Ling 425/709

Note 13

The Acquisition of Auxiliaries

Auxiliary verbs play an important role in the grammars of some languages, but are absent in others. Auxiliaries only appear in the first, second and final position in sentences, even in free word order languages like Warlpiri (Steele 1981). In English, auxiliary verbs have a prominent role in question formation:

The study of auxiliary development provides information about the acquisition of morphology and syntax.

Kuczaj & Maratsos (1975)

Their study involved one subject, Kuczaj’s son Abe. They summarize Abe’s auxiliary use as:

period I: 2;5(21)-2;7(15), MLU 3.01-3.25—Abe only produced 4 sentences with either ‘will’ or ‘can’ from 3,058 recorded sentences

   period II: 2;9(16)-(19)—Abe used auxiliaries in declaratives, but not in questions.

The authors gave Abe several sentences with the auxiliaries ‘will’ and ‘can’ to imitate

Results (Table 9.11)

Bellugi (1967/1971)

Bellugi proposed several developmental stages for auxiliaries for Brown’s subjects:

period A: (I, 28 months) Children do not have Aux.

      Yes/no questions are marked by intonation, e.g., ‘sit chair?’

Wh-questions are restricted to limited uses of ‘what’ and ‘where’, e.g., ‘what’s that?’, ‘where cookie go?’

period B: (II, III, 35 months)

period C: (IV, 38 months) Aux appears and is widely used. Inversion occurs in yes/no questions

                  but not in wh-questions, e.g. ‘does lions walk?’, ‘will you help me?’

                  Wh-questions, e.g., ‘what he can ride in?’, ‘why kitty can’t stand up?’

periods D to F: (V and later) Gradual onset of tag questions

Two other errors deserve mention:

   1. Double tense marking, e.g. ‘oh, did I caught it?’

   2. Double Aux marking, e.g., ‘whose is that is?’, ‘what did you did?’ (Hurford 1975)

Ingram & Tyack (1979)

They collected 225 questions from 21 children between 2;0 and 3;11

Results (Table 9.12)

1. The children display a gradual increase in their use of auxiliaries.

2. The children use auxiliary inversion with wh-questions as well as with yes/no questions

3. The children tended to not invert uncontracted auxiliaries

Several investigators report the gradual use of Aux inversion with specific wh-words (Labov & Labov 1978; Kuczaj & Brannick 1979; Erreich 1984)

Maratsos & Kuczaj (1978) provide data on double tense marking and double Aux marking

Hurford (1975) and Maratsos & Kuczaj (1978) report that double tense marking is usually restricted to irregular past forms (e.g. ‘broke’) instead of regular pasts (e.g. ‘missed’) or overgeneralized past forms (e.g. ‘breaked’).

Maratsos & Kuczaj claim that double Aux marking is extremely rare ‘Occasional errors such as ‘is this is the powder’ may not require any explanation at all, given their extremely low frequency ...’

The Explanation of Auxiliary Acquisition

Performance factors

Bellugi (1967), Brown, Cazden & Bellugi (1969) and Brown (1968) claim that children have the rules for Subject-Aux inversion and Wh-Movement, but cannot apply both rule to the same sentence. They predict:

   1. Children will use inversion in yes/no questions

   2. Children will not use inversion in wh-questions

Problems:

   1. What prevents children from using inversion, but not Wh-Movement in wh-questions?

      e.g. ‘can John do what?’

   2. What leads children to acquire both rules at Stage III?

   3. Why do children restrict the use of inversion to specific wh-words?

   4. What prompts the children’s double tensing and double Aux marking errors?

Competence factors

Erreich, Valian & Winzemer (1980) assume that children already know the rules for inversion and Wh-Movement.

They claim that all movement rules involve two operations: copying and deletion, in that order. They predict:

   1. The production of double Aux marking

Problems:

   1. They do not explain why it takes so long for children to eliminate double marking errors

   2. Children would have to use indirect negative evidence to eliminate double marking errors

   3. They predict the production of double Wh-word marking, e.g. ‘What did you see what?’

   4. They predict widespread double marking errors

Maratsos & Kuczaj present an explanation of auxiliary acquisition that combines both competence and performance factors.

They assume that children acquire auxiliaries very gradually and restrict their use, at first, to specific contexts. Children also restrict their use of Subject-Aux inversion, at first, to one or two wh-words, and only gradually generalize the rule to other wh-words.

They claim that children acquire auxiliaries in yes/no questions independently of their use in declaratives, and only gradually merge the auxiliaries into a single syntactic category. Abe, for example, used ‘would’ and ‘could’ in declaratives for several months, but would substitute ‘will’ and ‘can’ respectively in questions.

They also claim that children do not invert semi-auxiliaries such as ‘better’ or contracted auxiliaries ‘wanna’ and ‘gonna’ in questions. Children would be expected to overgeneralize auxiliary inversion to these words if they were forming a general rule of Subject-Aux inversion.

Maratsos & Kuczaj account for double tense marking through performance. Double tense marking is usually restricted to questions with a form of ‘do’ and an irregular verb. They propose that children simply retrieve the incorrect form of the irregular verb. This error only occurs with ‘do’ since ‘do’ only marks tense.

Maratsos & Kuczaj recognize that their retrieval account cannot account for children’s double Aux marking errors. These errors are so rare, though, that another performance factor may be at work.

Problems:

   1. They do not offer an account of how children acquire the general rules for auxiliaries.

2. They do not provide independent evidence for retrieval errors (past tense overgeneralization?)

   3. They do not provide a convincing explanation why retrieval errors are limited to ‘do’

Pinker (1984) treats auxiliaries as complement-taking verbs following the Lexical Functional Grammar model.

Additional phrase-structure rule

   S —> (V_AUX =c +) NP_SUBJ VP’_V-COMP

   Only auxiliary verbs can appear in the inverted sentence position

Children use semantic bootstrapping to recognize auxiliaries–they encode tense, aspect and modality. Auxiliaries also commonly occur in first, second or final sentence position.

They start to use auxiliaries at about the same time as they begin producing complement-taking verbs.

Double-tensing results from a violation of the constraining equation. Also find constraint violations with other complement-taking verbs:

   Adam (Sample 21)  She gonna fell out

Double-tensing is more likely to occur with irregular complements, and is more likely to occur in inverted rather than in non-inverted contexts.

Predicts double auxiliaries will not occur; the phrase-structure rule does not produce them.

Pinker’s model predicts lexical constraints due to word-by-word acquisition of auxiliaries.

Assessment

   The model agrees with the Maratsos & Kuczaj claims for lexical constraints on auxiliary acquisition. Pinker also claims that constraint violations may be due to performance (‘transient “slip of the tongue” errors’). The model predicts that children learn an extra rule for auxiliary inversion, but also have to learn which words are auxiliaries. The model does not explain why children take so long to learn the inversion rule.

Optimality Theory (OT) Prince & Smolinsky (1993)

Assume

1. A universal set of constraints

2. Constraints can be violated

3. Grammars require the constraints to be ranked

4. The rankings select an optimal form from the set of candidates

The candidate sentences are generated from the same words to express a single meaning

Grimshaw (1997, Projection, heads & optimality. LI 28.373)

Constraints

OpSpec (Operator is Specifier—syntactic operators must be in the specifier position)

ObHead (Obligatory Heads—every projection has a head)

Stay (Constituents don’t move)

Tableau

Have different possible rankings across languages

OpSpec     ObHead    Stay                English [+wh-move; inversion]

ObHead    OpSpec     Stay                ?    [only move in main clause]

Stay          OpSpec     ObHead          Japanese; Korean? [no movement]

Stay          ObHead    OpSpec                “    ”

ObHead    Stay          OpSpec             “     ”

OpSpec     Stay          ObHead          ?

Acquisition requires:

1. maturation of the constraints

2. discovery of the constraint order in the input

Advantage

The model includes all possible question forms

Problems

Still lack a specific set of universal constraints

The model does not account for the time needed to order the constraints

The model predicts ‘what you will read’ is better than ‘you will read what?’

Cross-linguistic differences

French allows the main verb to invert with pronominal subjects

   Pronominal Subject       Full NP

   Sait-il?                           * Sait Jean

   know he                         know John

   Does he know?              Does John know?

Spanish only allows the main verb to invert, not auxiliaries

   Main Verb                                             Auxiliary

   Partió él?            Partió Juan?                * Ha él partido

   Did he leave?     Did John leave?          Has he left?

K’iche’ adds an optional particle to the initial position of yes/no questions

   A xe7k

   Q left-s/he

   Did s/he leave?

Inversion became restricted in English in change from Middle English to Modern ~ 1500 CE

   Before 1500: Speak they the truth?

Wh-movement according to Guasti

Guasti cites Rizzi (1996) Wh-Criterion (189):

a. A wh-operator must be in a specifier-head relation with a head carrying the wh-feature.

b. A head carrying the wh-feature must be in a specifier-head relation with a wh-operator.

This statement raises some immediate issues:

   1. What is a wh-operator?

   2. What is a wh-feature?

   3. Which heads carry the wh-feature?

Guasti claims the wh-feature is generated in T (190) and that if nothing moves to C, the wh-criterion would not be satisfied.

Problems:

   1. Why not just move the wh-phrase to Spec T?

   2. What allows in-situ wh-questions, e.g. ‘You did what?’

Cross-linguistic differences:

   1. All verbs move to C in Italian (I-to-C movement or subject-aux inversion)

   2. Only auxiliary verbs (or the pleonastic verb do) move to C in English

   3. Wh-operators do not move overtly in Chinese, Japanese and Korean (they move at LF)

Guast claims that children have an early knowledge of wh-movement although she does not provide developmental data (191).

She provides some cross-linguistic data for verb movement in questions with overt subjects (Table 6.1)

Guasti cites data from Stromswold (1990) claiming that 12 children learning English inverted auxiliaries in 93.4% of their questions, although one child only used inversion in 54% of the questions. Figure 6.1 shows that Adam only inverts ~35% of questions at 30 months.

Negative questions

Stromswald (1990) reports that children only use aux inversion in 55.6% of negative questions.

Guasti cites data from Guasti, Thornton & Wexler 1995 (Table 6.2)

Auxless questions

Guasti also reports data on the use of non-subject questions without auxiliaries (Table 6.3)

Explanation according to Guasti

Guasti proposes a root null auxiliary to account for lack of auxiliaries in English questions.

This proposal is required to avoid contradicting the truncation hypothesis since questions like

   Where cookie go?

appear to have a wh-operator in Spec C, but lack an auxiliary.

Guasti claims the null auxiliary explains a subject-object asymmetry in the use of auxiliary verbs

   1. Adam, Eve and Sarah use null auxiliaries in 2% of their subject questions,

            e.g. Who laughing?

   2. They use null auxiliaries in 16% of their non-subject questions,

            e.g. What dat train doing?

Problems:

   1. The null auxiliary explanation assumes the wh-phrase moves to Spec FocusP. If the structure

is truncated at TP in children’s grammar, the wh-phrase could move to Spec TP and the null auxiliary could appear in T and satisfy the identification principle

   2. The explanation does not account for negative questions

   3. There is no explanation for the discontinuity between child and adult grammars

4. No account is provided for the difference, if any, between children’s use of wh-phrases and auxiliaries in questions.

5. No account is provided for the developmental trend in the use of auxiliaries evident in the Ingram & Tyack (1979) data.

Structural Constraints on Movement—Barriers

   * Who_i did you see William and t_i

            |_____________________|

   * What_i did you see the man that wore t_i

            |__________________________|

Subjacency Principle

   Movement may not cross more than one bounding node

   Bounding nodes are IP, CP and NP

   Languages differ over whether IP or CP is a bounding node

de Villiers & Roeper (JCL 1995)

Tested wh-movement constraints by telling children 3;5-6 picture stories and asking questions, e.g.,

A little boy is playing outside and he falls out of a tree and hurts his arm. When he is having a bath, he discovers the bruise on his arm and says to his father, ‘I must have hurt myself when I fell this afternoon.’

Children show 95% correct responses at 3-4 years of age.

Children also have to learn many ideosyncratic constraints on movement (after 4;0):

Negation         Why_i did he not say t_i his aunt was coming *t_i

Factive Verbs Why_i did he forget t_i his aunt was coming *t_i

Quantifiers      Why_i did he always say t_i his aunt was coming *t_i

There are also differences between adjunct and argument questions:

Adjunct     When_i did you ask t_i how to help *t_i

                  Short distance             Long distance

                  I asked last night         I asked how to help you next weekend

Argument Who_i did you ask t_i how to help t_i

                  Short distance             Long distance (violates subjacency!)

                  I asked Katie               I asked how to help Roy

Karin Stromswold (1988) PRCLD 27.107-114

Made computer search of the transcripts from 12 children including Adam, Eve and Sarah

On average:

   First who subject question appeared .5 months before the first who object question

   First what object question appeared 2.5 months before the first what subject question

   First which object question appeared 8.8 months before the first which subject question

   The differences were all marginally significant. She did not find a significant correlation between the frequency of the adult and child question types.

She reported more significant differences between the first appearance of argument and adjunct questions. She averaged the appearances for both types across all the wh-words.

   Argument questions appear 7.1 months before the first adjunct questions

   Argument where questions appear 6.8 months before adjunct where questions, e.g.,

            Argument: Where did Roy put the book?

            Adjunct: Where did Mary see John?

De Villiers & Roeper 1995 JCL Adjunct questions in relative clauses

Subject Relative                           How_i did the man [who hurt his leg *t_i] get home t_i

Object Relative                             How_i did the man rescue the cat t_i [who broke her let *t_i]

Extraposed Subject Relative         How_i did the boy drink t_i [who sneezed *t_i]

8 subjects ~ 3;7

Subject Relative                           How did the boy [who sneezed 0%] drink the milk 58%

Object Relative                             How did the woman help the man 61% [who won the race 0%]

Extraposed Subject Relative         When did the woman sleep 47% [who painted the picture 10.5%]

1. No children responded to the relative pronoun who as a question.

2. Children know that the relative clause is a barrier for adjunct questions

Their findings contrast with earlier research by Otsu

He reported that children do worse with argument questions.

What_i is James painting t_i a picture of a boy with t_i

   25-35% of 3-4 year-old children responded that he painted a picture of a boy with a book.

References

De Villiers, Jill & Roeper, Thomas. 1995. Relative clauses are barriers to wh-Movement for

      young children. JCL 22, 2.389-404.

Grimshaw, Jane.1997. Projection, heads & optimality. Linguistic Inquiry 28.373-422.

Pinker, Steven. 1984. Language Learnability and Language Development. Cambridge, MA:

      Harvard.

Steele, Susan. 1981. An encyclopedia of AUX : a study in cross-linguistic equivalence.

      Cambridge, Ma.: MIT Press.