10.4 Spoken Versus Written English


In both touchtone and speech applications, it is easy to find messages that reflect the norms of written rather than spoken language for example, "I must first verify your account number" and "You may now record your message." Spoken and written language differ for a number of reasons (Crystal 1992). Speech is dynamic and transient. No sooner is a word or phrase spoken than it's gone forever. Also, in spoken language, the participants are present. Typically, they interact face-to-face. The speaker usually has a clear notion of the person he or she is addressing.

Writing, in contrast, is static and permanent. Written words are there for as long as the medium lasts. Because of this permanence, written language is generally more formal than spoken language, and it ensures the survival of certain conservative words, phrases, and grammatical structures for example, "To whom it may concern." In addition, there is always a lag, often of unknown duration, between the time of the writing and the time of the reading, as well as different settings, which may also be unknown, where the writing and the reading take place.

Another important difference is that the writer is distant from the reader, cannot see the reader, and may not know who the reader is. Of course, this is also the case with speech applications. However, all speech applications crucially depend on spoken language, which has evolved in a different direction than writing because speech is prototypically face-to-face and prototypically more personal. To find yourself participating in a conversation, even one that has been engineered, with a persona who is reading formally written text at you, the content of which depends on your own unrehearsed, spontaneous responses, is unprecedented in authentic discourse. Because it cannot be likened to any real-world experience, we can actually consider it an "antimetaphor," and so this kind of artificial formality is undesirable for interface design. As we've mentioned before, the highest-level metaphor that will most easily usher the VUI user from recognition state to recognition state is that of an everyday conversation. VUI prompts should therefore adhere to spoken rather than written norms, still guiding and directing the user through the interaction.

In the sections that follow, we examine a few differences between speaking and writing in English and demonstrate their relevance to prompting. These differences motivate the methodological suggestion in Chapter 8 to read your prompts aloud; as you listen to them, you will react differently than when you merely read them on the printed page.

10.4.1 Pointer Words

One manifestation of the many differences between spoken and written language lies in their use of pointer words for example, this, that, here, and there. The most basic use of pointer words is to indicate position or location of things in physical space: "This car isn't for sale, but that red one over there is." As this example illustrates, the distinction is roughly equivalent to "near" versus "far." It is typical, however, for languages to extend the use of the physical pointers to the domain of discourse in other words, to point either forward or backward in a stretch of written or spoken language. Examples: "Here's what happened" (pointing forward to what is about to be said) and "There you go again!" (pointing backward to what someone else has just said).

Prompt writers should be aware that the directionality of this versus that, in particular, differs depending on whether they are used in spoken versus written language, as shown in Figure 10-3. These examples demonstrate that the back-pointer of choice in speech is that, whereas the back-pointer of choice in writing is this.

Figure 10-3. In speech, that points backward, but in writing, this is often preferred.

graphics/10fig03.gif

The next pair of examples (Quirk and Greenbaum 1973) relates to speaking only. As shown in Figure 10-4, that points backward, and this points forward.

Figure 10-4. In speech, that points backward, and this points forward.

graphics/10fig04.gif

In conversation, therefore, statements such as the ones in (32), (33), and (34) are most easily interpreted as forward-pointing. (In other words, "this is" is synonymous with "here's.")

(32)

This is the last item on the list.


(33)

This is your last message.


(34)

This is the end of your favorites list.


Although these statements are used and interpreted as forward pointers in everyday conversation, they are sometimes used in VUIs to point backward, in compliance with the norms of written English. The unfamiliar use of spoken this as a back-pointer in these dialogs would likely impede listening comprehension.

An example of a prompt that uses both this and that appropriately is shown in (35).

graphics/sound_icon.gif

(35)

This is the number I heard: 555 749 9591. Did I get that right?


This points forward to the phone number and could be replaced with "here." That points backward to the number and could be replaced with "it."

10.4.2 Contraction

Spoken language tends to favor contractions (you're, can't, it's, don't), whereas written language tends to avoid them (you are, cannot, it is, do not). Example (36) is a conversational prompt with three contractions.

graphics/sound_icon.gif

(36)

. . . Finally, if you're finished with your bookmark list, just say "I'm done," and I'll put it away.


Do not expect a professional voice actor to take the liberty of substituting contractions for uncontracted forms. Although your voice actor may do you this occasional favor, those in the business consider it unprofessional to edit scripted material extemporaneously unless there is some understanding or agreement to the contrary.

Contractions have occasionally been discouraged in VUI designs based on the incorrect assumption that they are characteristic of "lazy" speech. If contractions were lazy, then people who are feeling lazy should be able to respond affirmatively to a question such as, "Are you hungry?" with simply "Yes, I'm." Native speakers of English unanimously reject the contraction in such cases. Another impossible contraction is "you're" in "Sandy's tired, and you're, too." Such examples suggest that the distribution of contractions in speech probably has more to do with syntactic and rhythmic constraints than with undesirable personality traits such as sloth or carelessness.

In other cases, contractions have been avoided on the grounds that the content of the VUI is, for example, financial, and banking demands a certain level of formality. There is no evidence, however, that people stop using contractions in speech just because they work in a bank or are trading shares in a volatile market. Contractions should be used in your VUI to the extent that they are a defining characteristic of everyday spoken language, regardless of the conversation's level of formality. The point is to provide users with a familiar, comfortable language experience.

10.4.3 Must and May

Must and may have two basic meanings each. One can be loosely described as social/interactive, and the other one as a logical/probability meaning (Celce-Murcia and Larsen-Freeman 1999). The social/interactive meaning of must is "obligation" or "necessity," and the logical meaning is "deduction" or "inference" (see Figure 10-5).

Figure 10-5. Must has social as well as logical meanings.

graphics/10fig05.gif

According to Melrose (1999), in speech, native speakers of North American English reserve must for inferences, which is the logical/probability meaning. In other words, although people use must to signal obligation in writing, they avoid it in speech, even in formal settings. Instead of must, the expression they use for obligation or necessity is generally have to (e.g., "Sorry, but I have to leave now").[4] Finally, Melrose notes that the social must seems to be more common in British English for example, "You must come over for dinner soon" and "We must correct that problem as soon as possible."

[4] Melrose also found that (have) got to or have gotta is used in both social and logical contexts, especially when there is some sense of urgency for example, "You gotta lend me ten dollars" (social use signaling obligation) and "You've gotta be kidding me" (logical use signaling inference).

The stilted quality of the prompts in (37) and (38) can therefore be repaired by replacing social must with more conversational alternatives, as in (39) and (40), respectively.

(37)

You must say your PIN one digit at a time for example, two one zero zero.


(38)

We must first get your starting point.


graphics/sound_icon.gif

(39)

Go ahead and say your PIN one digit at a time for example, two one zero zero.


graphics/sound_icon.gif

(40)

First, let's get your starting point.


Like must, the verb may also has social/interactive versus logical/probability uses, and also like must, it is only the logical/probability use that we find in everyday conversations. The social/interactive use of may is to ask for or grant permission, whereas the logical use is to express possibility (see Figure 10-6).

Figure 10-6. Like must, may has social as well as logical meanings.

graphics/10fig06.gif

With social may, the speaker is implicitly referencing a position of authority and an inequality in status, so this is a relatively formal way to grant or request permission. In these cases, can is more egalitarian and friendlier. In North America, there is less social stratification than in other language communities around the world, so, not surprisingly, can is more prevalent than social may, and it is preferred in contexts such as, "You can (may) go now" (Bailey 1999; Melrose 1999). Presumably, North Americans are put off by the implicit social inequality, formality, and coldness of may in its social use. In contrast, social roles and setting are irrelevant to the use of may in the logical/probability context. In Figure 10-6, the speaker is assessing the probability of rain, independent of the social aspects of the conversation (Celce-Murcia and Larsen-Freeman 1999).

In the interest of projecting a more favorable persona, social may in (41) and (42) should be reworded as in (43) and (44), respectively.

graphics/sound_icon.gif

(41)

. . . At any time, you may also ask for help.


(42)

When you are finished recording, you may hang up or press pound for more options.


(43)

. . . At any time, you can also ask for help.


(44)

When you're finished recording, feel free to hang up. Or to hear some more choices, press the pound key.


Note in (44) the contraction "you're" and the restructuring of information to accommodate the End-Focus Principle.

10.4.4 Will Versus Going To

In general, will is more formal and impersonal as an indicator of futurity, whereas going to is more informal and personal. Because it is more formal, will is often preferred in writing. For native speakers of English, however, sometimes these verb forms cannot be interchanged. In conversation, will is commonly used to signal spontaneous volition, as in (45) and (46), and future contingency, as in (47) and (48). In contrast, going to signals planned intention, as in (49) and (50), and "the future on its way," as in (51) and (52).

(45)

A:

That was my mother's favorite vase.

B:

Oops! I'll buy you another one.


(46)

A:

I can't really talk right now.

B:

Okay, I'll call you tonight.


(47)

If you buy a house, you'll get a huge tax write-off.


(48)

If you put your pawn there, he'll win the game.


(49)

A:

What's Harvey doing with my pliers?

B:

He's going to fix the TV antenna.


(50)

A:

What's on your agenda for tomorrow?

B:

I'm gonna get a haircut.


(51)

Stop the car! I'm going to be sick!


(52)

Look, it's going to rain.


If going to in these examples is replaced with will, the result is very unnatural, and yet will is often used in VUI prompts where we would conversationally expect going to. Take, for example, the prompt in (53), which is part of a lengthy enrollment process in a subscriber-based service and is followed by instructional information.

(53)

I will now record your account number. Say the number one digit at a time.


This prompt can be rewritten to conform to conversational norms by replacing "I will" with "I'm going to." In addition, the placement of the adverb "now" between the two verbs ("will" and "record") yields another structure particular to written discourse or in some formal spoken genres, such as when a professional magician says, "I will now saw my lovely assistant in half." We can rewrite (53) as (54).

graphics/sound_icon.gif

(54)

Now I'm going to record your account number. Tell me the number one digit at a time.


Another example comes from a VUI that requires several steps to provide the caller with driving directions. At one point in the sequence, we receive some helpful orientation, shown in (55).

(55)

We will now collect your address.


Although this prompt is intended to be heard over the telephone, it is representative of written discourse. First, we hear will instead of going to, even though planned intention is a better metaphor in this context than spontaneous volition. Again, we find the placement of "now" between the auxiliary ("will") and the main verb ("collect"). Finally, the use of "collect" in the sense of "elicit information" is unnecessarily technical. This prompt can be recast as shown in (56).

graphics/sound_icon.gif

(56)

Now I'm going to ask you a few questions to find out where you're going.


This is longer than the original, but it is easier to listen to, because it avoids those discourse features that are alien to spoken language and instead exploits familiar linguistic conventions.

10.4.5 "Romans Perspire, Anglo-Saxons Sweat"

We've noted that written language often favors certain words, phrases, and grammatical structures that are avoided in speech. This section focuses on vocabulary in spoken versus written language.

The title of this section, "Romans perspire, Anglo-Saxons sweat," alludes to the richness of the vocabulary of the English language. Owing to our dual inheritance of words from the Romance and Germanic language families, we often have recourse to two words that are roughly synonymous, as in the case of verbs such as sweat/perspire, put out/extinguish, and leave/exit, nouns such as fish tank/aquarium, crowd/multitude, lunch room/cafeteria, and drink/beverage, adjectives such as kingly/regal, funny/humorous, friendly/amicable, and fat/corpulent, adverbs such as yearly/annually, and so on. In terms of usage, however, Latin-derived alternatives (such as perspire and extinguish) usually have a more formal, technical ring, whereas their Anglo-Saxon counterparts sound more informal and colloquial.

When we write, especially in the case of more careful genres, we gravitate toward more formal (Latin-derived) alternatives, given the relative permanence or persistence of written messages. The writing of prompts, which are intended to be spoken, should avoid this tendency. The list of verbs in the left column in Table 10-1 was culled from a review of several speech and touchtone applications; the right column shows more colloquial, contextually appropriate equivalents.

Table 10-1. Formal Versus Colloquial Verbs

FORMAL

COLLOQUIAL

acquire

get

activate an account

set up an account

create an account

set up an account

create a bookmark

make, add a bookmark

encounter [information]

find, come across

encounter [difficulty]

have problems

exit list

be done with a list

experience difficulties

have problems

obtain

get

pause

take a break

provide

give

receive

get

request, collect

ask for

respond

answer

return

go back

select

choose, pick

terminate

end, finish



Voice User Interface Design 2004
Voice User Interface Design 2004
ISBN: 321185765
EAN: N/A
Year: 2005
Pages: 117

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net