) 
TRYING JUMP NPIHEAD ARC 
JUMP TRANS t61 FROM 31:NP/DET:23(36) TO 
30:NP/HEAD:23(39) 
SELECTED CONFIGS (52 48 44) FOR EXTENSION 
PICKING UP CONFIG 52:VP/VP:23(9) 
TRYING JUMP S/VP ARC 
JUMP TRANS #62 FROM 52:VP/VP:23(9) TO 59:SIVP:23(12) 
PICKING UP CONFIG 48:VP/NP:23(9) 
TRYING JUMP VP/VP ARC 
JUMP TRANS #63 FROM 48:VP/NP:23(9) TO 52:VP/VP:23(11) 
PICKING UP CONFIG 44:VP/V:23(9) 
TRYIRG JUMP VPIHEAD ARC 
JUMP TRANS P64 FROM 44:VP/V:23(9) TO 60:VP/HEAD:23(13) 
SELECTED CONFIGS (60) FOR EXTENSION 
PICKING UP CONFIG 60:VP/HEAD:23(13) 
TRYING JUMP VP/NP ABC 
JUMP TRANS 865 FROM 60:VP/HEAD:23(13) TO 48:VP/NP:23(16) 
SELECTED CONFIGS (59) FOR EXTENSION 
PICKING UP CONFIG 59:S/VP:23(12) 
TRYING JUMP S/S ARC 
JUMP TRANS C66 FROM 59:S/VP:23(12) TO 61:S/S:23(14) 
D CONFIGS (61) FOR EXTENSION 
~T~UP CONFIG 61:S/S:23(14) 
TRYING POP ARC 
POP TRANS 867 FROM 61:S/S:23(14) 
EXECUTING PATH (1 2 60 35 34 64 65 63 62 66 67) 
BEGINNING AT TRANS 34, CONFIG 43 
**** 
MADE 85 FROM 0 TO 23:. 
S Q 
SUBJ NP DET ART THE 
ADJ NP N REGISTRATION 
NU SG 
N FEE 
FEATS NU SG 
AUX TNS PRESENT 
VOICE ACTIVE 
VP V BE 
OBJ NP N WHAT 
FEATS NU SQ/PL 
#YO* 
{This iu the complete parve of the utterance, but 
SPARSER continues the operations it has pending bePore 
raturninc to Control.) 
NO SEMANTICS FOR HEAD 
{This is a comment from the sarnantic component 
indicating that it cannot currently interpret the 
construction.) 
SYK WEIGHT + SEM WT = 25 + O = 25 
S/ WAS NEVER PUSHED FOR 
PUSH S/ TRANS 868 FROM 62:COMPL/NTYPE:O(1) TO 
63:COMPL/S:23(15) 
S/ WAS NEVER PUSHED FOR 
PUSH S/ TRANS #69 FROM 64:S/THEN:O(1) TO 
65:S/IFTHEN:23(:5) 
S/ WAS NEVER PUSHED FOR 
PUSH S/ TRANS #70 FROM 66:VP/HEAD:O(1) TO 
52:VP/VP:23(13) 
JUMP TRANS #71 FROM 67:VP/V:0(1) TO 66:VP/HEAD:O(1) 
JUMP TRANS #72 FROH 68:S/AUX:0(1) TO 67:VP/V:0(1) 
JUMP TRANS #73 FROM 69:S/NO-SUBJ:O(1) TO 67:VP/V:0(1) 
JUMP TRANS #74 FROM 68:S/AUX:0(1) TO 69:S/NO=SUBJ:O(l) 
JUMP TRANS 875 FROM 69:S/NO=SUBJ:O(l) TO 67:VP/V:0(1) 
JUMP TRANS #76 FROM 70:S/THERE:O(l) TO 6?:VP/V:0(1) 
{One of the pending operations is to check thz othar 
arcs which caused monitors for the verb "isf1 1 
"IS" TRYING (CAT V --) FROM STATE S/NP TO CONFIG 45 
"IS" TRYING (CAT V --$ FROI-I STATE FOR/TO TO CONFIG 49 
CAT V TRANS 1/77 FROM 71:FOR/TO:3(5) TO 49:VP/V:4(20) 
"IS1' TRYING (CAT V --) FROM STATE VP/V TO CONFIG 49 
CAT V TRANS t78 FROM 72:VP/V:3(5) TO 49:V~/~:4(20) 
JUMP TRANS #79 FROM 73:S/AUX:3(1) TO 72:VP/V:3(5) 
JUMP TRANS #80 FROM 74:S/NO-SUBJ:3(1) TO 72:VP/V:3(5$ 
JUMP TRANS #81 FROM 73:S/AU~:3(1) TO 74:S/NO-SUBJ:3(1) 
JUMP TRANS 882 FROM ~~sS/NO-SUBJ:~(I) TO 72:VP/V:3(5) 
JUMP TRANS 883 FROM 75:S/THERE:3(1) TO 72:VP./V:3(5) 
CREATING THEORY 4: 
0 WHAT 3 IS 4 THE 6 REGISTRATION 19 FEE 23 
WITH SYN SCORE 15 
{This processing took 34.45 seconds.) 
This example was run with a vary simple, mechanical control 
structure. After the processing of the initial theory, the 
proposals which had bean made by SPARSER were processed by th+ 
lexical retrieval component and the results added to the word 
lattice -- a process which can sat off monitors and result in the 
creation of event notices. The ~v*nts are scored by a 
combination of the monitor score assigned by SPARSER and the 
lexical score asslpnad by the word match component. In this 
sentence, syntax and lexical score alone ware suffhient to make 
ths besb scorinp event at each step be one which resulted in a 
correct extension of the theory. 
Vz now ahow how the same utterance use@ in the prdvious 
example cah be recognized when dkfferznt theories qre crsated and 
when avants and theories are processed in a diffarent orda? from 
that in Example 2. Suppose that after the initial scan of the 
Ptterance the semantic component created two thzorias, one for 
the words llwhatw and "feeff and the other fori the wobds I'vhat1' and 
nrzristrationtl Let us see what happen9 in SPARSER when we hepin 
by procassipg these two tbsories in sequence. 
Page 68 
SPARSER PROCESSING THEORY 1: 
0 WHAT 3 19 FEE 23 
{The processing of this thzory iq very similar to that 
of the first theory in the previous ~xample, and will 
not be commented upon here. The purpoae in show in^ it 
is to provide a map, part of which the next call to 
SPARSER will trace. ) 
STARTING AN ISLAND 
STARTING AT LEFT END OF SENTENCE 
SELECTED CONFIGS (1) FOR EXTENSION 
PICKING UP CONFIG 1:S/:0(1) WITH WOHD WHAT 
TRYING PUSH PP/ ARC 
TRYING JUMP S/Q ARC 
JUMP TRANS #I FHOM 1:S/:0(1) TO 2:S/Q:0f6) 
TRYING WRD IF AHC 
TRYING JUMP S/IblP ARC 
TRYING JUMP S/DCL ARC 
SELECTED CONFIGS (2) FOR EXTENSION 
PICKING UP CONFIG 2:S/Q:0(6) WITH WORD WHAT 
TRYING PUSH NPI ARC 
MONITORING [ NP/ ] 
SETTING UP CONFIG 3:NP/:0(11) 
TRYING WRD HOW ARC 
TRYING CAT QWORD ARC 
CAT QWORD TRANS #2 FRO11 2:S/Q:0(6) TO Q:S/NP:3(11) 
TRYING CAT QADV ARC 
TRYShJG JUMP S/NP ARC 
SELECTED CONFIGS (3 4) FOR EXTENSION 
PICKING UP CONFIG 3:NP/:0(11) WITH WORD WHAT 
TRYING WRD ONLY ARC 
TRYING CAT QDET ARC 
CAT QDET TRANS #3 FHOM 7:NP/:O(ll) TO 5:NPiORD:3(16) 
TRYING PUSH DATE/ ARC 
TRYING TST ARC 
TRYING JUMP NPIONLY ARC 
PICKING UP CONFIG 4:S/NP:3(11) 
STARTING AT 3: 
MONITORING [ MODAL ] 
MONITORING [ V ] 
TRYING POP ARC 
POP TRANS #4 FROM 4:S/NP:3(11) 
EXECUTING PATH (1 2 4) 
BEGINNING AT TRANS 1, CONFrG 1 
DOING JUMP ARC WITH WHAT FROM l:S/ TO 2:S/Q 
DOING CAT ARC WITH WHAT FROM 2:S/Q TO 4:S/NP 
DOING POP ARC FROM 4:S/NP 
TEST FAILED 
SELECTED CONFIGS (5) FOR EXTENSION 
PICKING UP CONFIG 5:NP/ORD:3(16) 
STARTING AT 3: 
PJONITORING [ QUANT/ 1 
SETTING UP CONFIG 6:QUANT/:3(16) 
MONITORING [ INTEGER 1 
TRYING JUMP NP/QUANT ARC 
JUMP TRANS #5 FROM 5:NP/ORD:3(16) TO 7:NP/QUANT:3(21) 
SELECTED CONFIGS (7) FOR EXTENSION 
PICKING UP CONFIG ~:NP/QUANT:~(~~) 
TRYING JUMP NP/DET ARC 
JUMP TRANS U6 FROM 7:N~/QUAblT:3(21) TO 8:NP/DET:3(26) 
SELECTED CONFIGS (8) FOR EXTENSION 
PICKING UP CONFIG 8:NP/DET:3(26) 
STARTING AT 3: 
{MONITORING [ NPR/ NPR/ NPR NPH N ADJ N V ADV I) 
SETTING UP CONFIG g:NPR/:3(26) 
TRYING JUMP NP/HEAD ARC 
JUMP TRANS #7 FROM 8:NP/DET:3(26) TO 10:NP/MEAD:3(29) 
SELECTED CONFIGS (10) FOR EXTENSION 
PICKING UP CONFIG 10:NP/HEAD:3(2?) 
STARTING AT 3: 
IMONITORZNG c R/ PP/ RINIL PREP WHOSE WHO WHICH THAT WHOM 11 
SETTING UP CONFIG 11 : R/ : 3 ( 29) 
{PROPOSING IIWHOSE~~ lf~~~w ll~~~~~n 11~~~~11 tl~~~~n~ 
SETTING UP CONFIG 12:PP/:3(29) 
MONITORING PREP 1 
SETTING UP CONFIG 13:R/NIL:3(2Q) 
MONITORING [ THEaE ] 
PROPOSING llTHERE'l 
TRYING POP ARC 
POP TRANS #8 FROM 10:NP/HEAD:3(29) 
EXECUTING PATH (3 5 6 7 8) 
BEGINNING AT TRANS 3, CONFIG 3 
DOING CAT ARC WITH WHQ FROM 3:NP/ TO 5:NP/OHD 
DOING JUMP ARC FROM S:NP/ORD TO 7:NP/QUANT 
DOING JUMP ARC FROM 7:NP/QUANT TO 8:NPfDET 
DOING JUMP ARC FROM 8:NP/DET TO 10:NP/HEAD 
TEST FAILED 
STARTING AN ISLAND 
"FEEf1 TRYING CAT N ARC FROM NP/DET TO NP/DET 
CAT N TRANS #9 FROM 14:NP/DET:19(1) TO 15:NP/DET:23(4) 
ENDING AT 19: 
MONITORING [ NPR/ ] 
MONITORING [ ADJ ] 
MONITORING [ N ] 
NOTICING "REGISTRATIONv 
NOTICING "REGISTRATION" 
{There are two instances of the word llragistrationn in 
the word lattice, hence two notices are created.) 
MONITORING [ V 1 
JVMP TRANS #I0 FROM 16:NP/QUANT:19(1) TO 14:NP/DET:19(1) 
ENDING AT 19: 
t N TOR NG WANT/ ] 
JU~  IRAN^ 1 FROM 17:NP/ORD: 19( 1) TO 16:RP/QUANT: 19( 1) 
ENDING AT 19: 
{MONITORING [ ORD QDET ONLY 1) 
PROPOSING "ONLY" 
JUMP TRANS R12 FROM 18:NP/ART:19(1) TO 17:NP/OHD:l?(l) 
ENDING AT 19: 
{MONITORING [ ART QUANT POSS WHOSE 1) 
PROPOSING "WHOSE" 
JUMP TRANS 113 FROM 10:NP/ONLY:1?(1) TO 18:NP/ART:lo(l) 
ENDING AT 19: 
MONITORING [ ONLY 1 
PROPOSING I'ONLY" 
JUMP TRANS #I4 FROM 20:NP/:19(1) TO 1Q:NY/ONLY:19(1) 
tlFEE1' TRYING CAT N ARC FROM NP/DET TO NP/I!EAD 
CAT N TRANS #I5 FROI-i 14:NP/DBT: 1Q( 1) TO 21 :NP/E!EtZD:2?(6) 
SELECTED CONFIGS (21) POH EXTENSION 
PICKING UP CONFIC 21:NP/I1EAD:23(6) 
TRYING POP ARC 
POP TRANS #16 FROM 21:NP/HEAD:2?(6) 
EXECUTING PATH (14 13 12 11 10 15 16) 
BEGINNING AT TRANS 14, CONFIG 20 
TEST FAILED 
PREDICTIONS: 
NOTICING (1 REGISTRATION 6 19 277 O), SCORE -5 
NOTICING (10 REGISTRATIOIJ 7 19 103 O), SCORE -5 
PROPOSING (ONLY WHOSE) ENDING AT 19 
PHOPOSING (WHOSE WHO WHICH THAT WHO14 THERE) STARTING AT 3 
{EIONITORING [ ADJ N V ORD QDET ART QUANT POSS 1 
ENDING AT 10, SCORE O 
MONITORING [ WHOSE ONLY ] ENDING AT 19, SCORE 5 
MONITORING MODAL V INTEGER NPH N ADJ V.ADV P3EP ] 
STARTING AT 3, SCORE 0 
MONITORING [ WHOSE WHO WHICH THAT WHOM TWEHE ] 
STARTING AT 3, SCORE 5) 
FINISHED THEORY 1 WITH SYN SCORE 0 
{This processing took 12.5 seconds.) 
{Now wz will process the second theory.) 
SPARSER PROCESSING THEORY 2: 
0 WHAT 3 6 REGISTRATION 19 23 
STARTING AN ISLAND 
STARTING AT LEFT END OF SENTENCE 
SELECTED CONFIGS (1) FOR EXTENSION 
{Upon picking up this configuration to extend it, 
SPARSER finds thd transitions which were created durin~ 
the processing of the word I1whatfr by tha previous 
theory . It tttracesll them all, that is, it does not 
recreate them but simply puts the transition numbers on 
a list which will form part of the syntactic infornation 
associated with the current theory. The tracing process 
also involves the creation of monitors (and notices, 
uherr applicable) for constituants along the path. 
These monit~rs and notices must be remade, since the 
previous monitors will activate only th6 previous 
t heary . 
Due to the recursive nature of the tracin~ process, 
the transitions a not necessarily followad in t.he same 
order that they were originally created, nor are the 
monitors made in exactly the same order. 
Notice that the many arcs which we tried but 
which did not result in the creation of transitions in 
the previous theory are not retried here.} 
PICEING UP CONFIG 1:S/:0(1) WITH WORD WHAT 
TRAGING JUMP S/Q TRANS 1 FROM I:s/:o(I) TO 2:~/~:0(6) 
NONITORING [ NP/ 1 
TRACING CAT QWORD TRANS 2 USING "WHATu FROM 2:S/Q:0(6) 
TO 4:S/NP:3(11) 
STARTING AT 3: 
MONITORING [ MOBAL 1 
MONITORING [ V 1 
TRACING POP TRANS 4 FROM 4:S/NP:3(11) 
TRACING CAT QDET TRANS 3 USING "WHQ1I FROM 3:NP/:0(11) 
TO 5:MP/ORD:3(16) 
STARTING AT 3: 
MONITORING [ QUANT/ ] 
SETTING UP CONFIG 6:QUANT/:3(16) 
[This does not mean that confiauration 6 was just 
created. Since it already existed in the map, having 
been craated during the p~ocessinc of the previous 
theory, the configuration number is merely put on the 
list of configurations in the current theory.) 
MONITORING [ INTEGER 1 
TRACING JUMP NP/QUANT TRANG 5 FROM 5:~p/ORD:3(16) TO 
7:NP/QUANT:3(21) 
TRACING JUMP NP/DET TRANS 6 FROM 7:NP/QUANT:3(21) TO 
8:NPIDET:3(26) 
STARTING AT 3: 
{MONITORING [ NPR/ NPR/ NPH NPR N ADJ N V ADV 1) 
SETTING UP CONFIG 9:NPR/:3(26) 
TRACING JUMP NP/HEAD TRANS 1 FROM 8:NP/DET:3(26) TO 
10:NP/HEAD:3(29) 
STARTING AT 3: 
{!¶OtJITORING [ R/ PP/ R/NIL,PREP WHOSE WHO WHICH 
THAT WHOM PREP THERE 1) 
SETTING UP CONFIG 11:R/:3(29) 
{NOTICING 'WHOSE" "WHOw ) 
SETTING UP CONFIG 12:PP/:3(29) 
SETTING UP CONFIG 13:R/NIL:3(29) 
{No proposals were made here because proposals are not 
theory dependent; that is, the word proposals which 
were made during the processing of the previous theory 
resulted in some words baing placed in the word lattice 
whLch were noticed here. Remaking the proposals would 
not lead to the discovery of any new information.) 
TRACfNG POP TRANS 8 FROM 10:NP/HEAD:3(29) 
{The processinp of thz island for llregistrationw is 
idBntical to that in the last example, so the remainder 
of tha trace will be omitted. Thz total processing took 
12.2 seconds.] 
{Let us now process the event which adds the word "theu 
to ths thleory just processed. This will result in the 
creation of a conutituent event.) 
SYNTA PROCESSING EVENT FOR THEORY#2 
WIT tr NEW WQRD (4 THE 6) 
TO GET NEW THEORYf3: 
0 WHAT 3 4 THE 6 REGISTRATION 19 23 
"THEJ' TRYING (CAT ART --) FRO14 STATE NP/ONLY TO CONFIG 25 
CAT ART TRANS 825 FROM 32:NP/ONLY:4(?) TO 25:NP/ART:6(6) 
ENDING AT 4: 
MONITORING [ ONLY ] 
PROPOSING "ONLY~~ 
JUMP TRANS C26 FROM 33:NP/:4(1) TO 32:NP/QNLY:4(3) 
EXECUTING PATH (26 25 20 19 18 17 15 16) 
BEGINNING AT TRANS 26, CONFIG 33 
**** 
MADE #I FROM 4 TO 23: 
NP DET ART THE 
ADJ NP N REGISTRATION 
NU SG 
N FEE 
FEATS NU SG 
*+** 
NOTIFYING THEORY 3 ABOUT CONSTITUENT #l 
 his constituent cannot ba used immediately by this 
theory because it contains a word ("feevf) which is not 
in the thaory. Therefore a noticd is sdnt to Control 
which may be turned into an event at some later time. 
Nothing further is done with this constituznt at this 
time, i.e., no transitions using it are created. It is, 
howevsr, placed in the WFST for later use.) 
EXECUTING PATH (26 25 20 19 18 23 24) 
BEGINNING AT TRANS 23, CONFIG 22 
(?he creation of transition #26 complctas another path.) 
+**+ 
MADE 92 FROM 4 TO 19: 
NP DET ART THE 
N REGISTRATION 
FEATS NU SG 
**** 
SYN WEIGHT + SEM WT = 10 + 0 = 10 
{Tljis constituent is completely consistent with the 
current theory, that is, it is composed only of word 
matches already in the theory, and there are no nonitors 
in the WFST for it, so it is proce~sed bottom up as we 
have swp before. ) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #27 FROM 34:FOR/FOR:4(1) TO 35:T0/:19(7) 
ENDING AT 4: 
k¶ONITORING [ FOR ) 
PROPOSING "FORu 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #28 FROM 36:PP/pREP:4(1) TO 
37:PP/NP:19(7) 
ENDING AT 4: 
MONITORING [ PREP 1 
NP/ WAS NEVER PUSHED FOR 
PUSH NPI TRANS #29 FROM 38:R/N1L:4(1) TO 39:S/NP:19(6) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #30 FROM 4O:R/WH:4(1) TO 39:S/NP:19(6) 
ENDING AT 4: 
MONITORING [ R/WHOSE 1 
MONITORING [ WHICH THAT WHO M~OH WHICH WHOM ] 
(NOTICING "WHOlt llWHOM1l "WHICHw llWHOM'l 1 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 831 FROM 41:S/DCL:4(1) TO 39:S/NP:19(6) 
JUMP TRANS #32 FROM 42:S/:4(1) TO 41:S/DCL:4(1) 
ENDING AT 4: 
MONITORING [ PP/ 1 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 833 FROM 43:S/NOwSUBJ:4(1) TO 
44:VP/V:19(6) 
JUMP TRANS #34 FROM 45:S/AUX:4(1) TO 43:S/NO-SUBJ:4(1) 
ENDING AT 4: 
{MONITORING [ MODAL NEG V 1) 
{NOTICING "IS" IIARE" l'PAYW) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 135 FROM 46:S/Q:4(1) TO 39:S/NP:19(7) 
ENDING AT 4: 
MONITORING [ QADV ] 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 136 FROM 47:VP/HEAD:4(1) TO 
48:VP/NP:19(7) 
ENDING AT 4: 
MONITORING [ PARTICLE ] 
MONITORING [ V ] 
NOTICING "PAYn 
JUMP TRANS 837 FROM 49:VP/V:4(1) TO 47:VP/HEAD:4(1) 
ENDING AT 4: 
{MONITORING [ NP/ NP/ V ADV V ]] 
(NOTICING "IS1' "ARE" "PAY" l'IS" "AREt1 "PAY") 
JUMP TRANS 138 FROM 45:S/AUX:4(1) TO 49:VP/V:4(1) 
JUMP TRANS #39 F~~OM 43:S/NO-SUBJ:4(1) TO 49:VP/V:4(1) 
JUMP TRANS #40 FROM 43:S/NO=SUBJ:4(1) TO 49:VP/V:4(1) 
JUMP TRANS 141 FROM 50:S/THERE:4(1) TO 49:VP/V:4(1) 
ENDING AT 4: 
MONITORING [ THERE .J 
PROPOSING "THEREw 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS 842 FROM 51:VP/NP:4(1) TO 52:VP/VP:19(6) 
ENDING AT 4: 
MONITORING [ NP/ ] 
JUMP TRANS #43 FROM 47:VP/HEAD:4(1) TO 51:VP/NP:4(1) 
NP/ WAS NEVER PUSHED FOR 
PUSH NP/ TRANS #44 FROM 49:VP/V:4(1) TO 44:VP/V:1?(7) 
SELECTED CONFIGS (35 37 39 44 48) FOR EXTENSION 
PICKING UP CONFIG 35:TO/:19(7) 
STARTING AT 19: 
MONITORING [ NEG 1 
MONITORING [ TO 3 
PROPOSING If TOtf 
ALL ARCS TRIED AT THIS CONFIG 
PICKING UP CONFIG 37:PP/NP:19(7) 
TRYING POP ARC 
POP TRANS #45 FROM 37:PP/NP:19(7) 
PICKING UP CONFIG 39:S/NP:19(7) 
STARTING AT 19: 
MONITORING [ MODAL 1 
MONITORING [ V 1 
TRYING POP ARC 
POP TRANS #46 FROM 39:S/NP:19(7) 
EXECUTING PATH (32 31 46) 
BEGINNING AT TRANS 32, CONFIG 42 
**** 
MADE $3 FROM 4 TO 19: 
S NPU 
NP DET ART THE 
N REGISTRATION 
FEATS NU SG 
WITH FEATURES (NPU) 
+*** 
SYN WEIGHT + SEM WT = 10 + 0 = 10 
PICKING UP CONFIG 44:VP/V:19(7) 
STARTING AT 19: 
{MONITORING NP/ N-QDET ADJ INTEGER ARE QUANT 
PRO NPR POSS V V ADV TEST(N0T (CAT V)) I} 
SETTING UP CONFIG 20:NP/:19(7) 
NOTICING "FEE" 
ALL ARCS TRIED AT THIS CONFIG 
PICKING UP CONFIG 48:VP/NP:19(?) 
STARTING AT 19: 
{MONITORING [ COMPL/ TO/ COMPL/ NP/ FOR THAT TO FOR THAT 
N QDET AD3 INTEGER ARE QUANT PRO NPR POSS 
V PARTICLE 1) 
SETTING UP CONFIG 53:COMPL/:19(7) 
SETTING UP CONFIG 35:T0/:19(7) 
SETTING UP CONFIG 53:COMPL/:19(7) 
Page 75 
SETTING UP MNFIG 20:NP/:19(7) 
NOTICING "FEE" 
TRYING JUMP VP/VP ARC 
JUMP TRANS #47 FROM 48:VP/NP:19(7) TO 52:VP/VP:19(9) 
SELECTED CONFIGS (52) FOR EXTENSION 
PICKING UP CONFIG 52:~~/~P:19(9) 
STARTING AT 19: 
MONITORING [ PP/ 1 
SETTING UP CONFIG 30 : PP/ : 19 ( 9) 
MONITORJNG [ PREP 1 
MONITORING [ PREP ] 
TRYING JUMP S/VP ARC 
JUMP TRANS 648 FROM 52:VP/VP:lQ(g) TO 54:S/VP:19(12) 
SELECTED CONFIGS (54) FOR EXTENSION 
PICKING UP CONFIG 54t S/VP: 19 ( 12) 
TRYING JUMP $IS AHC 
JU!lP TRAMS #49 FROll 54:S/VP:lQ(12) TO 55:S/S:la(lll) 
Sl2LECTED CONFIGS ( 55 ) FOR KXTl<i!SION 
PICKfNG UP CONFIC 55;S/S:19(14) 
TRYING POP ARC 
POP TRANS #50 FROM 55:S/3:10(14) 
PREDICTIONS: 
NOTICING (4 FEE 19 23 155 O), SCORE 5 
NOTICING (19 WHO 3 4 -180 O), SCORE 10 
NOTICING (21 IS 3 4 -39 O), SCORE 10 
NOTICING (23 ARE 3 4 -128 O), SCORE 10 
NOTICING (25 PAY 3 4 -146 O), SCORE 5 
PROPOSING (TO) STARTING AT 19 
PROPOSING (ONLY FOR WHOM WHICH THERE) EIJDI~~K AT 4 
PROPOSING (V IIODAIa) FHOi1 3 TO 4 
PROPOSING (MODAL PREP) STARTING AT 3 
PROPOSING (PREP MODAL NEG QADV) ENDING AT 4 
[The lengthy summary of monitor,a set by this event is 
omit tad. ) 
CREATING THEORY 3: 
0 WHAT 3 4 THE 6 REGISTRATION 19 23 
WITH SYN SCORE 15 
{This took 30.8 seconds.) 
{Now we will process the constituent event for the 
theory just created. Because of the constituent for 
"the registrationm there art3 now monitors in the WFST 
for a noun phrase beginning at position 4, so ths 
appropriate transitions are made.) 
SYNTAX PROCESSING EVENT FOR THEORYiI3 WITH CONSTITUENT #1 
TO GET NEW THEORYf4 
0 WHAT 3 4 THE 6 REGISTRATION 19 FEE 23 
{Processing begins exactly where it left off when the 
constituent was made -- thz constituent is semantically 
evaluated with rwspact to this theory so that the 
constituent weight may be altered. In this case, 
howevar, Semantics has been turned off, so them is no 
increment in the score.) 
SYN WEIGHT + SEM WT = 15 + 0 = 15 
NP/ WAS PUSHED FOR AT CONPIG 34 
PUSH NP/ TRANS #51 FHOll 34:FOR/FOH:I+(l) TO 96:T0/:27('11) 
{Similar transitions are set up for all 9 other 
confipurationu where an NP/ was used in the previous 
theory. The monitors set by these path8 are copied from 
the previous theory, 80 there is no indication h+rz of a 
new monitor beinp created.) 
SELECTED CONFIGS (58 57) FOR EXTENSION 
PICKING UP CONFIG 5R:S/NP:23(10) 
TRYING POP ARC 
POP TRANS 961 FRO11 58:S/NP:23(10) 
EXECUTING PATH (32 55 61) 
BEGINNING AT TRANS 55, CONFIO 41 
DOING PUSH ARC WITH #1 FROM 41:S/DCL TO 58:S/NP 
DOING POP ARC FROt1 58:S/NP 
**** 
MADE #4 FROM 4 TO 23: 
S NPU 
NP DET ART THE 
ADJ NP N REGISTRATION 
NU SC 
N FEE 
FEATS NU SG 
WITH FEATURES (NPU) 
**** 
SYN WEIGHT + SEM WT = 15 4 0 = 15 
PICKING UP CONFIG 57:PP/NP:23(11) 
TRYING POP ARC 
POP TRANS #62 FROM 57:PP/NP:23(11) 
PRQDICTIONS : 
NOTICING (19 WHO 3 4 -180 Q), SCORE 10 
NOTICING (21 IS 3 4 -39 O), SCORE 10 
NOTICING (23 ARE 3 4 -128 O), SCORE 10 
NQTICING (25 PAY 3 4 -146 O), SCORE 5 
PROPOSING (V MODAL) FROM 3 TO 4 
PROPOSING (MODAL PREP) STARTING AT 3 
PROPOSI~~G (PREP MODAL NEG QADV) ENDING AT 4 
{Again, the monitor list is omitted for considerations 
of space. ) 
CREATING THEORY 4: 
0 WHAT 3 4 THE 6 REGISTRATION 19 FEE 23 
WITH SYN SCORE 15 
{This avant took only 9.7 seconds.) 
The procrssinp of the final event, that which adds the word 
"isw to the theory Just created, will not bz shown. 
ThesB examples have shown that SPARSER is a u~oful tool in 
th* automatic reco~nition df speech. The t iminp measurements 
indicate that considerable procesain~ iu done when the parser is 
forced to work in bottom ~p mode, especially with a large 
Eranmar Of course there is some implementaion ovdrhead involvdd 
in doins the timin~s themselves. If the paruinc al~orithm were 
to be carefully recoded in assembly langua~e a speed up of at 
least a factqr of 20 (and perhaps much nore) could be achieved. 
Another way to cut down the time-con sum in^ processing mipht be to 
at-tempt to obtain gar6 semantic ~uidance. For sample, if the 
semantic hypothesis asuociated with a theory indicates that a 
particular noun is likely to be used in a noun phrase modifier 
(ern& lttomorrowtt), than SPARSER should be able to take advanta~r 
of this information by scorinp the PUSH NP/ transition from a 
confi~uration for atata PP/ (i.e. to pet something like 'Iby 
tomarroww) hi~har than those PUSH NP/ transitions for other 
syntactic slots. In fact, th* others may not need' to be 
constructed at all. The Erammar could also be further tuned to 
eliminate soma spurious predictions and reduce the time spent 
following erroneous paths. 
Section 7 
Conclusions and Further Research 
*/.I STHENGTHS AND WEAKNESSES OF SPARSER 
One of the weak points of the current system is the fact 
that some context Information is not uued until a path is 
complete, result in^ in the creation of false paths and 
predictions which should not have been nade. This is partly 
miti~ated by the fact that this avoids a too  rea at dependence on 
left context and allows the creation of partial paths which may 
be followed if an earlier word is changed. 
It is important, however, to rininize the number of 
predi'ctions which are made and to ~ake the predictions as 
accurate as possible. In this r~~ard, it is unfoqtunatc that the 
currqt system makes predictions on the left of an island solely 
on the bavis of the first word in the island and a makes 
predictions 6n the right end from confi~urations which, if 
context sensitive tests had besn done alone the path, would never 
nave been created. 
One way to help tighten the predictions would be to take 
each context free path threu~h an island and walk it in a special 
mode after the island has been processed but before predictions 
are communicated to the cohtrol component, This mode would set 
and check registers, assuming that any tests which require 
unknown left context are true. Only if the path did not fail 
under this mode of operation would the pr~dictions at aither end 
pf it be made. If a really efficient way of handlinp unknown 
left context and of storinp this informafion ware developed, it 
could be uued in place of the context free pass in the first 
place, thus a1 lminatinp a 11 inconsistent paths. 
The problem with st or in^ all possible contexts is that they 
must be ~ecomputed each time a new atep is added to the path. 
Nia ie relatively easy if th* next step is taken to the ripht of 
an 2xistin~ path, since ATN s are mora auitdd to left to ripht 
processing, but it becomes extremely complex when a transition is 
added to the left end of a path (or set of paths) or when a 
transition Joins two sets of paths togethzr. To be absolutely 
sure that no contexts haw been misued, all the paths would have 
to be walked and their contexts reprocessed and copied in whole 
or in part (since the new step may be wronR, the old context rnuvt 
be preserved.) Of eoursa this is not the only approach which 
could be .used -- a merninp technique like that of Earley's 
algorithm might be feasible, if the structure of the crammar were 
also chanped to make it less left to ri~ht oriented. 
One ~reat ytren~th of the system is its ability to store and 
merge information in such a way that it does not hava to be 
redone when tha context is changed. For example, once an arc has 
bean tried with a particular word match, a transition will be 
created if the arc may be taken and the arc will ba removed from 
further consideration if it may not be taken. Then, if the 
configuration should ever be reached with the same word match 
apai'n (perhaps in a later theory) not only will any relevant 
transitions be reco~nized without havinp to po throuyh the work 
of r*-creatin~ them, but also ne arc8 whic_h had prev,Lou3+ly failed, 
Y~ ec hk ~&~kd* 
Another feature of SPARSER. in the fact that it \~i4~ ~!trsiyned 
and implemented with many unsolved problar~ :jnd unsviiiluble bats 
in mPnd, and therefore many wl~oleall have been lef't on which to 
lfhooku further developments. For exnmle, a1 t h~urh pro::~l!ic 
verification of constituents is not yet available, the scorinr 
rntrchani:~~ for ~wnst ituenty is structured in such a way that it 
would be easy to include the results of verification by prcsodics 
(or any other component). Th* oripinsl implenen'tation of SPAHSEH 
used a depth first search but Qas implemented in such a way that 
the chanre to dodified breadth first was quite simple. This 
foresight has paid off in a flexible systen rrhich has shown that 
it can be readily experimented with in o~der to explore many 
still unsolved problems ccrncoerning the nature and use of 
syntactic information in understandinp 
A tremendous amount of information in speech is conveyed by 
proscdic features: stress, intonation duration, loudness, 
pauses, pitch. For example, if John mumbles to Bill, ??The 
mailman left something for you," Bill may reply aithar IrWhat?" 
with much energy and a sharply risih'p intanation or "What?I1 with 
a flat o'f. falling intonation. In th~ first case John is vzry 
likely to shout 111 said, Ths mailman left something for you 
interpreting lvWhat?lv to mean "What did you say?" whereas in the 
second case he is likely to say something like *A package from 
youp rnothw," interpreting ltWhat?vv to mean "What is it?" To 
i~nora prosudics iu to ipnore a sourca of information which has 
bssn shown repeatedly to be an extremely important factor in 
human underatandinp. 
Consider the following examples of sentences and sentence 
fhpmehts which illustrate some of the ways prosodie8 are used: 
1. I ~trppad on the man with black yhoes. (Who wau wearing 
the ~h~as?) 
2a. The new gnu knew news. 
2b. Ths Rnu knew new Rnus. 
3 I m ~oing to move on Thursday. (stress on "move1n 
implies moving to a new house; stress on fronw imp1 ies 
traveling to a new place.) 
4a. Can you swim to Daddy? 
4b. Can you swim too, Daddy? 
5a. . .. two-fifty for . . . 
5b . . . two-fifty-four . , . 
Prosodic vmification could help a lot in reJect ing 
semantically correct, syntactically consistent phrases which are 
nonethsless wrong. If the constituent "speech understandingtf 
were identified and relied upon, it might be very difficult to 
produce a correct analysis of the utterancs: wBd~a~~d of 
peculiarities in his speech, understanding Joa is not easymN 
Besides indicatinp syntactic boundaries and/or providinp 
intonation contours for certain constituents, prosodic features 
can be used to mark emphasis, introduce new topics, cpnvey 
information about the speaker s internal mental and t:notion:il 
state (e.p. whether hl is teasinp or ~erioua), nnd probably 
more. It is particularly interestin~ to note that some well 
known phenomena ~uch us l'pranouns are almost never atrt*suedU :in3 
"ip discourse wh~n new topic word 1:: gention~bd it is alcost 
always stressedm hhve very naturtil explainat ions in ll~ht of what 
we know about acoustic processing. Stressed war-da are ~anerdlly 
easier to identify because there is less acoustic aztbicuity, but 
unstressed words may differ creatdy f rorn their ideal 
pronunciation and hence are harder to reliabl y identify 
Pronouns refer to antecedents which are presu~ably known to the 
listener, so he can anticipate them or at leaut verify them 
easily, hence they need not have mod acoustic characteristics. 
A new topic may not have been anticipated, so the listener will 
have to depend heavily on identify in^ the word from acoustic 
information alone and the speaker can provide this extra reliable 
information by stressine the word. 
Unfortunately, not a przat deal is known about either the 
acoustic correlat~s of prosodic features or the ways in which 
they are used. Many- of the rules which have bezn developed thus 
far are speaker dependent and are snfficient for convey in^ 
information but are not necessary. This makes them difficult to 
use in thz analysis mode. Althou~h a good start has been made in 
exploring pr-osodies (see, for example, Lea [52, 131 and Eates and 
Wolf [8]), much more work remains to be done before prosodic 
Page 83 
information can be reliably used by speech understandinc systems. 
SPARSER could use prosodic information in several ways. 
Verification of constituents would be a prest help, but local 
proaicid information could be used ev*n earlier in the parsing 
p~oceas. For example, if maJor constituent boundaries could be 
accurately determined, then inatead of both POPing a constituent 
and continuing it in parallel, as is done now, one alternative 
could be chosen inatdad of the other on the badis of prosodic 
informal ion. If, as is more likely, yome major boundariey could 
be reliably detected, then it would be easy to revise SPARSER to 
begin procsssinp at such places even #within an island at states 
which can begin constituents. This would a~ain reduce the number 
of partial paths created when pars in^ an island. 
7.3 EXTENSIONS AND FURTHER RESEARCH 
One of the obvious extension8 to a basic speech 
underatandinp . . system is to relax the restrictions on the input to 
the system. Syntactically, this can mean removing the 
requirement that the initial utterance be prammatical. Since 
people frequently speak unprammatically in informal discourse, 
this is a natural step to want to take. 
In order to extend SPARSER to handle such input, sevaral 
approaches are possible. Certain types of errors may be called 
errors of style (and may not be called errors at all by some 
people) such as the use of tlain'tlt and the occurrence of a 
prepositjon at thd and of a sentence. These resularit ies r?:ly 
simply be declared ~ramrnatical by ~0difyin.r the pr&+nnur to accept 
them. Hany speech errors have hern shown to fQcllc-\: rt~\:l~~:* 
piitterns and hence gay ht? :~vn:tble to t!lis :ippl*~;~ck. 
t her PC~V~!OII 1 spec- i 1"ic t tbst s \~?:ic!? :jp;-kS:+!- p .! 
the arcs at' the r, f 1 t ?r*~hikit i+t~:~-l~- 
net is n t 0 check for nu~ber a~~trtwven t bet wet'11 s1:5.-~vt :iz:q 
verb or between determiner itnd noun (c.~. *I is SPY Y=~~*V 
severe restrictions on this rule."). In this case, rather t!?ap 
renovinp the tests from the yramar it w~uld ~CJ rcre suita?le tc 
& 
modify  the^ SP th;it if t 1 t -arc 1 still hr L~.<C*Y, 
thourh  it h a :nvch reeuce~ w~~ir!~t or \:it h ;in icbicat icn in sere 
re~ister that an error has occurred. Cne way to in2lzxent this 
would be to have all tests return a nu~bttr as their value 
indicating how well they succeeded on some sca& fr~a "perf"ect1y" 
to ltnot at all". 
Not all arc tests are of this reIaxablc! nature, however, 
since certain types of errors are so rare, if they occur at all, 
that they may be Judged ~nacceptati~. Examples of such tests are 
the case checks for pronouns (2.~. *"I pavz it to he1'] and the 
requirement that a verb modifyinp. a noun must be in either the 
present or past participle form (2.g. Itthe sincinc brookn vs. 
*"ths sinp brook1'). 
These methods would not allow all tvpes of qran~atical 
errors to be handled (in particular it irnores the problen of 
constituent ordering errors such as ItThrow !lama from the train a 
kissw), but uould handle many of the most common syntactic 
errors. 
& experiment 
hoepinr in mind that SPARSER is not intended to be a mo'dd 
~f human svntactic analysis, it is nonetheless reasonable to ask 
whether there are anv ~imilarities which may be seen. The 
followinp experiment is suppest ed with the hypothesis tkit it 
will indicate that people do considerable processing at the end 
of svntactic constituenks in a way similar to-some repister 
set tin^ and testing actions and szmantic (or othw) verification 
The experiment is this: a subject is seated in front of a 
switch which hi is asked td press whenever ha is surd that he is 
hear in^ an anomalous sentence, He is then presented with a 
pulaber of recorded utterances, some of which are incorrect, e.~. 
{'The cat and dog which live3 next door are friendly. 
I sawma red big barn on f he farm. 
1 hypothesize that the subject will indicate the presence of an 
ewer at a point shortry after the end of the constituent in 
uhich the error occurred more often thand shortly after the 
earliest possible place where the error coul be detected. 
In conclusion, it is obvibds t~gf there is much work yet to 
be done in the problem.of speech undzrgtanding, but it is hoped 
Dha( the system presented be- has no.t, only advadczd our current 

Page 87 
APPENDIX I 
MINIGRAMMAR 
This appendlx contains a listing (slightly edited for 
clarity) of theqrammar called MINIGRAMMAR which was discussed in 
Section Three (irlustrated in Fi~ure 3.3) and which was used in 
Seatdon Six. 
(NP/ 
(CAT ART (T T) 
5 
(SETR#ART (BUILDQ ((ART *)))) 
(TO' NP/ART) ) 
(JUMP NP/ART (T T) 
4)) 
(N'P/ADJ 
(CAT N (T T) 
5 
(SETR #'*I 
(SETR NU (GETFEATURE NUMBER)) 
(TO NP/N)) 
(CAT N (T T) 
2 
(ADDL ADJS (BUILDQ tADJ (NP (N *) 
( N.U 
(GETF;EATURE NUMBER))) 
(TO NPIADJ))) 
( NPIART 
(CAT QUINT (T, T) 
4 
(SETR QUANT (BUILDQ ((QUANT *))I) 
(TO NPIQUANT) 1 
(JUMP NPIQUANT (T T) 
5)) 
(NP/QUANT 
(CAT ADJ (T T) 
4 
(ADDR ADJS (BUILDQ (@ (ADJ) 
(*I 
FEATURES)) 
(TQ NP/QUANT)) 
(JUMP NP/ADJ (T T) 
'4 1) 
(NP/N 
(PUSH PP/ ((PPSTART) 
7.' TI 
4 
(ADDL NMODS *) 
(TO NP/N)) 
(POP (BUILDQ (Q (NP) 
+ + + ((N +)I 
((NU +)I 
+ 
ART QUANT ADJS N NU NMODS) 
IT (DETAGREE)) 
5) 
(PP/ 
(CAT PREP (T T) 
5 
(SETR PREP *) 
(.TO PP/PREP) ) ) 
(PP/PREP 
(PUSH NP/ ( (NPSTART) 
T T) - 
5 
(SETR NP *) 
(TO PP/NP))) 
-C 
(PP/NP 
(POP (BUILDQ (PP (PREP +) 
+) 
PREP NP) 
(T TI 
5)) 
APPENDIX I1 
Vocabulary and Syntax CIaaaes 
This appendix lista the 351 words which wwa fn the 
dictionary of the BBN speech understanding ayatem wtkn the 
examples in Chapter Six ware run (July 19n). (A 569, word 
dictionary and one with 1000 entries arc now available. After 
the listing of the wards in the dictionary, Obey ara.broken into 
syntactic classes, with the number of words in. each alas3 
indicated beside the class name. Finally, the ~yntactic features 
are given together with a list of the word8 which carry each 
feature. Features may be of the form FEATURE, (FEATURE), or 
(FEATURE VALUE). 
This is not a listing of tha dictionary as dt appears to the 
systerrt, but rather a derived crosv reference which indicates the 
various parts of speech and ~yntactic features for each word 
Tha words: 
(A ABOUT ABOVE ACL ACOUSTICAL ACOUSTICS ADDITIONAL AFFORD AFTER 
A1 AIR AIRPLANE ALL ALREADY ALSO AM AMHERST AMOUNT AN AND 
ANTICIPATE ANY ANYONE ANYWHERE APRJL ARE ARPA ARRANGE ASA ASK 
&SSOOIATION ASSUME ASSUMPTION AT ATTEND AUGUST AVAILABLE BATES 3E 
BECAUSE BEEN BEFORE BEGINNING BEING BIG BILL BpNNfE BOSTON BDTH 
BREAKDOWN BUDGET BUS BY CALIFORNIA CAN CANCEL CAR CARMEGIE CENT 
CHANGE CITY COLARUSSO COMPUTATIONAL CONFERENCE CONTXWUE COSELL 
COST COSTING COSTS COUNTRY CRAIG CURRENT DATE DAY-E IAY DECEMBER 
DENNIS DID DIVISION DO DOES DOLLAR DONE DUE@TO DURING EACH EIGHT 
EIGHTEEN EIGHTEENTH EIGHTH EIOHTY EITHER ELEVEN ELEVENTH END 
ENGLAND E~OUGH ESTIMATE-N ESTIMATE-V EVERY EVERYONE EXPECT 
EXPENSE EXPENSIYE FALL FARE FEBRUARY FEE FIFTEEN FIFTEENTH FIFTH 
FIFTY FIGURE FINAL FIRST FISCAL FIVE FOR FORTY FOUR FOVRTEEN 
FOURTEENTH FOURTH GET GETS GETTING GIVE GIVEN GIVES GIVItJG GO 
GOES CO3'NG GONE GOT GOTTEN CROUP WAD HALF HALVES HAS HAVE HAVING 
HE HER HIM HIS HPw HOWMANY HOWMUCH HUNDRED 1 IF IFIP IJCAI IN 
INTERNATIONAL IS IT JANUARY JERRY JOHN JULY JUNE KNOW L.A. 
LAST LATE LEFT LINDA L,INGUISTICS LIST LONDON LONG LOSeANGELES LYN 
LYNN. MADE MAKE .MAKES MAKMOUL MA~INC MANY MARCH MASSACtIUSETTS MAY 
ME HEETING MEMBER MISCELLANEOUS MONEY MONTH MORE MOST MUCH MY 
NEED NEW@YORK NEXT NINE NINETEEN NINETEENTH NINETY NINTH NQ NOT 
NOTE NOVEMBER NOW. OCTOBER OF ON ONE ONLY OH OTIIER OUT OVERHEAT, 
PAJARRDPDUNES PARTICIPAIJT PAUL PAY PENNSYLVANIA PEOPLE PER PEHSON 
PHONOLOGY PITTSBURGH PLACE PLEASE PLUS PRINT PROJECT-N PROJECT-V 
PURPOSE QUARTER REGISTRATION REMAIN REST REVISE RICH RICHARD 
ROUNDeTRhIP SANTAeBARBARA SCHEDULE SECOND SEND SENDING SENDS SENT 
SEPTEMBER SkVEN SEVENTEEN SEVENTEENTlI SEVENTH SEVENTY SHE SINCE 
SETE SIX SIXTEEN SIXTEENTH SIXTH SIXTY SO SOCIETY SOME SOMEONE 
SPEECH SPEND SPENDING SPENDS SPENT SPRING ST.LOUIS START STATUS 
STOCKHOLH SUMMER SUPPOSE SUPPOSED SUPPOSITION SUN SWEDEN TAKE 
TAKEN TAKES TAKING TEN TENTH THAN THANKQYOU THAT THE THEIH THEM 
THERE THESE THEY THIRD THIRTEEN THIRTEENTH THIRTIETH THIRTY THIS 
THOSE THOUSAND THREE TIME TO TOO TOOK TOTAL TRAVEL TRIP TWELFTH 
TWELVE TWENTIETH TWENTY TWO UNANTICIPATED UNDUDCETED UNSPEilT 
UNTAXEN NUS VARIOUS VISIT WANT WAS WASHINGTON WE WEEK WENT WERE 
WHAT WHEN WHERE WHICH iJHO WHOM WHOSE WILL WINTER WISCONSIN WITH 
WITHIN WORKSHOP YEAR YES YOU) 
The syntactic categories: 
(ADJ 23 (ACOUSTICAL ADDITIONAL AVAILABLE BIG COI-IPUTATIONAL 
CURRENT EACII ENOUGH EXPZNSIVE FINAL FISCAL INTERNATIONAL 
LATE LEFT LONG MANY MISCELLANEOUS OTHER UNANTICIPATE~ 
UNBUDGETED UNSPENT UNTAKEN VARIOUS)) 
(ADV 18 (ALREADY ALSO ANYWHERE EITHER ENOUGH WOW [,ATE LONG NORE 
MOST MUCH NOW ONLY PLEASE SO THERE TOO YES)) 
(ART 8 (A AN NO THAT THE THESE THIS THOSE)) 
(CONJ 8 (AND BECAUSE BOTH IF OR PLUS SINCE SO)) 
(IN.TEQER 27 (EIGHT EIGHTEEN EIGHTY ELEVEN FIFTEEN FIFTY FIVE 
FOR*TY FOUR FOURTEEN NINE NINETEEN NINETY ONE SEVEN 
SEVENTEEN SEVENTY SIX SIXTEEN SIXTY TEN THIRTEEN THIRTY 
THREE TWELVE TWENTY TWO)) 
(MODAL 5 (CAN DID DO DOES WILL)) 
(N 70 (ACOUSTICS AIR AIRPLANE AMOUNT ASSOCIATION ASSUMPTION 
BEGINN-ING BREAKDOWN BUDGET BUS CAR CENT CHANGE CITY 
CONFERENCE COST COUNTRY DATE DAY DIVISION END ESTIMATE-N 
EXPENSE FALL FARE FEE FIGURE GROUP HALF HALVES LINGUISTICS 
LIST MEETING MEMBER MONEY MONTH MUCH NEED NOTE OVERHEAD 
PARTICIPANT PEOPLE PERSON PHONOLOGY PLACE PROJECT-N 
PURPOSE QUARTER REGISTRATION REST ROUNDQTRIP SCHEDULE SITE 
SOCIETY SOME SPEECH SPRING STATUS SUMMER SUPPOSITION 
TH4NKeYOU TIME TOTAL TRAVEL TRIP VISIT WEEK WINTER 
WORKSHOP YEAR)) 
(NEG 1 (NOT)) 
(NPR 53 (ACL A1 AMHERST APRIL ARPA SA AUGUST BATES BILL BONNIE 
BOSTON CALIFORNIA CARNEGIE C:ILARUSSO COSELL CRAIG DECEMBER 
DENNIS ENGLAND FEBRUARY IFIP IJCAI JANUARY JERRY JOHN JULY 
JUNE L.A. LINDA LONDON LOS@ NGELES LYN LYNN MAKHOUL MARCH 
MASSACHUSETTS MAY NEWCYORK NOVEMBER OCTOBER PAJARROeDUNE-S. 
PENNSYLVANIA PITTSBURGH RICH RICHARP SANTA~BARBAHA 
SEPTEMBER ST.LOU1S STOCKHOLM SUR SWEDEN WASHINGTON 
WISCONSIN)) 
(ORD 23 (EIGHTEENTH EIGHTH ELEVENTH FIFTEENTH FIFTH FIHST 
F~URTEBNTH FOURTH LAST NEXT NINETEENTH NINTH SECOND 
SgVENTEENT11 SEVENT.H SIXTEENTH SIXTH TENTH' THIRD THIIjTEENTH 
Tflf RTIETH TWELFTH TWENTIETH) ) 
(PARTICLE 3 (IN ON OUT)) 
(POSS 5 (H~R HIS MY THEIR WHOSE)) 
(PRECONJ 2 (BOTH EITHER)) 
(PREP 18 (ABOUT ABOVE AFTER AT BEFORE BY DUEeTO DURING FOR IN OF 
ON OUT PER SINCE TO WITH WITHIN)) 
(PRO 23 (ANYONE EVERYONE HE HER HIM I IT ME ONE SHE SOMEONE THAT 
THEM THESE THEY THIS THOSE US WE WHAT WHO WHOM YOU)) 
(QADV 2 (WHEN WHERE)) 
(QDET 5 (HOWNANY HOWMUCH WHAT WHICH WHOSE)) 
(QUANT 14 (ALL ANY BOTH EACH EITHER ENOUGH EVERY HOWMANY HOWMUCH 
MANY MORE MUCH OTHER SOME)) 
(SPECIAL 8 (DOLLAR HUNDRED K NO THAN THANKdlOU THOUSAND YES) ) 
(.V 85 (AFFORD AM ANTICIPATE ARE ARRANGE ASK ASSUME ATTEND BE BEEN 
BEGINNING BEING BUDGET CAN CANCEL CHANGE CONTINUE COST 
COSTING COSTS DID DO DOES DONE END ESTIMATE-V EXPECT 
FIGURE GET GETS GETTING GIVE GIVEN GIVES GIVING GO GOES 
GOING GONE GOT GOTTEN HAD HAS HAVE HAVING IS KNOW LAST 
LEFT LIST MADE MAKE MAKES MAKING NEED NOTE PAY PRINT 
PROJECT-V REMAIN REVISE SCHEDULE SEND SENDING SENDS SENT 
SPEND SPENDING SPENDS SPENT START SUPPOSE TAKE TAKEN TAKES 
TAKING TOOK TOTAL TRAVEL VISIT WANT WAS WENT WERE WILL)) 
Ths syntactic features: 
(INGCOMP (CANCEL CONTINUE GIVE START START)) 
(INTRANS (CONTINUE GO GO GO START)) 
(PASSIVE (CANCEL CONTINUE FIGURE GET GIVE MAKE MAKE SEND SEND 
START START TAKE) ) 
(QCOMP (CANCEL CONTINUE FIGURE GIVE SEND SEND TAKE)) 
(THATCOMP (END FTGURE)) 
(TRANS (CANCEL CONTINUE END FIGURE GET GIVE MAKE MAKE SEND SEND 
START START TAKE)) 
((ANAPHORIC) (WHICH)) 
((DETERMINED) (ANYONE I IT ME THAT THESE THIS THOSE US WE WHO 
WHOM YOU)) 
((INDOBJ FOR) (MAKE HAKE)) 
((NUMBER PL) (HALVES PEOPLE THEM THESE THEY THOSE US WE)) 
((NUMBER SO) (ANYONE ME HER HIM I IT ME ONE SHE SOMEONE THAT THIS 
WHO WHOM WHOM)) 
((NUMBER SG/PL) (WHAT Wi3AT WHICH WHO YOU)) 
((PARTICLEOF (LEAVE PUT ADD)) (IN)) 
((PARTICLEOF (ADD CONTINUE)) (ON)) 
((PARTICLEOF (LEAVE PRINT SEND MAKE CANCEL FIGURE FIND GO)) 
(OUT) ) 
((PASTPART) (BEEN COST DONE GIVm GONE GOTTEN HAD LEFT MADE SENT 
SPENT TAKEN ) ) 
((PNCODE 13SG) (WAS)) 
((PNCODE 1SG) (AM)) 
((PNCODE 3SG) (COST COST COST COST COSTS DOES DOES GETS GIVES 
GOES HAS IS MAKES SENDS SPENDS TAKES)) 
((PNCODE ~IJ (CAN CAN DID)) 
((PNCO'DE X13SG) (ARE WERE)) 
((PNCODE X3SG) (COST DO WILL)) 
((PRESPART) (BEGINNING BEING COSTING GETTING GIVING GOING HAVING 
MAKING SENDING SPENDING TAKING) ) 
((ROLE OBJ) (HER HIM ME THEM US WHOM WHOM)) 
((ROLE SUBJ) (HE I SHE WE)) 
((ROLE SUBJ/OBJ) (WHAT WHICH WHO WHO)) 
((TNS FUTURE) (WILL WILL)) 
( (TNS PAST) (COST DID DID GOT HAP MADE SENT SPENT TOOK WAS WENT 
WERE) ) 
((TNS PRESENT) (AM ARE CAN CAN LJST COST COST COST COST COSTS DO 
DOES DOES GETS GIVES GOES HAS IS MAKES SENDS SPENDS TAKES 
WILL) ) 
((UNTENSED) (BE WILL)))) ) 

BIBLIOGRAPHY 
C1] Baker, James K. (1975b) 
St-ochastic Modeling as a Means of Automatic Speech 
Rewgai tAw 
PhD thesis, Speech and Computer Sciepca, 
Carne~ia-Mallon Univ., AprW 1975 

r2] Barnztt, Jeffrey (1973) 
A Vocal Data Management System 
IEEE Trans. on Audio and Elactroacoustics, 
AU-21:3 (June 1973) p. 185-188 

[31 Bates, M. (1974) 
The UY~ of Syntax in a Speech under st end in^ System 
in Erman IEEE Symposlum on 'Speech Rclcogni t ion, 
p. 226-233 

[4] Pates, M. (1975) 
Syntactic Analysis in a Speech Understandine System 
Ph.D. thesis, Harvard Univeruity,l975 
also BBN Report No. 3116, Bolt Beranek and 
Nswman Ind., Cambridge, Mass. Aucust 1975 

[5] Bates, N..and 901f, J. (1975) 
Prosodies 
technical report in BBN Speech Understanding 
system QPR noi 2, BBN Report no 3080 (A1 
Report no. 30), Bolt- Bargnek and Newman Inc, 
Cambridge, Mass. (MEW, 1975) 

161 Bruce, Bertram (1976) 
Pragnoat ics. in Speaeh Understanding 
Procaedinps of the IJCAI, 1976 

C71 Erman, Lee (ad.) (1934) 
IEEE Symposium on Speech hzcsgnition 
IEEE, Inc. (1974) 

18) Forgria, James W. (1974b) 
Speech Understanding Systems;-Semiannual Technical Summary 
Lincoln Laboratory, Lexington, Mass., 31 May 1974 

C91 Fu; K.C. (1976) 
SYNTACTIC PATTERN RECOGNITION 
Springer-VerJan* N.Y.0 

[lo] Hyde, S.R. (4368) 
Automatic Speech Recognition: Literatbre Surveye and 
Discussion 
Report No. 45, Post Office Research Dept. 
Dollis Hill, London, 1968 

[I11 Klovstad, John W. (1975) 
personal communication 

12. W. Lea, M. Medress, and T. Skinner. 1972. Prosodic aids to speech recognition: I. Basic algorithms and stress studies, Univac report no. PX7940, Oct. 1972.

13. W. Lea, M. Medress, and T. Skinner. 1973. Prosodic Aids to speech recognition: II. Syntactic segmentation and stressed syllable location, Univac report no. PX10232, 1973.

14. Richard J. Lipton and Lawrence Snyder. 1974. On the optimal parsing of speech, research report 37, Oct. 1974, Yale.

15. Bonnie Nash-Webber. 1974. Semantic Support for a speech understanding system, IEEE trans. on ASSP, 1975.

16. R. B. Neely. 1973. On the use of syntax and semantics in a speech understanding system. PhD. thesis, Stanford, 1973.

17. A. Newell, et. al., 1973. Speech understanding systems: final report of a study group, North-Holland, Amsterdam.

18. William H. Paxton, 1974. A best-first parser, in Erman, IEEE Symp. on Speech Recognition.

19. William H. Paxton and Ann E. Robinson, 1975. System integration and control in a speech understanding system. Tech. note 111, Artifical intelligence center, Stanford, 1975.

20. D. R. Reddy, L. Erman, and R. B. Neely. 1970. The CMU speech recognition project, IEEE System Sciences and Cybernetics Conf., 1970.

21. D. R. Reddy, et. al., 1973. The HEARSAY speech understanding system: an example of the recognition process. IJCAI, Stanford, 1973.

22. D. R. Reddy, ed. 1975. Speech Recognition. IEEE Symposium, Academic Press, NY.

Rovnzr, P., Nash-Wabber, B., and Woods, W. (1974) 
Control Concepts in a Speech Undervtanding System 
IEEE Trans. on ASSP,, 32:1 (Fab: 1975), p. 136-140 

Aovn~r, P., at .al. (1974) 
Where the Words are: 
Lexical Rrtriaval*in a Spaech Undervtanding Sy~tem 
in Erman, IEEE Symposium on Speech Recopnition, 
p. 160-164 

Rustin, Randall (ad.) (1973) 
NATURAL LANGUAGE PROCESSING 
Alcorithmica Pray, NwY. 1973 

Schwartz, Rw and Makhoul, Jm (1074) 
Where the Phoneme8 are: 
Dealinn with Arnbirulty in Acouutic-Phonetic Heco~nition 
IEEE Trans. on ASSP, 231 (Feb. 11)75), p. 50-53 

Shirai, K. and Fujiuawa, H. (197?) 
An Al~orithm for Spoken Sentence Recognition and its 
Application to the Speech Input-Output Sy~tern 
IEEE Trans. on Systemu, Man, and Cybrenetics, 
Sept 1924, p. 475-479 

Walker, Donald Ew (1972) 
Speech Understanding Research 
SRI Annual Technical Report,'~ct 1071 - 0ct 1972 

Walker-, Donald E. (1973aJ 
Speech Understanding Throuph Syntactic 
and Semantic Analysis 
Prac. 3rd IJCAI, Stanford, Aup. 1973, p. 208-215 

Walker, Donald E. (1973b) 
Automated Lan~uape Proce~sinfl 
in Cuadra (ed.), Annual Review of Information Science and 
Trchnolohy, vol 8 

[31] Wolf, Jared (1976) 
Spaach Recognition and Understanding 
in Fu, Syntactic Pattern Recognition 

[32] Woods, W.A-1 (-193Q) 
Transition Network Grammars for Natural Language Analysis 
CACM, 13:10 (Oct 1970) p. 591'9606 

1331 Woods, WeAw (1974) 
Motivatiod and Overview of BBN'SPEECHLIS.: 
An Experimental Prototype far Speach Undarstandihg Rassarch 
IEEE Trans. on ASSP, 23:1 (Fzb. 1975), p. 2-10 

[34] Woods, W.A. (1975) 
Syntax, Semantics, and Speach 
in qeddy, Spaech Recognition: Invited Papers 
Presented at the IEEE Symposium, Acadsmic Press 

[35] Woods, W.A. et. el. (1974) 
Natural Languajia Communication with Computeru, 
Final Report, Volume I, Spadch Undsrstanding Research 
at BBN 
BBN Report 2976, Bolt Beranek and Newman Inc., 
Cambridpa, Mauu., Dec., 1974 

[36] Woods, WoA,, et .ale (1?72) 
The Lunar Sciencds Natural Lan~uare Information 
System: Final Report 
Report No. 2378, Bolt Reranek and Newman Inc., 
Cambridge, Mass. 
