Syntax 1 Wiki: Week 4: Difference between revisions

From English Grammar
Jump to navigation Jump to search
 
(123 intermediate revisions by 2 users not shown)
Line 1: Line 1:
<font face="Arial, Helvetica, sans-serif">
<font face="Arial, Helvetica, sans-serif">
<font size="3">
<font size="3">
<!--
=== Valence Properties of Verbs ===


All verbs in English need to have a subject. But verbs differ from each in other in whether they need to have complements, how many complements they need, and what kinds of complements they are. All the elements which a verb (or a word of another part of speech) can occur with are called its <span style="color: blue>arguments</span>. The sum of all the arguments are called the word's <span style="color: blue>valence</span>.
== Looking under the hood ==


In feature structures, the arguments of the verb are described by three so-called <span style="color: blue>valence features</span>: SUBJ, SPR, and COMPS. We will explain SPR later and here concentrate on the other two. In school you learned that traditional grammar distinguishes between intransitive verbs like ''snore'' and transitive verbs like ''kiss''. These verbs can be used as follows:
<span style="color: blue>Exercise</span>


Intransitive verb:
Go to the [http://141.2.159.95:7001/wt/ Online Grammar] and do the following:


(1) Lilly snored.<br>
* Type "snores" into the ''Sentence'' textbox and press the "Parse" button or RETURN on your keyboard.
(2) *Lilly snored Fido.
* Click once on the top node of the tree.
* Inspect the result and try to understand as much as possible.
* <p style="margin-bottom:0mm;">Now click on the boxed "0" after AGR.</p>
* Inspect the result and try to understand as much as possible.
* <p style="margin-bottom:0mm;">Finally, click on the boxed number below the NP in the SPR list.</p>
* Again, inspect the result and try to understand as much as possible.
* <p style="margin-bottom:0mm;">Click on the new boxed "0" after the AGR that appears.</p>
* Now there are two boxed "0"s on the screen. What seems to be their function?


Transitive verb:
<span style="color: blue>Exercise</span>


(3) *Lilly kissed.<br>
Go to the [http://141.2.159.95:7001/wt/ Online Grammar] and do the following:
(4) Lilly kissed Fido.


(1)-(2) show that an intransitive verb can (in fact, must) have a subject, but is not allowed to have a direct object. (3)-(4) show that, in contrast, transitive verbs must have both a subject and a direct object.
* Parse the sentence Lilly dances".
* Open the S node and the V node. Don't open any numbered boxes for now.
* Compare the two feature structures.
* For every feature (= attribute), compare the values of mother and daughter.
* Now open the N node, leaving its numbered boxes alone again.
* Compare the three feature structures. What relationships do you detect? Do the self-test exercises below.


With the use of the valence features, we can capture the similarity and difference between intransitive and transitive verbs. All valence features take a list as value. To express that all verbs must have one and only one subject, we require their SUBJ list to be as follows:
<span style="color: blue>Self-test exercises</span>


SUBJ <NP>  
(1) <span style="color: blue>The part of speech features of the mother are also the part of speech features of</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
the verb.
</div>
</div><br>


The COMPS lists of intransitive and transitive verbs must look different, however:
(2) <span style="color: blue>The verb is looking for ______ complements.</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
no
</div>
</div><br>


Intransitive: COMPS < ><br>
(3) <span style="color: blue>The sentence is looking for ______ complement(s).</span>
Transitive: COMPS <NP>  
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
no
</div>
</div><br>


When we put the subject and complement information together for both types of verbs, we get the following result:
(4) <span style="color: blue>The verb is looking for ______ specifier(s).</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
one
</div>
</div><br>


Intransitive verbs:
(5) <span style="color: blue>If the verb requires a specifier, then it selects a specifier of part of speech ________.</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
NP
</div>
</div><br>


SUBJ <NP> <br>
(6) <span style="color: blue>The sentence is looking for ______ specifiers.</span>
COMPS < >
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
no
</div>
</div><br>


Transitive verbs:
(7) <span style="color: blue>Formulate the relationship between the valences of S and V, and the relationship between the NP daughter and the V daughter.</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
A. The verb is complement-empty and selects one NP-specifier.<br>
B. The NP acts as the verb's specifier.<br>
C. As the verb's specifier requirement has now been satisfied, its mother S has an empty SPR list (as well as an empty COMPS list).
</div>
</div><br>


SUBJ <NP><br>
== Head-Specifier Phrases ==
COMPS <NP >


Convince yourself that this is true by looking at these two verbs in the [http://141.2.159.95:7001/wt/ Online Grammar].
We can combine all the observations above into a Phrase Structure Rule. In short form, the rule can be formulated as follows:


Now we are going to do an exercise together:
(1) XP -> Specifier X


We will look at a number of verbs in the [http://141.2.159.95:7001/wt/ Online Grammar]. Their SUBJ lists will all have the same length but their COMPS lists may be of different lengths.  
We call this kind of phrase a '''head-specifier phrase'''.


<span style="color: red>Exercise</span>
Instead of formulating the full-fledged PS rule, we look directly at the local tree that the rule licenses:  


Find a systematic relationship between the semantic relation that the verb expresses and its valence (we continue to ignore the valence feature SPR for now). You find the semantic information as the value of the feature CONT.
[[File:H-SPR phrase.PNG|400px]]<br>
In words: A head-specifier phrase can be formed by combining <i style="color:blue;">a sign</i> [= the head daughter] with <i style="color:blue;">a second sign</i> [= the specifier daughter] whose syntactic category can be unified with the syntactic category on the word’s SPR list. The mother and its head daughter are both COMPS-saturated (= have empty COMPS lists).


To find the generalization, you might want to ask yourself the following questions:
Comparing the tree in the online grammar with the tree above (the rule), you will notice that they aren't completely identical.


* How many elements are on the SUBJ list?  
(2) <span style="color: blue>What does the rule contain that is missing from the tree?</span>
* How many elements are on the COMPS list?
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
The rule marks the righthand daughter as the '''head''' (= '''H''') and the lefthand daughter as a '''specifier''' (= '''SPR''').
</div>
</div><br>
 
(3) <span style="color: blue>What does the tree contain that is missing from the rule?</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
The tree contains the information that the mother node and the head have the same part of speech features.
</div>
</div><br>
 
== The Head Feature Principle ==


Let us deal with the second difference first. It will turn out that this will also take care of the first one.


==== Homework for Week 5 ====
Head-specifier phrases are not the only kind of phrase. We will soon see others. All phrases have one thing in common, however:


1. Work through the following sections of the Wiki:<br>
'''The Head Feature Principle (HFP)'''
The part of speech features of a phrase are also the part of speech features of its head daughter.


: [[Grammatical_Functions|Grammatical Functions]]<br>
Formally, this is expressed as follows:


-->
  '''The Head Feature Principle (formal)'''
[[File:HFP.PNG|300px]]
 
The formal version of the head feature principle is a '''constraint.''' Constraints are all of the form "If a feature structure F has a property A, then F is well formed only if F also has property B." The double arrow expresses the if-then.
 
Consequently, the constraint above says: if a feature structure F is a phrase, then F is well formed only if F's value for POS is also the value of POS of F's head daughter.


== A Systematic Way of Representing Information ==
With this, we are ready to return to (3) in the previous section. Remember that we wanted to take care of the fact that the head-specifier rule lacks the information of the identity of head features of mother and head daughter. This problem is now solved: since head-specifier phrases are phrases, they are well formed only if their POS features are identical to the POS features of their head daughter.


You have probably encountered the word ''database''. A database is a way of collecting data about objects (people, birds, planets, etc.) where the objects have similar properties (e.g. first name, last name, date of birth, etc.), but different values for these properties. Here are a few examples.
The difference between the online grammar and the head-specifier rule thus is explained as follows: the online grammar shows the sum of information contained in the head specifier rue and the head feature principle!


The page [https://qis.server.uni-frankfurt.de/qisserver/rds;jsessionid=5C3620A59C4CF824B88845E7DF634C45.waldmarie01?state=verpublish&status=init&vmfile=no&moduleCall=webInfo&publishConfFile=webInfoPerson&keep=y&publishSubDir=personal&personal.pid=10183 QIS-Webelhuth] lists some information about one of the professors of the University of Frankfurt. The page [https://qis.server.uni-frankfurt.de/qisserver/rds?moduleCall=webInfo&personal.pid=14417&publishConfFile=webInfoPerson&publishSubDir=personal&state=verpublish&status=init&topitem=members&vmfile=no QIS-Sailer] lists the corresponding information about a second professor. Let us, to simplify matters, extract some pieces of the information from both pages. Then we can represent the information as follows:
The effect of the head feature principle shown in the online grammar also takes care of the observation (2) in the previous section. You can always tell which daughter of a phrase is its head daughter: it is the daughter which shares the phrase's POS features.


[[File:employees.jpg]]
<span style="color: blue>Exercise</span>


We see a table of information enclosed by straight brackets. In the table there are two columns: the first column contains FEATURES of employees of the University of Frankfurt and the second one the ''values'' of those features for two employees, Prof. Webelhuth and Prof. Sailer. This description format is called a <span style="color: blue>feature structure</span> for obvious reasons: we describe objects in terms of the values they have for certain features.
Go to the <span class="newwin">[http://141.2.159.95:7001/wt/ Online Grammar]</span> and do the following:


Here is a second example. If you go to the online book catalog of Goethe University and type in the two names "Gert Webelhuth" and "Julika Griem", you will find a list of books authored by these two professors. If you click on a link describing one of these books, then the information about the book is also given to you in the form of a feature structure. Using the same format as above, you will find the following two descriptions:
* Parse the string "a cat".
* Click on the three category nodes.
* Interpret what you see.


[[File:books.jpg]]
== Head-Complement Phrases ==


Here is a third and last example. The following description of mobile phones can be found on the internet:
<span style="color: blue>Exercise</span>
<br>
<br>
[[File:iphones.png]]
<br>
<br>
Do you realize that the information above is in the form of a table? In the first column, we find FEATURES of phones and in each following column we find the values for the features of one individual model of phone. It would be easy to represent this information in the form of feature structures.


== Types ==
Go to the <span class="newwin">[http://141.2.159.95:7001/wt/ Online Grammar]</span> and do the following:


Look at the feature structures for employees, books, and mobile phones one more time. Since they describe different things, they also contain different FEATURES. The employees have an OFFICE, but the books and the mobile phones do not. The books have an AUTHOR and a PUBLISHER, but employees and mobile phones do not. In contrast, only the phones have the features BATTERY LIFE and CAMERA SIZE. The moral of the story is that <span style="color: blue>the inventory of FEATURES in a feature structure depends on the object that the feature structure is meant to describe.</span> This information somehow needs to be represented in the feature structure as well. To this end, we add a so-called ''type'' to feature structures in the top lefthand corner which signals <span style="color: blue>the type of the feature structure</span>, i.e. what the feature structure is about.
* Parse the string "likes lilly".
* Click on the three category nodes.
* Interpret what you see and answer the following questions.


The first employee feature structure and the first book feature structure above will then really look as follows:
(1) <span style="color: blue>Which daughter is the head daughter of the head-complement phrase?</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
the lefthand daughter
</div>
</div><br>


[[File:employee-Sailer.jpg]]
(2) <span style="color: blue>What is the valence relationship between the daughters of the head-complement phrase?</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
the head daughter selects the non-head daughter as a complement.
</div>
</div><br>


[[File:book-Webelhuth.jpg]]
(3) <span style="color: blue>How are the COMPS valences of a head-complement phrase and its head daughter related?</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
The head daughter selects a complement, which is realized as its sister in the tree.<br>
The head-complement phrase is COMPS-saturated, since there are no further complements to be realized.
</div>
</div><br>


In general, feature structures will then have the following form:
(4) <span style="color: blue>How are the SPR valences of a head-complement phrase and its head daughter related?</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
They are identical. The head-complement phrase selects the same specifiers as its head daughter.
</div>
</div><br>


[[File:generic-fs.jpg]]
<span style="color: blue>Exercise</span>


Based on the properties of head-complement phrases just discovered, draw a schematic non-local tree for head-complement phrases in the same format as was done for head-specifier phrases above.


Here is the solution:


<span style="color: blue>The Structure of Head-Complement Phrases</span>
<div class="toccolours mw-collapsible mw-collapsed" style="width:800px">
Check your answer
<div class="mw-collapsible-content">
[[File:H-C phrase 1.PNG|400px]]
</div>
</div><br>


<br>
<br>
<br>
<br>
Navigation:
<div align="center">
<div align="center">
   [[Syntax_1_Wiki |'''Main page''']]|[[Syntax_1_Wiki:_Week_1| '''Week 1''']] |[[Syntax_1_Wiki:_Week_2| '''Week 2''']]|[[Phrases| '''Week 3''']]| '''Week 4'''|[[Syntax_1_Wiki:_Week_5| '''Week 5''']]| Week 6| Week 7  
   [[Syntax_1_Wiki |'''Main page''']] [[Syntax_1_Wiki:_Week_1| '''Week 1''']] [[Syntax_1_Wiki:_Week_2|'''Week 2''']] [[Syntax_1_Wiki:_Week_3|'''Week 3''']] '''Week 4''' [[Syntax_1_Wiki:_Week_5|'''Week 5''']] [[Syntax_1_Wiki:_Week_6|'''Week 6''']] [[Syntax_1_Wiki:_Week_7|'''Week 7''']] [[Syntax_1_Wiki:_Week_8|'''Week 8''']]
</div>
</div>
<!-- |[[Phrases| '''Week 3''']]| '''Week 4'''|[[Syntax_1_Wiki:_Week_5| '''Week 5''']]|[[Syntax_1_Wiki:_Week_6| '''Week 6''']]| Week 7
-->

Latest revision as of 07:29, 12 July 2024

Looking under the hood

Exercise

Go to the Online Grammar and do the following:

  • Type "snores" into the Sentence textbox and press the "Parse" button or RETURN on your keyboard.
  • Click once on the top node of the tree.
  • Inspect the result and try to understand as much as possible.

  • Now click on the boxed "0" after AGR.

  • Inspect the result and try to understand as much as possible.

  • Finally, click on the boxed number below the NP in the SPR list.

  • Again, inspect the result and try to understand as much as possible.

  • Click on the new boxed "0" after the AGR that appears.

  • Now there are two boxed "0"s on the screen. What seems to be their function?

Exercise

Go to the Online Grammar and do the following:

  • Parse the sentence Lilly dances".
  • Open the S node and the V node. Don't open any numbered boxes for now.
  • Compare the two feature structures.
  • For every feature (= attribute), compare the values of mother and daughter.
  • Now open the N node, leaving its numbered boxes alone again.
  • Compare the three feature structures. What relationships do you detect? Do the self-test exercises below.

Self-test exercises

(1) The part of speech features of the mother are also the part of speech features of

Check your answer

the verb.


(2) The verb is looking for ______ complements.

Check your answer

no


(3) The sentence is looking for ______ complement(s).

Check your answer

no


(4) The verb is looking for ______ specifier(s).

Check your answer

one


(5) If the verb requires a specifier, then it selects a specifier of part of speech ________.

Check your answer

NP


(6) The sentence is looking for ______ specifiers.

Check your answer

no


(7) Formulate the relationship between the valences of S and V, and the relationship between the NP daughter and the V daughter.

Check your answer

A. The verb is complement-empty and selects one NP-specifier.
B. The NP acts as the verb's specifier.
C. As the verb's specifier requirement has now been satisfied, its mother S has an empty SPR list (as well as an empty COMPS list).


Head-Specifier Phrases

We can combine all the observations above into a Phrase Structure Rule. In short form, the rule can be formulated as follows:

(1) XP -> Specifier X

We call this kind of phrase a head-specifier phrase.

Instead of formulating the full-fledged PS rule, we look directly at the local tree that the rule licenses: 

H-SPR phrase.PNG

In words: A head-specifier phrase can be formed by combining a sign [= the head daughter] with a second sign [= the specifier daughter] whose syntactic category can be unified with the syntactic category on the word’s SPR list. The mother and its head daughter are both COMPS-saturated (= have empty COMPS lists).

Comparing the tree in the online grammar with the tree above (the rule), you will notice that they aren't completely identical.

(2) What does the rule contain that is missing from the tree?

Check your answer

The rule marks the righthand daughter as the head (= H) and the lefthand daughter as a specifier (= SPR).


(3) What does the tree contain that is missing from the rule?

Check your answer

The tree contains the information that the mother node and the head have the same part of speech features.


The Head Feature Principle

Let us deal with the second difference first. It will turn out that this will also take care of the first one.

Head-specifier phrases are not the only kind of phrase. We will soon see others. All phrases have one thing in common, however: 

The Head Feature Principle (HFP)
The part of speech features of a phrase are also the part of speech features of its head daughter.

Formally, this is expressed as follows: 

 The Head Feature Principle (formal) 
HFP.PNG

The formal version of the head feature principle is a constraint. Constraints are all of the form "If a feature structure F has a property A, then F is well formed only if F also has property B." The double arrow expresses the if-then.

Consequently, the constraint above says: if a feature structure F is a phrase, then F is well formed only if F's value for POS is also the value of POS of F's head daughter.

With this, we are ready to return to (3) in the previous section. Remember that we wanted to take care of the fact that the head-specifier rule lacks the information of the identity of head features of mother and head daughter. This problem is now solved: since head-specifier phrases are phrases, they are well formed only if their POS features are identical to the POS features of their head daughter.

The difference between the online grammar and the head-specifier rule thus is explained as follows: the online grammar shows the sum of information contained in the head specifier rue and the head feature principle!

The effect of the head feature principle shown in the online grammar also takes care of the observation (2) in the previous section. You can always tell which daughter of a phrase is its head daughter: it is the daughter which shares the phrase's POS features.

Exercise

Go to the Online Grammar and do the following:

  • Parse the string "a cat".
  • Click on the three category nodes.
  • Interpret what you see.

Head-Complement Phrases

Exercise

Go to the Online Grammar and do the following:

  • Parse the string "likes lilly".
  • Click on the three category nodes.
  • Interpret what you see and answer the following questions.

(1) Which daughter is the head daughter of the head-complement phrase?

Check your answer

the lefthand daughter


(2) What is the valence relationship between the daughters of the head-complement phrase?

Check your answer

the head daughter selects the non-head daughter as a complement.


(3) How are the COMPS valences of a head-complement phrase and its head daughter related?

Check your answer

The head daughter selects a complement, which is realized as its sister in the tree.
The head-complement phrase is COMPS-saturated, since there are no further complements to be realized.


(4) How are the SPR valences of a head-complement phrase and its head daughter related?

Check your answer

They are identical. The head-complement phrase selects the same specifiers as its head daughter.


Exercise

Based on the properties of head-complement phrases just discovered, draw a schematic non-local tree for head-complement phrases in the same format as was done for head-specifier phrases above.

Here is the solution:

The Structure of Head-Complement Phrases

Check your answer

H-C phrase 1.PNG





Retrieved from "https://www.english-linguistics.de/grammarparadise/wiki/index.php?title=Syntax_1_Wiki:_Week_4&oldid=5469"