Syntax and speech
المؤلف:
Paul Warren
المصدر:
Introducing Psycholinguistics
الجزء والصفحة:
P29
2025-10-29
32
Syntax and speech
One intriguing aspect of the formulation processes is how grammatical encoding relates to phonological encoding. That is, are aspects of the grammatical structure of sentences reflected in the way in which those sentences are spoken What is the relationship between syntax and speech? Do speakers employ some mapping between syntactic structures and speech structures which results in the marking of the constituent structure of the sentence, potentially of benefit to listeners?
The sentences with eat and persuade discussed in the preceding section involve superficially similar sentences that are actually quite different. The same can be said of other sentences containing structural ambiguity, i.e. which have differences in their sentence structures that correspond to different meanings of the sentences. The example in 2.25 is a global or standing ambiguity, since there is nothing in the sentence itself that resolves the ambiguity, while 2.26 and 2.27 contain a local or temporary ambiguity, hinging on whether or not the book is the grammatical object of read, which is confirmed by the final words of the sentences.
The differences between the two interpretations of each ambiguity correspond to differences in their assumed syntactic structure. In terms of the syntactic trees often used to describe sentence structures in linguistics, and with the omission of irrelevant detail, the two different structures for the prepositional phrase PP ambiguity in 2.25 can be represented as in 2.28 and 2.29. Note that these are not the only tree structures proposed by linguists for these sentence types – see sidebar for implications. The sentence tree in 2.28 shows that the PP with the binoculars attaches to the verb phrase as a modifier of the verb – this is the adverbial use of the PP, describing how the spy watched the man. The tree in 2.29 shows that the PP is part of the noun phrase NP headed by a, because in this reading the PP is a property of the NP the man has the binoculars. These tree structures show that the difference between the two meanings of the sentence in 2.25 is related to a difference in the height of attachment of the PP into the syntactic tree. The boundary before the PP is a more major one i.e. is within a higher-level sentence constituent in the adverbial treatment in 2.28 than when the PP modifies the N in 2.29.
In experiments where participants have to produce sentences like 2.25 with one or other of the meanings, this structural difference is reflected in a greater likelihood of pausing prior to the PP with the adverbial meaning, along with pre-pausal lengthening of the immediately pre ceding word, compared to the same word in the noun modification reading Cooper Paccia-Cooper, 1980. Many studies have also included an acoustic measure of voice pitch, known as fundamental frequency e.g. Cooper & Sorensen, 1981. Voice pitch falls towards the end of a constituent and re-sets to a higher level at the beginning of the next constituent, often with a continuation rise’ after the falling tune in the first constituent. Auditory impressions of intonation patterns like this are also used as a measure of prosodic disambiguation Speer, Warren Schafer, 2011.
Studies of local ambiguities like that in 2.26 and 2.27 show that these too differ in rhythmic breaks, and also in their intonation patterns, even though the sentences become disambiguated before they end. Chapter 10 discusses the comprehension of such ambiguities and shows that these speech cues are important for listeners working out the intended meaning as efficiently as possible.
Other speech cues that reflect the syntactic differences between these kinds of utterances include the incidence of connected speech processes. These are changes to the speech sounds of words that occur late in the production process, as a consequence of the phonetic environment that the sounds find themselves in. For instance, many varieties of English exhibit flapping, i.e. the pronunciation of /t/ and /d/ as sounds with the briefest of contacts between the tongue and the teeth ridge. This occurs between vowels, and might be found in words like rider or writer. It even occurs between vowels that are in different words, such as met Anne, as long as the boundary between the two words is not a strong syntactic one. So for speakers whose English typically features flapping, this might be found in last time we met Anne, she told us a great story, but not in Last time we met, Anne told us a great story (Cooper & Paccia Cooper, 1980).
Such processes even respect the derivational history of a sentence. Take the ambiguity in 2.30. On the one hand this sentence could mean that Max wants to leave someone, while on the other it could mean that Max wants someone else to do the leaving.
It is argued that these two meanings correspond to different underlying sentences. In the first case, the question in 2.30 is derived from a sentence that might have the structure in 2.31. The question word who is asking about the object of the underlying sentence, and if someone answered this sentence with Bob’, this would be taken to mean Max wants to leave Bob. This relationship of the question word to the rest of the sentence is shown in 2.31 by the use of a trace indicated by in italics which follows leave i.e. in the object position. The trace is co-indexed with the word who as indicated by the subscript on both who and t. The second meaning corresponds to the representation in 2.32, where the question word is now co indexed with a trace in the subject position before leave. The answer Bob’ to this question would be understood to mean Max wants Bob to leave. It is argued that the trace is somehow present in the surface structure of the sentence, and that the version of 2.30 that corresponds to 2.32 therefore has a trace between want and to leave that blocks the operation of a connected speech process known as wanna -contraction. Wanna -contraction results in the realisation of want to as wanna. In those dialects that regularly have wanna -contraction, it is more likely when the sentence has the meaning in 2.31 than when it has the meaning in 2.32 Nagel, Shapiro Nawy, 1994; Straub, Wilson, McCollum Badecker, 2001.
The examples we have looked at show that the timing structures of speech and the incidence of connected speech processes reflect syntactic sentence structure. They also show that syntactic structure is indicated at more boundaries than those marked by punctuation in corresponding written sentences. But other examples are more complicated. For example, take a sentence from part way through the nursery rhyme the house that Jack built. Those who know this rhyme will know that it can have much longer sentences than this. Typically, this would be recited with breaks as shown in 2.33, but in a traditional syntactic analysis the tree structure for the first part of the sentence would look like that shown in 2.34.
Notice that the main break in the syntactic tree, i.e. the highest point at which the tree starts to divide, is between this and is, the next highest break in the tree is between is and the cat, and so on. In addition, if pauses are primarily used to mark the ends of clauses, it is clear from the tree that there would be no pause until the end of the sentence, since the right-branching structure of this sentence means that no clauses finish until then. In short, the pattern of the syntactic breaks does not match the pattern of the breaks in the spoken sentence. In some treatments, this discrepancy is dealt with through adjustment rules that give the more even structure in 2.33 Chomsky, 1965; Cooper, 1980. Note that it could also be dealt with by having a different theory of sentence structure than the one represented in 2.34, i.e. one that gives different and flatter syntactic trees. Whatever the solution for this particular sentence, it is clear that the relationship between syntax and speech may not always be straightforward see also Gee &Grosjean, 1983.
One criticism that has been levelled against the kinds of studies that have been summarised in this section is that they are based on what speakers do when they read sentences aloud Allbritton, Mckoon & Ratcliff, 1996; Haywood, Pickering & Branigan, 2005. This is quite a different task from spontaneous speech, as we have seen above. Not only are there differences in the planning involved in these different tasks, but also the goals of a reader are quite different from those of someone speaking spontaneously, and could include the goal of wanting to produce a clear difference between, say, the two meanings of 2.25, because this is what the reader perceives the experimenter to want. Recent studies have attempted to address this criticism by using speech tasks in which the participant’s goals are much more closely aligned to ordinary interactions. These include collaborative game tasks, in which the outcome of the game is dependent on spoken inter action between two participants Speer et al, 2011. Such studies – on the whole – confirm that speakers produce distinctions such as those discussed in this section, distinctions that are linked to syntactic sentence structure.
الاكثر قراءة في Syntax
اخر الاخبار
اخبار العتبة العباسية المقدسة
