The Ups and Downs of Child Language

Experimental Studies on Children's Knowledge of Entailment Relationships and Polarity Phenomena

Andrea Gualmini

• Publisher: Taylor & Francis
• Year: 2014
• eBook ISBN: 9781135875367
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

The new experimental evidence presented in The Ups and Downs of Child Language shows that it is possible to extend research on child language to children's semantic competence, adopting the same theoretical framework that has proven useful to the study of children's syntactic competence. Andrea Gualmini investigates the role of entailment relations for child language in a series of interconnected experiments assessing children's negation and their interpretation of words like or, every, and some. Comparing his study to other models of language acquisition and characterizing the observed differences between children and adults, Gualmini asserts that even in the domain of semantic competence there is no reason to assume that child language differs from adult language in ways that would exceed the boundary conditions imposed by Universal Grammar.

CHAPTER 1

Language and Acquisition

Much contemporary research on child language is guided by the poverty of the stimulus argument. This argument is for the most part based on two observations. First, young children possess a rich linguistic competence. Second, the input to which young children are exposed undermines their linguistic competence in many respects. To reconcile the mismatch between the rich linguistic competence shown by children and the impoverished input to which they are exposed, it is inferred that innate linguistic knowledge assists the child in the process of language acquisition. The role of innate linguistic principles is to restrict the child’s hypothesis space, so that the child can make the most. out of the impoverished input (see Chomsky, 1980a).

In recent years, the basic premises of the poverty of the stimulus argument have been challenged. Some researchers have denied that young children possess adult linguistic competence (see Tomasello, 2000). Other researchers have denied that the input to which children are exposed is seriously impoverished (see Pullum and Scholz, 2002). In this chapter, we consider both (kinds of) challenges to the poverty of the stimulus argument. We also examine one line of research that arguably constitutes the strongest nativist position within the generative framework. This is the Continuity Assumption, which proposes that child language is constrained by the boundaries of Universal Grammar (UG) throughout the entire course of language development.

The three models of language acquisition that we discuss will be evaluated throughout the present dissertation. In conducting this evaluation, we will consider one semantic property, which underpins various phenomena of natural languages. This property has to do with entailment relations between different sentences. The relevance of entailment relations for natural languages takes on different forms. Entailment relations regulate the use of particular expressions that show a limited distribution, e.g., so called negative and positive polarity items. Moreover, entailment relations play a role in more subtle phenomena, namely the interpretation of expressions like the disjunction operator or and a subclass of positive polarity items. The relevant facts will be introduced in the next section. We will argue that this cluster of phenomena provides a useful yardstick for evaluating the alternative models of language acquisition which are the focus of current debate.

1.1 Entailment Relations, Polarity Items and Inferences across Natural Languages

This section introduces the linguistic phenomena that will be considered throughout the present study. For expository purposes, we start by considering the following pair of questions.

(1)	a.	Would you like something to drink?
	b.	Would you like anything to drink?^1

Both questions in (1) are grammatical. Furthermore, they are quite similar meaning. Let us now consider the possible answers to the questions in (1).

(2)	a.	Yes. please. I would love something to drink!
	b.	*Yes, please. I would love anything to drink!
(3)	a.	*No, thank you. I don’t want something!^2
	b.	No, thank you. I don’t want anything!

The question is why a contrast is witnessed in (2) and (3), but not in (1). This puzzle has received considerable attention, beginning with Klima (1964). Over the last thirty years, several phenomena have been discussed in conjunction with the facts summarized above. As a result of this research, different properties have been held responsible for those facts. One property plays a prominent role in most accounts, however. This property is downward entailment.

1.1.1 Entailment Relations across Natural Languages

Downward entailment is a semantic property common to a variety of linguistic expressions. The defining property of downward entailing expressions is the licensing of inferences from a set to its subsets. Consider the examples in (4). Adult speakers of English recognize the inferences in (4) as valid. In each example, the difference between the two sentences is that a noun phrase, e.g., paper, is replaced by a second noun phrase, e.g., good paper, which picks out a subset of the set denoted by the first.

(4)	a.	John has not written a paper yet
		⇒ John has not written a good paper yet.
	b.	None of the students has written a paper yet
		⇒ None of the students has written a good paper yet.
	c.	None of the students who wrote a paper has received a grade yet
		⇒ None of the students who wrote a good paper has received a grade yet.
	d.	John graduated without writing a paper
		⇒ John graduated without writing a good paper.

As the invalid inferences in (5) suggest, it is easy to identify linguistic contexts that do not license valid inferences from a set to its subsets.³

(5)	a.	John wrote a paper
		*⇒ John wrote a good paper.^4
	b.	Some of the students wrote a paper
		*⇒ Some of the students wrote a good paper.
	c.	Some of the students who wrote a paper received a grade
		*⇒ Some of the students who wrote a good paper received a grade.
	d.	John graduated with a paper
		*⇒ John graduated with a good paper.^5

It is worth pointing out the existence of linguistic expressions which, though closely related in meaning, license different inferences (i.e., they differ in the direction of entailment relations). One clear contrast of this sort arises with the prepositions without versus with, as shown in (6).

(6)	a.	John passed the class without a grade
		⇒ John passed the class without a good grade.
	b.	John passed the class with a grade
		*⇒ John passed the class with a good grade.

The temporal subordinate conjunctions before and after exhibit the same contrast, as shown in (7).

(7)	a.	John graduated before he wrote his first paper
		⇒ John graduated before he wrote his first good paper.
	b.	John graduated after he wrote his first paper
		*⇒ John graduated after he wrote his first good paper.

In addition to presenting a minimal pair, the contrast in (7) is interesting for a second reason. Many of the downward entailing contexts illustrated in (4) seem to have a negative component which the non-downward entailing contexts in (5) lack. In the case of (7)a, however, this is not the case. Neither sentence in (7) has a negative flavor, but the two sentences generate different entailments. Similarly, a contrast that does not involve a negative component is witnessed in (8).

(8)	a.	Every student who wrote a paper received a grade
		⇒ Every student who wrote a good paper received a grade.
	b.	Every student wrote a paper
		*⇒ Every student wrote a good paper.

The contrast in (8) allows us to comment on another feature of downward entailment. Downward entailment is not a property of a sentence as a whole. Rather, downward entailment is limited to a specific environment. As (8)a shows, the universal quantifier every is downward entailing on the subject noun phrase. However, as witnessed by (8)b, the universal quantifier every does not create a downward entailing environment for the object noun phrase.

1.1.2 Polarity Items and Entailment Relations

Thus far, the property of downward entailment seems to have limited consequences for natural languages. Downward entailment only describes the environments in which a noun phrase can be substituted with one that picks out a subset of the denotation of the original noun phrase without affecting the truth of the original sentence. The relevance of entailment relations for natural languages becomes more tangible if we turn to the distribution of negative polarity items (NPIs), e.g., the words any and ever in English. As shown by the grammaticality of the sentences in (9) and (10), these negative polarity items are permitted in the same linguistic environments that license inferences from sets to their subsets (e.g., the linguistic environments in (4)).

(9)	a.	John didn’t write any paper.
	b.	None of the students wrote any paper.
	c.	None of the students who wrote any paper received a grade.
	d.	John graduated without any paper.
(10)	a.	John didn’t ever write a paper.^6
	b.	None of the students ever wrote a paper.
	c.	None of the students who ever wrote a paper received a grade.
	d.	John graduated without ever writing a paper.

Moreover, linguistic environments that fail to license inferences from sets to their subsets (i.e., the linguistic environments in (5)), also ban negative polarity items, as shown by the ungrammaticality of the sentences in (11) and (12).

(11)	a.	*John wrote any paper.
	b.	*Some student wrote any paper.
	c.	*Some student who wrote any paper received a grade.
	d.	*John graduated with any paper.
(12)	a.	*John ever wrote a paper.
	b.	*Some student ever wrote a paper.
	c.	*Some student who ever wrote a paper received a grade.

Still further parallels can be drawn between downward entailment and the distribution of negative polarity items. Examples (13) and (14) indicate that negative polarity items are tolerated with the preposition without, but not with, and with the temporal conjunction before, but not after.

(13)	a.	John passed the class without any grade.
	b.	*John passed the class with any grade.
(14)	a.	John graduated before he wrote any paper.
	b....

Table of contents

• Cover
• Original Title Page
• Title Page
• Copyright Page
• Table of Contents
• Acknowledgments
• Introduction
• Chapter 1 - Language and Acquisition
• Chapter 2 - Entailment and Polarity Phenomena in Child Language
• Chapter 3 - The Structure of Child Language
• Chapter 4 - Asymmetries of Child Language
• Chapter 5 - Structure and Beyond
• Chapter 6 - The Structure of the Universal Asymmetry and Beyond
• Appendix
• References
• Index of Names
• Index of Topics