Deriving the English Verb 1
Linguistics 323
As in the case of the noun, we adopt the hypothesis that all grammatical features originate here as a bundle of features together with the verb stem. How the grammatical features end up in the feature bundle is a lexico-syntactic problem which is covered in L322.
The English verb is more complex than the English noun. There is one uninflected form and five inflected forms. The uninflected form is often called the infinitive, and it is identical with the non-third person singular form of present tense form of the person. There is one exception: be. be is an infinitive form only, and it forms the stem for the progressive and the non-progressive participles: be+ing; be+en. All the inflected forms are irregular: am, are, is, was, were.
Let us start with the regular verb cook. First, consider the following present tense paradigm:
singular | plural | |
cook | cook | first |
cook | cook | second |
cooks | cook | third |
Note that all six of these forms bear the feature [-Past] (for present tense). It is nearly universally accepted that cooks is bimorphemic: book+s. Rather than adopting an analysis the considers the suffix '+s' as bearing the feature [-Past], let us assume that it is the morpheme COOK that does. That is, in all six forms, cook = COOK+[-Past]. '+s' contains the features [-Personal] (for third person) and [-Pl]. This gives a neat analysis. The present tense is associated with the stem and not with a suffix. On the other hand, the past tense is always associated with a suffix (-ed in regular forms) or it isn't realized as a suffix in the class of strong verbs to be discussed below.
singular | plural | |
cooked | cooked | first |
cooked | cooked | second |
cooked | cooked | third |
The third person here considered to be the feature [-Pers(onal)].
[+Personal] refers to the first and second person only. The first
person either refers only to the speaker, or to the speaker plus
some other person in the plural. This form is the marked form.
It should be written as [+First] (first person) and [-First] (second
person). Unfortunately, a large part of my writing marks the second
person as [+2nd], indicated that it is the most marked form. Actually,
the first person is the most marked form. Throughout anything
that I've written on the web, [+First] = [-2nd]. Let us revised
the present tense paradigm as follows:
[-Pl] | [+Pl] | |
cook | cook | [+First, +Pers] |
cook | cook | [-First, +Pers] |
cook+s | cook | [-Pers] |
Note that [-Pers] does not expand to [First].
Given the above analysis, [-Past] is by default, [-Suf]:
[-Past] --> [-Past, -Affix].
There are no exceptions to the above default rule. The third person singular is expanded by the following default rule:
[-Pers, -Pl] --> [-Pers, -Pl, +Suf].
With the exception of the verb to be, there are no exception to the 3PS default rule. [{BE},-3rd, -Pl] --> "is" in the lexicon, which means it occurs before the default rule applies. Three verbs have allomorphic variants int he context of the third person singular: has, does, says. These forms have weak allomorphs, meaning that the final consonant is deleted or there is a weakening of the stem vowel:
{HAVE}: ha+s,
{DO}: do+es (here the high back tense vowel weakens to a mid central vowel (the carot),
{SAY}: say+s (here the mid front tense vowels weakens to a mid front lax vowel.
None of these rules is phonetically predictable; they must occur in the lexicon as determined by lexical properties. In the case of SAY, English spelling is inconsistent (what else is new?). In the past tense, the weak allomorph occurs and the spelling reflects it: said (with a mid front lax vowel). In other words, if English spelling were consistent, the present 3PS would be spelled " *sais".
Default rules.
Many languages have default rules of the following sort that assign grammatical features to a suffix:
Let alpha (a) represent all grammatical features that may occur in a given language, LEXICAL represent any lexical stem, and X is a syntactic category:
{X LEXICAL, alpha} --> [X LEXICAL, alpha [XSTEM LEXICAL], [XSUF alpha]].
What this means is that a bundle containing a lexical entry is converted into a structure which contains a the lexical entry as a stem and a grammatical suffix (inflection). We can represent this in a tree structure configuration:
In English this more general default rule does not apply. Only certain grammatical categories are realized as suffixes. One of the is the past tense:
{V LEXICAL, [+Past]} --> [V [LEXICAL, [+Past]]: [Vstem LEXICAL] [Vsuffix [+Past]].
In tree structure form, this appears as:
Now, let's apply it to a specific verb. We will chose the regular verb typed. The following lexical entry for type includes enough information for its morphological and phonological representation:1
{TYPE} |
lexical form |
type |
orthographic form |
/tajp/ |
phonemic form |
V |
category |
|
The basic (and underspecified) lexical form next includes the feature [+Past] (derived in the syntax):
{TYPE} |
lexical form |
type |
orthographic form |
/tajp/ |
phonemic form |
V |
category |
[+Past] |
grammatical feature |
The above default rule for English is now applied. The rule states that the grammatical feature [+Past] is realized as a suffix:
Often, such structure are appreviated as follows:
Sometimes, even this structure may appear abbreviated; however, we will use the above abbreviation: no information is missing in the tree. To spell out the suffix [+Past], we need a dafualt rule:1
[+Past. +Suf] --> "ed" / /îd/.
The penultimate tree structure is better here to illustrate the application of the above rule. It states that [+Past] and [+Sif] are spelled out as "ed":
Techically, the orthographic form need not be given in the lexicon, except for irregular forms. There exists a set of phoneme-to-orthography rules. These apply after the phonological rules have applied. However, if there is an irregularity, it must be listed in the lexical entry. In the word "type", the grapheme "y" is not predictable. One would expect "i": "*tipe". But the phoneme /aj/ is othographically spelled out as :"y" rather than "i". This rule would have to be listed in the lexical entry of TYPE:
/aj/ --> "y" |
Irregular orthographic rule |
Another rule would add a mute "e" after the final consonant:
/tajp/ --> /t/ "aj" /p/ --> /t/ "aj" /p/ "e" --> "type".
The first expansion is determined by the irregular orthographic rule; the remainder are regular default rules of English orthography.
Rather than go into the formality of these spell-out rules, we will simply list the orthographic form in the lexicon commening on certain cases. The rules of English orthography combine the graphemes of a word into a linear string with no internal blanks:
typeed.
Another rule of English orthography deletes a mute "e" if it precedes another vowel:
typed.
"type" is determined in the lexical entry of TYPE. In English if "e" follows a consonant in a word that contains an inner vowel, the final vowel is mute as a default rule. If "e" is not mute, then it must be marked somehow. We don't go into how to do this here. Note that the condition that there is an inner vowel renders the final "e" in the following words as pronounced and not mute: be, he, the, she, me, ye (obsolete).
Some verbs take either "t" or "d" as the past
tense marker. These form a somewhat irregular class. This class
is typically marked by the deletion of a final obstruent or the
weakening of the stem vowel: make+[+Past] --> made; sleep+[+Past]
--> slept. This feature is marked in the lexical entry of the
verb. The feature occurs in the form of a rule:
/i:/--> /e/ / ____X+[+Past] | Irregular vowel mutation in context of the past tense. |
// --> /t/ / [+Past] | Irregular past tense rule |
Here, these rules must apply in a strict order; the upper rule applies first. The following shows the application of the lexical rules deriving the past tense form. Assume that SLEEP contains the above lexical non-default rules:
SLEEP, [+Past], /sli:p/ | Basic form plus grammatical category |
SLEEP, [+Past, +Suf, /sli:p/] | Default Past-Suffix rule |
[V [Vstem SLEEP, /sli:p/] + [V Suf +Past,+Suf]] | Suffix Splitting Rule |
[V [Vstem SLEEP, /slep/] + [V Suf +Past,+Suf]] | Irregular vowel mutation in context of the past tense. |
[V [Vstem SLEEP, /slep/] + [V Suf +Past,+Suf, /t/]] | Irregular past tense rule |
slep + t | Spell-out rule |
slept | word convergence rule |
Verbs whose past tense formis irregularly /d/ are derived in exactly the same way.
Footnotes
1. Because non-standard fonts cannot be used universally on the web, we use the following system to represent English vocalic phonemes:
i, e, u, o, a = lax vowels
i:, e:, u:, o: = tense vowels.
æ = low front vowel (technically a tense vowel). If this doesn't work, we will use /a:/ to represent the low tense vowel.