Cultural differences. Cultural differences in telling the Pear Stories are very small, to the surprise of the original designers. The very simple experimental story succeeds all too well in being universally accessible, without dramatic cultural variation. It is also too short, without well-known cultural heroes such as Monkey or Superman to trigger distinctive tales. Grammatical differences in language are too small and too technical to produce differences in world view.

Minor differences in vocabulary appear. The Taipei speakers all talk about 'guavas' because pears were not available in the tropical Taiwan of the 1970's. The one consistent cultural difference was that the Californians differed from all the other speakers by seldom including any moral comment, such as 'stealing is wrong' or 'those kids were really bad'. The U.S. students were very familiar with experimental methods and probably wanted to appear 'objective'. Speakers from every other country, especially the Greeks, included many moral comments. They were also much more likely to use evaluative nouns such as 'thief' rather than 'boy' (Erbaugh 1990, Tannen 1980. See Table 1).

Table 1 Cultural Evaluation

Total mentions English Mandarin
Interpretive naming of man*
('landlord' vs. 'man')
6 26

Social interpretation *
(label 'theft', 'thanks', proverbs)

14 69

(* is statistically significant above the .001 level.)


Universals in brain and information processing produce the short, choppy bursts of speech, bounded by pauses, which Chafe calls 'intonation units'(1980, 1993). These appear in all the languages studied, including all the Chinese dialects. At first glance, the transcripts look shockingly disorganized. Readers often ask if they come from kindergarten children. In fact, the short idea units are organized into divisions which are indispensable for the brain to produce and understand speech. The average 'intonation unit' is 4 - 10 morphemes, regardless of language. Often, though not always, an intonation unit corresponds to a clause. The brain plans speech not word-by-word, but in grammatically coherent units. Pauses typically come at a grammatical boundary, not inside one. That is, people will say, 'a man... was picking pears', not 'a ... man... was... picking pears'. (Beginning language students must, of course, pause more often.)

Pauses offer listeners a clue to sentence divisions, as well as necessary time for analysis. The very large amount of repetition in spoken language is also universal across languages (with some variation among individual speakers). Silent pauses typically indicate that the speaker is making extra effort in planning what to say. Filled pauses, with an 'umm' or an 'uhh' indicate that the speaker is trying to hold the floor, while searching for the right word (Levelt 1983).

Speech errors also reveal mental processing. The majority of errors anticipate a coming phrase, indicated that we are planning ahead, sometimes several sentences in advance. Most errors are corrected immediately after the phrase; but we may remember and correct an error from some sentences, paragraphs, or even days back (Levelt 1983, Shen 1992).

A universal monitoring process tallies how well and how correctly both native and non-native speakers are speaking. Both first and second language speakers pause and err at the same grammatical junctures. Surprisingly, planning the main clause for the primary story line is most difficult, producing the most pauses and errors. Backgrounded and relative clauses are more optional. If speakers attempt them, they tend to be fluent and error free (Erbaugh 1986).

At the rhetorical level, most adults tell a well-organized Pear Story. The scenes closely match the cuts in the film. Speakers describe virtually identical actors and events, in near-identical prominence and sequence. The amount of information varies considerably among speakers, but not across languages. Some tell very detailed stories, while others summarize. One German told the whole story in one sentence intonation. The Californians told longer stories, but only because they included so many comments on film making. With film making comments removed, their stories were statistically the same length as the Mandarin and Greek speakers. The more discursive speakers also included a fair number of descriptive comments (Table 2).

Almost all adults, regardless of language, begin by identifying the main actor, the setting, and his actions. A few provide an overview, such as 'this story was about theft' or 'the story is very simple'. Speakers proceed chronologically through main actors and actions for each scenes. Important scenes, which lasted longer on film, produce more commentary. Personal comments such as 'this part was interesting' also occur throughout. Most adults then signal that the story is over with a simple phrase such as 'that's all'. Many non-English speakers also add an evaluation or a moral. This basic story grammar also underlies the stories of even beginning foreign language student adults. Story grammar is also robust enough to survive even stroke and many other causes of severe aphasia. Children, however, master story grammar only slowly and with effort throughout the primary school years (Berman, Slobin and Aksu 1994).

Table 2 English, Mandarin and Greek Narratives

  English Mandarin Greek
intonation units* 125 88 84
Film making clauses ('the camera angle
as good')*
17 3 3
Opening scene clauses 4 4 -
Narrative action clauses
('the boy got on his bike')
64 60 -
Descriptive clauses
('the pears green')
15 14 -
Personal comments*
('I liked this')
9 5 -

(Mean per speaker. * indicates statistically significant difference at .05 or above. Other figures are statistically identical. Not all figures for Greek are available. Based on Erbaugh 1990, Tannen 1980.)


Universal differences between spoken and written language emerge in the stories for which we have both spoken and written versions (Chafe and Danielewicz 1987, for Mandarin see Christensen 1994, 2000, Li and Thompson 1987 for Mandarin). We must speak fairly rapidly, or hearers will lose track of what we are saying. Writers, however, do a great deal of planning and editing. Written stories are less than half as long, containing much less repetition. Writing also uses much more complex grammar, for relative clauses and backgrounding, 'while the man was in the tree, the boy who had stolen the fruit passed by again'. Written texts make immediate and fewer demands on memory because readers, unlike listeners, can go back and look again at unclear sections.

Individual differences are striking. Typically, differences are greater among speakers of a particular language than across languages. The most striking difference is the amount of detail and summary. This is largely a matter of style and choice among adults. It does not measure language ability, for a fluent speaker may choose to tell a brief story. One measure of organization is events added out of sequence, 'oh, I forgot, he hit the rock before his hat blew off'. These do indicate some lack of pre-planning. But speakers may also add them because they view the task as a test of memory.

Language differences emerge largely at the level of grammar. Before the study, one hypothesis held that, because Chinese lacks verb tense, Chinese speakers would tell less time-oriented, less chronologically precise stories. This proved false. The Mandarin speakers told temporally very precise stories. They were also less than half as likely to add events out of sequence as the Californians (who were also a younger and less highly educated).

Table 3 Events out of Sequence

  English Mandarin
Number of speakers with events out of sequence 6 3

Number of revisions

8 3

(from Erbaugh 1990)


Another hypothesis held that Chinese might be vague in identifying actors, because of a neuter spoken pronoun for 'he/she/it', lack of verb inflections for person and number, and frequent (though grammatical) omission of subjects and objects ('zero anaphora'). In fact, analysis shows Chinese to be as referentially clear as English (Chen 1986 a, b, 1987, Cumming 1984). Japanese shows considerable zero reference, however (Clancy 1980, Downing 1986, 1996).

Mandarin uses a good deal of topicalization, with or without object fronting. Word order, however, is rather rigid, much like English. German, Italian, Mayan and Cantonese show much more variable order.

DIFFERENCES ACROSS THE CHINESE DIALECTS are only beginning to be analysed for connected speech, although T'sou's work (1999) indicates that spoken Cantonese uses about 70% different vocabulary from Mandarin. Cantonese also shows much more flexible word order and many more mood particles. More subtle differences in reference and temporal location are still being analysed (e.g. Matthews and Yip 1994). But the Cantonese speakers use over five times more specific noun classifiers per noun than Mandarin speakers. The content of the stories is almost identical, including the relative prominence of the objects mentioned. The greater use of classifiers reflects not greater mental precision, but more grammatical contexts requiring classifiers, especially for possessives and for repeated mentions of the same object (Erbaugh 2002). Classifier use in the other dialects is still being investigated.