Chapter 7: Acquisition in the prepositional domain 0 Introduction

28
Chapter 7: Acquisition in the prepositional domain 0 Introduction The purpose of this chapter is to present the findings from a longitudinal study of preposition acquisition; these findings support the fine-grained approach to the prepositional domain detailed in Chapters 3 and 4. While a number of studies in different fields provide partial evidence for the analysis offered here, they do not provide comprehensive, conclusive evidence (Chapter 5). The fact that children are known to acquire language in developmental stages where lexical classes of elements are acquired before functional ones provides a rich testing ground for the view that syntactic domains are comprised of four discrete categories. If syntactic domains in general, and the prepositional domain in particular, are comprised of elements with various designations for lexical and functional information, then the purely lexical elements of the domain (adverbs, in the prepositional domain) should be acquired first, and the purely functional elements (functional prepositions) should be acquired last. The ordering of the intermediary categories (particles and semi-lexical preposition) will determine which of two features [±Lexical] and [±Functional] is more highly ranked; the order of acquisition of these two categories cannot be predicted based on anything internal to the model, and is left to emerge from the data. Thus there are two possible developmental orders that are possible. In the first, particles are acquired before semi-lexical prepositions (1), and in the second, the particles follow the semi-lexical prepositions (2).

Transcript of Chapter 7: Acquisition in the prepositional domain 0 Introduction

Chapter 7: Acquisition in the prepositional domain

0 Introduction

The purpose of this chapter is to present the findings from a longitudinal study

of preposition acquisition; these findings support the fine-grained approach to the

prepositional domain detailed in Chapters 3 and 4. While a number of studies in

different fields provide partial evidence for the analysis offered here, they do not

provide comprehensive, conclusive evidence (Chapter 5). The fact that children are

known to acquire language in developmental stages where lexical classes of elements are

acquired before functional ones provides a rich testing ground for the view that syntactic

domains are comprised of four discrete categories. If syntactic domains in general, and

the prepositional domain in particular, are comprised of elements with various

designations for lexical and functional information, then the purely lexical elements of the

domain (adverbs, in the prepositional domain) should be acquired first, and the purely

functional elements (functional prepositions) should be acquired last. The ordering of

the intermediary categories (particles and semi-lexical preposition) will determine which

of two features [±Lexical] and [±Functional] is more highly ranked; the order of

acquisition of these two categories cannot be predicted based on anything internal to the

model, and is left to emerge from the data. Thus there are two possible developmental

orders that are possible. In the first, particles are acquired before semi-lexical

prepositions (1), and in the second, the particles follow the semi-lexical prepositions (2).

192

(1) Predicted order of acquisition:–Functional +Lexical Adverb–Functional –Lexical Particle+Functional +Lexical Semi-lexical preposition+Functional –Lexical Functional preposition

(2) Predicted order of acquisition:–Functional +Lexical Adverb+Functional +Lexical Semi-lexical preposition–Functional –Lexical Particle+Functional –Lexical Functional preposition

In both of these two possible orders, the purely lexical elements (adverbs) are acquired

first and the purely functional elements (functional prepositions) are acquired last; the

exact order of the two remaining elements (semi-lexical prepositions and particles) is left

to discovery in the data.

In examining the data from the five children studied (see Chapter 6 for the details

of the subjects and methodology), strong evidence of the predicted order of elements is

found. First, the general results will be presented, followed by a brief discussion of the

adult data. Then we turn to an examination of the rates of accuracy and the order of

acquisition that was found. Finally, additional patterns in the data that support

differences in the four prepositional elements will be discussed.

1 Results

To begin with, a total of 38,052 contexts in the data of the five children examined were

identified that contained or should have contained material homophonous with

prepositional elements. As explained in Chapter 6, a number of utterances were

excluded from the coding process because they were unintelligible (as indicated in the

transcript), ambiguous, or didn’t necessarily reflect the child’s own analysis (like

193

recitations and perfect imitations); these excluded contexts are listed as ‘other’ in the

tables showing the overall totals, and aside from these initial tables, will not be discussed

further. Setting aside the excluded ‘other’ contexts, a total of 25,695 obligatory

prepositional contexts (both correct and incorrect) were coded for status as

prepositional elements (adverb, particle, semi-lexical preposition, or functional

preposition) on the basis of the distributional patterns presented earlier (Chapters 2 and

4, especially, and as reviewed in Chapter 6). The general breakdown of the contexts can

be found in the appendix, in Tables (9.1), (9.2), and (9.3) for the main group of children,

the adult data, and the two additional children, respectively. Tables listing the frequency

of word-forms for each prepositional category are also provided in the appendix.

Of the 25,697 prepositional contexts, Adam’s data accounted for 7783 of these

contexts, and Eve’s contained 2062 total contexts. In Naomi’s data, 2325 prepositional

contexts were identified, and Nina’s contained 8355 contexts. 5172 contexts were

identified in Sarah’s data. The distribution of correct and incorrect uses across MLU

Groups and across prepositional type is shown in Tables (7.1) and (7.2), respectively.

194

Correct ContextsCorrect ContextsCorrect ContextsGroup 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+ Totals

AdamAdverb * * 267 511 137 1,006 409 2,330Particle * * 16 58 24 130 83 311

Semi-Lexical Prep * * 184 668 206 2,268 716 4,042Functional Prep * * 1 39 23 274 87 424

Total 0 0 468 1,276 390 3,678 1,295 7,107

EveAdverb * 66 53 160 290 * * 569Particle * 1 2 8 57 * * 68

Semi-Lexical Prep * 31 46 223 760 * * 1,060Functional Prep * 1 9 16 88 * * 114

Total 0 99 110 407 1,195 0 0 1,811

NaomiAdverb 49 42 107 248 204 151 96 897Particle 1 3 11 20 32 27 13 107

Semi-Lexical Prep 0 13 78 201 324 311 204 1,131Functional Prep 0 0 0 1 31 48 25 105

Total 50 58 196 470 591 537 338 2,240

NinaAdverb * 114 264 638 206 356 269 1,847Particle * 8 3 17 23 86 26 163

Semi-Lexical Prep * 212 376 1,679 879 1,505 806 5,457Functional Prep * 9 14 138 94 171 91 517

Total 0 343 657 2,472 1,202 2,118 1,192 7,984

SarahAdverb * 52 226 165 488 511 * 1,442Particle * 10 19 18 74 83 * 204

Semi-Lexical Prep * 79 266 435 859 1,324 * 2,963Functional Prep * 0 8 93 84 160 * 345

Total 0 141 519 711 1,505 2,078 0 4,954

TOTAL 50 641 1,950 5,336 4,883 8,411 2,825 24,096* No data available

Table (7.1): Correct contexts, child data

195

Error ContextsError ContextsError ContextsGroup 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+ Totals

AdamAdverb * * 12 6 3 1 3 25Particle * * 0 0 0 2 2 4

Semi-Lexical Prep * * 191 220 19 64 7 501Functional Prep * * 24 62 26 24 10 146

Total 227 288 48 91 22 676

EveAdverb * 0 1 2 1 * * 4Particle * 0 0 0 2 * * 2

Semi-Lexical Prep * 34 83 58 32 * * 207Functional Prep * 3 11 6 18 * * 38

Total 37 95 66 53 251

NaomiAdverb 0 0 0 3 1 0 1 5Particle 0 0 1 0 2 1 0 4

Semi-Lexical Prep 0 5 17 20 10 9 0 61Functional Prep 0 1 4 5 2 0 3 15

Total 0 6 22 28 15 10 4 85

NinaAdverb * 5 2 7 4 7 11 36Particle * 0 0 3 0 0 0 3

Semi-Lexical Prep * 73 42 137 18 20 6 296Functional Prep * 4 8 15 3 5 1 36

Total 82 52 162 25 32 18 371

SarahAdverb * 2 1 0 0 6 * 9Particle * 0 0 0 0 0 * 0

Semi-Lexical Prep * 38 48 16 24 23 * 149Functional Prep * 5 9 5 22 19 * 60

Total 45 58 21 46 48 218

TOTAL 0 170 454 565 187 181 44 1,601* No data available

Table (7.2): Incorrect contexts, child data

The adult data contained 13,501 total prepositional contexts. Adam’s mother

accounted for 25% (3,426 contexts) of the adult data, Eve’s mother accounted for 13%

(1,813 contexts), Naomi’s parents accounted for 6% (866 contexts), Nina’s mother

accounted for 40% (5,487), and Sarah’s mother accounted for 14% (1,897 contexts).

196

The distribution of total correct contexts for each prepositional category is provided in

Table (7.3). No totals are given for errors, as only 12 errors occurred (evenly distributed

across all adults), giving the adults an error rate of .08%.

Group 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+ Totals

AdamAdverb * * 187 254 48 125 97 711Particle * * 13 35 9 27 14 98

Semi-Lexical Prep * * 469 800 107 537 314 2,227Functional Prep * * 90 121 19 113 47 390

Total 0 0 759 1,210 183 802 472 3,426

EveAdverb * 94 86 32 189 * * 401Particle * 4 4 1 30 * * 39

Semi-Lexical Prep * 213 298 108 554 * * 1,173Functional Prep * 40 45 13 102 * * 200

Total 0 351 433 154 875 0 0 1,813

NaomiAdverb 10 54 46 28 26 41 20 225Particle 3 4 14 7 13 15 5 61

Semi-Lexical Prep 25 83 71 78 87 107 46 497Functional Prep 4 17 16 13 18 8 7 83

Total 42 158 147 126 144 171 78 866

NinaAdverb * 108 91 350 81 183 121 934Particle * 6 15 22 7 31 11 92

Semi-Lexical Prep * 301 313 1,402 349 935 402 3,702Functional Prep * 59 87 266 96 150 101 759

Total 0 474 506 2,040 533 1,299 635 5,487

SarahAdverb * 34 88 93 147 107 * 469Particle * 7 23 24 26 23 * 103

Semi-Lexical Prep * 97 191 202 377 257 * 1,124Functional Prep * 19 29 44 58 51 * 201

Total 0 157 331 363 608 438 0 1,897

TOTAL 42 1,140 2,176 3,893 2,343 2,710 1,185 13,489* No data available

Table (7.3): Correct contexts, adult data

197

2 Adult data

In this section, the distribution of the four prepositional categories in the adult

data will be considered. By examining the use of these four elements in the adult

language, a baseline of adult patterns is established, so that the linguistic target that the

children are aiming for is known. This may also help to differentiate between patterns

of normal prepositional usage and patterns due to acquisition in the child data.

In order to establish the patterns of adult usage of the four prepositional

categories, the proportion of use of each element in each MLU Group was calculated

(the total number of uses of a prepositional element in a specific MLU Group was

divided by the total number of prepositional uses found in that MLU Group). The

results of this calculation can be seen in Table (7.4).

Interestingly, all adults show identical patterns of use for the four prepositional

elements: all parents use far more semi-lexical prepositions than any other prepositional

elements. This category accounts for an average across parents of 62.4% of all

prepositional contexts. The second most frequently used prepositional elements were

the adverbs, which accounted for 22.3% of all adult contexts. This was followed by the

functional prepositions, which accounted for an average of 11.4% of usage across all

adults, and particles, which accounted for 3.8% of all adult prepositional usage. As can

be seen in Table (7.4), all parents follow this pattern very consistently, with only slight

variations between MLU Groups and individuals. Looking at the means across parents

in each of the MLU Groups (Figure 7.1), the pattern shows very little variation over

198

time: adverbs show a slight decrease in frequency over time, but the other three elements

remain essentially flat from MLU Group 1 through MLU Group 7.

* No data available

Group 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+ Mean St. Dev.

AdamAdverb * * 25% 21% 26% 16% 21% 22% 4%Particle * * 2% 3% 5% 3% 3% 3% 1%

Semi-Lexical Prep * * 62% 66% 58% 67% 67% 63% 4%Functional Prep * * 12% 10% 10% 14% 10% 12% 2%

Total 100% 100% 100% 100% 100% 100%

EveAdverb * 27% 20% 21% 22% * * 22% 3%Particle * 1% 1% 1% 3% * * 2% 1%

Semi-Lexical Prep * 61% 69% 70% 63% * * 66% 4%Functional Prep * 11% 10% 8% 12% * * 10% 1%

Total 100% 100% 100% 100% 100%

NaomiAdverb 24% 34% 31% 22% 18% 24% 26% 26% 5%Particle 7% 3% 10% 6% 9% 9% 6% 7% 2%

Semi-Lexical Prep 60% 53% 48% 62% 60% 63% 59% 58% 5%Functional Prep 10% 11% 11% 10% 12% 5% 9% 10% 2%

Total 100% 100% 100% 100% 100% 100% 100% 100%

NinaAdverb * 23% 18% 17% 15% 14% 19% 17% 3%Particle * 1% 3% 1% 1% 2% 2% 2% 1%

Semi-Lexical Prep * 64% 62% 69% 65% 72% 63% 66% 4%Functional Prep * 12% 17% 13% 18% 12% 16% 14% 3%

Total 100% 100% 100% 100% 100% 100% 100%

SarahAdverb * 22% 27% 26% 24% 24% * 24% 2%Particle * 4% 7% 7% 4% 5% * 6% 1%

Semi-Lexical Prep * 62% 58% 56% 62% 59% * 59% 3%Functional Prep * 12% 9% 12% 10% 12% * 11% 2%

Total 100% 100% 100% 100% 100% 100%

Table (7.4): Proportion of prepositional elements in each MLU Group, Adult data

199

70%60%50%40%30%20%10%0%

Adverbs Particles Semi-Lex Preps Funct Preps

Grp 1

Grp 2

Grp 3

Grp 4

Grp 5

Grp 6

Grp 7

Figure (7.1): Average proportion of use across MLU Group, adult data

3 Child acquisition patterns

In order to determine what the order of acquisition of prepositional elements is,

the generally accepted threshold of 90% correct usage in obligatory contexts (along the

lines of Brown (1973)) will be used. Brown’s (1973) actual criterion of 90% accuracy

level in three consecutive files was not adopted here. The files were examined using this

criterion first, but because the particles occurred so infrequently, three files rarely

occurred consecutively. Using this criterion in Naomi’s data showed that particles were

acquired in MLU Group 5, despite the fact that they occurred intermittently at accuracy

level of 100% since the first MLU Group. In Sarah’s case, the semi-lexical prepositions

would have been accepted as acquired in MLU Group 2, despite the fact that in the

immediately following files, semi-lexical prepositions are used with the following

accuracy levels: 0%, 33%, 40%. Her ability to correctly use semi-lexical prepositions

continues to vacillate through MLU Group 3 (13 of 25 files are below the 90% rate, and

of those, 6 show accuracy levels of 50% to 67%), and in MLU Group 4 her usage is

consistently above 90% (with one file falling below the 90% mark). A similar situation

200

occurs in Adam’s data, with his use of functional prepositions. Outside of these

problems, the pattern of acquisition shown by the three-consecutive files approach and

the MLU Groups averaging approach were identical. In one child, Eve, the timing of her

acquisition of functional particles becomes clearer with a file-by-file approach, as will be

discussed below.

Table (7.5) shows the percent of correct usage in each MLU Group for each

child’s use of the prepositional elements (rates above 90% are shaded). Each score

reported in the tables below is the result of dividing the number of correct uses of a

prepositional element by the total of all obligatory contexts for that element. In the next

five subsections, the acquisition patterns of each child will be briefly outlined, followed

by a summary of the patterns identified.

201

Group 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+

AdamAdverb * * 96% 99% 98% 100% 99%Particle * * 100% 100% 100% 98% 98%

Semi-Lexical Prep * * 49% 75% 92% 97% 99%Functional Prep * * 4% 39% 47% 92% 90%

EveAdverb * 100% 98% 99% 100% * *Particle * 100% 100% 100% 97% * *

Semi-Lexical Prep * 48% 36% 79% 96% * *Functional Prep * 25% 45% 73% 83% * *

NaomiAdverb 100% 100% 100% 99% 100% 99% 99%Particle 100% 100% 92% 100% 94% 100% 100%

Semi-Lexical Prep n/a 72% 82% 91% 97% 97% 100%Functional Prep n/a 0% 0% 17% 94% 100% 89%

NinaAdverb * 96% 99% 99% 98% 98% 96%Particle * 100% 100% 85% 100% 100% 100%

Semi-Lexical Prep * 74% 90% 92% 98% 99% 99%Functional Prep * 69% 64% 90% 97% 97% 99%

SarahAdverb * 96% 100% 100% 100% 99% *Particle * 100% 100% 100% 100% 100% *

Semi-Lexical Prep * 68% 85% 96% 97% 98% *Functional Prep * 0% 47% 95% 79% 89% *

* No data availableTable (7.5): Percent correct usage, child data

3.1 Adam

Of all the children, Adam’s data begins the latest: in MLU Group 3. At this

point, he has fully acquired adverbs and particles, using them accurately in 96% (279

total obligatory contexts) and 100% (16 total contexts), respectively. These highly

accurate levels of use continue throughout the remaining MLU Groups. In contrast,

Adam’s semi-lexical and functional prepositions in MLU Group 3 are not fully acquired.

The semi-lexical prepositions are used correctly 49% (375 contexts) of the time at this

202

stage, and the correct uses continue to increase until they reach 92% (225 contexts)

correct usage in MLU Group 5; the rate of correct usage continues to increase in MLU

Groups 6 and 7, with 97% (2332 contexts) and 99% (723 contexts) rates of correct use.

The functional prepositions in Group 3 are used correctly only 4% (25 contexts) of the

time, and continue to increase, with 39% (101 contexts) and 47% (49 contexts) in the

next two stages. Adam’s use of functional prepositions reach 92% (298 contexts)

accuracy in MLU Group 6, and remains at 90% (97 contexts) in MLU Group 7.

Overall, the adverbs and particles in Adam’s speech are both used appropriately in over

90% of all contexts throughout all MLU Groups, and no conclusions can be made with

regard to the order of acquisition of these two elements. However, it is clear that the

semi-lexical prepositions are not fully acquired until later (MLU Group 5), and the

functional prepositions are fully acquired even later, in MLU Group 6.

3.2 Eve

Eve’s use of prepositional elements mirrors Adam’s in many ways. Her first

MLU Group 2 shows full acquisition of adverbs and particles, with 100% correct use

for both (66 adverb contexts and 1 particle contexts), and these high rates of correct use

remain at or near 100% in all subsequent stages. The semi-lexical prepositions are used

correctly 48% (65 contexts) of the time in MLU Group 2, and increase in correct usage

until they are fully acquired in MLU Stage 5, with a rate of 96% (792 contexts). The

functional prepositions are used correctly in only 25% of the four obligatory contexts in

Eve’s initial MLU Group, and gradually increase in correct usage until MLU Group 5,

where they are used correctly 83% (106 contexts) of the time. While this is not above

203

the 90% threshold for full acquisition, a closer look shows that she has reached the point

of full acquisition by the end of MLU Group 5. Her first two files in the group show

accuracy levels of 40% and 50%; her rate of usage increases over the next couple of files,

and her last four consecutive files are all above 90%. As with Adam’s data, no

conclusions can be reached about the acquisitional order of adverbs and particles.

However, it is clear that in Eve’s data, semi-lexical prepositions are acquired after

adverbs and particles, and before functional prepositions.

3.3 Naomi

Of all the children examined, Naomi’s data offers the most complete view of the

acquisition of prepositional elements, because her data extends fully from MLU Group

1 to MLU Group 7. In her earliest MLU Group, she used both adverbs and particles

correctly 100% of the time (49 adverb contexts and 1 particle contexts), showing her

acquisition of these elements. The use of these two elements remains well above 90% in

all subsequent MLU Groups. Naomi used no semi-lexical prepositions in MLU Group

1, and in MLU Group 2, her rate of correct usage is 72% (18 contexts). This rate

increases over time, and surpasses the 90% threshold of acquisition in MLU Group 4,

with a rate of 91% (221 contexts). The rate of correct usage continues to increase, and

finally reaches 100% (204 contexts) in MLU Group 7. No contexts requiring functional

prepositions occur until MLU Group 2, and her rate of correct usage in this stage as well

as the next is 0% (1 context in MLU Group 2, and 4 contexts in MLU Group 3): all

obligatory contexts contain errors in these two stages. Her ability to use functional

prepositions continues to improve over time, and reaches the rate of full acquisition in

204

MLU Group 5 with a rate of 94% (33 contexts). This rate reaches 100% (48 contexts)

in MLU Group 6, and dips slightly to 89% (28 contexts) in the final MLU Group.

Importantly, Naomi’s data display exactly the same pattern that was exhibited in Adam

and Eve’s data: semi-lexical prepositions are acquired later than adverbs and particles,

but before functional prepositions. While the examination of accuracy rates indicate that

adverbs are particles are acquired at the same time, a closer examination of Naomi’s data

suggests that there is in fact a difference between these two elements. This point will be

returned to later, when the earliest child files are examined.

3.4 Nina

Similar to the previous three children, Nina’s use of both adverbs and particles is

above 90% in her first available MLU Group, with adverbs being used correctly in 96%

of 119 obligatory contexts, and particles being used correctly in all eight obligatory

contexts. Her rate of correct usage generally hovers around 100%, except in MLU

Group 4, where the particles are used correctly only 85% (20 contexts) of the time.

Nina uses semi-lexical prepositions correctly in 74% of 285 required contexts in MLU

Group 2, and in the next MLU Group she has them fully acquired, using them correctly

90% (418 contexts) of the time. In the subsequent stages her correct usage increases,

and reaches 99% (812 contexts) accuracy in MLU Group 7. Functional prepositions are

used correctly at a rate of 69% (13 contexts) in MLU Group 2, and are fully acquired

with a 90% (153 contexts) of accuracy in MLU Group 4. Nina’s rate of correct use

continues to increase in all subsequent groups, and reaches 99% (92 contexts) in MLU

Group 7. Once again, this child’s pattern mirrors the others: semi-lexical prepositions

205

are used at fully acquired rates after adverbs and particles are acquired, but before

functional prepositions are; no difference between adverbs and particles can be drawn.

3.5 Sarah

As expected, Sarah’s data matches the patterns of the other four children. Her

use of adverbs and particles begins at the fully-acquired rates of 96% (54 contexts) and

100% (10 contexts) in MLU Group 2, and both remain at or near 100% in all subsequent

files. Semi-lexical prepositions are used correctly in 68% of the 117 required contexts in

Sarah’s MLU Group 2, increasing to 85% (314 contexts) in MLU Group 3, and finally

surpassing the threshold for acquisition in MLU Group 4 with a rate of 96% (451

contexts). In all remaining stages, her use of semi-lexical prepositions continues to

climb. Sarah’s MLU Group 2 contains only five incorrect contexts for functional

prepositions, giving her a rate of 0% correct usage. This increases to 47% (17 contexts)

in MLU Group 3, and jumps to 95% (98 contexts) in MLU Group 4, followed by 79%

(106 contexts) and 89% (179 contexts) rates of correct use in MLU Groups 5 and 6. It

doesn’t appear that the 95% accuracy rate found in MLU Group 4 truly establishes the

start of Sarah’s full acquisition, as she never again reaches that rate of use in the

remaining stages.

In fact, the sudden increase in Sarah’s functional preposition use at this stage is

due to one particular file (File 53), where Sarah and her mother are looking at a catalog

together, and Sarah points at every item, saying phrases like “get one of those” or “I

want one of those”. These two phrases alone occur a total of 33 and 10 times,

respectively; Sarah’s use of of is so high in this file (51 instances) that is accounts for

206

15% of all of her functional prepositions. As seen in Table (7.6), if this file is taken out

to adjust for the unusually high number of uses, her patterns of development appear

more level, and don’t show the drastic increase, followed by an immediate fall to 79% in

the next file. However, in both MLU Groups 4 and 6, Sarah is very close to reaching

90% accuracy.

Group 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+

SarahAdverb * 96% 100% 100% 100% 99% *Particle * 100% 100% 100% 100% 100% *

Semi-Lexical Prep * 68% 85% 96% 97% 98% *Functional Prep * 0% 47% 89% 79% 89% *

* No data availableTable (7.6): Sarah’s adjusted data

As with the other children, Sarah’s data clearly show that semi-lexical prepositions are

acquired after adverbs and particles, and they also seem to be acquired before functional

prepositions. Again, no conclusion regarding the order of acquisition of adverbs and

particles can be drawn.

3.6 Summary of acquisition patterns

While slight variations occur from child to child with regard to the exact timing of

when semi-lexical and functional prepositions are acquired, all show identical patterns

with regard to linear ordering. In all five children, the adverbs and particles are used rates

well above 90% from their earliest stages, indicating that these elements are acquired

very early on. After the adverbs and particles, the level of 90% correct usage is attained

for semi-lexical prepositions. The functional prepositions are acquired after the semi-

lexical prepositions. Three children (Adam, Naomi, and Nina) clearly reach the 90%

207

threshold, and Eve reaches the 90% threshold in her last four files in MLU Group 5 (her

last available group). Sarah, except for the exceptional ‘catalogue shopping’ file, never

seems to reach the 90% threshold, but remains on the cusp. These patterns are laid out

schematically in Table (7.7), where the MLU Group of acquisition is listed.

Importantly, these data support the conclusion that semi-lexical and functional

prepositions represent different types of prepositions, and that there is an acquisitional

difference between adverbs/particles, semi-lexical prepositions and functional

prepositions. The probability1 that all five children would follow the same acquisitional

pattern with three elements is p = .00463. However, because of the similarity between

the acquisitional patterns of adverbs and particles, no conclusion can be made from this

data about these two elements relative to one another. This will be explored further in

the next section.

Adverbs/ Semi-Lexical FunctionalParticles Prepositions Prepositions

Adam 3 5 6

Eve 2 5 (last 4 files of 5)

Naomi 1 4 5

Nina 2 3 4

Sarah 2 4 (after 6)

Table (7.7): Acquisitional ordering of prepositional elements, child data

3.7 Geraldine, Melissa, and Naomi

So far, no differences seem to emerge in the patterns of acquisition of adverbs

and particles. Part of this problem is related to the fact that all but Naomi’s data begin

1 The probability is based on the fact that there are 3! possible orders that can occur. If each of these 6orders is equally likely, then the probability of five kids following the same order is 1/6 to the thirdpower.

208

at MLU Group 2 or later. Additionally, Naomi’s data in the previous discussion of

acquisition rates seemed to indicate that she uses adverbs and particles in identical ways.

A closer examination of her earliest MLU Group shows that there is, in fact, evidence

for a difference between adverbs and particles. In her first MLU Group, there is one

single particle that is used correctly, giving her 100% accuracy in her use of particles.

This MLU Group contains seven files spanning ages 1;2,29 to 1;9,10. What is not clear

from the charts and discussion above is that this single particle occurs in the last file of

this MLU Group. By contrast, the 49 adverbs that are used are distributed among the

different files. In short, it seems that this one particle is very close to emerging in MLU

Group 2.

Because none of the other children selected for the study have data that were

collected this early on, confirmation of this pattern cannot be found in this dataset.

Instead, we turn to two other children, whose data begin during the MLU Group 1 and

continue into MLU Group 2: Melissa and Geraldine. As can be seen in Table (7.8),

these two children mirror Naomi’s pattern, as described previously. Note that this table

shows all contexts that occurred in the three children’s data during MLU Groups 1 and

2; as no errors occurred in this MLU Group, all are instances of correct usage.

209

Naomi GeraldineGeraldineGeraldine MelissaMLU Adverb Particle MLU Adverb Particle MLU Adverb Particle1.83 0 0 1.21 7 0 1.20 1 01.31 27 0 * 1.29 1 01.14 2 0 * 1.22 1 01.39 10 0 * 1.25 0 01.47 7 0 * 1.39 2 01.26 3 0 * 1.43 2 01.40 0 1 * *

Grp 1 49 1 Grp 1 7 0 Grp 1 7 01.62 1 0 1.70 14 0 1.67 13 41.58 19 3 1.78 9 0 1.84 8 01.79 10 0 1.72 9 0 1.71 10 31.54 8 0 1.52 7 0 *1.67 3 0 2.88 6 1 *1.72 1 0 1.96 0 3 *

Grp 2 42 3 Grp 2 45 4 Grp 2 31 7* No data available

Table (7.8): Distribution of early adverb and particle uses

Importantly, all three children show a delay in the production of particles. Adverbs are

produced for a period of time, and then the particles start to be produced. When

particles do appear, they are used perfectly with rates of 100% accuracy.

Of course, one cannot rely on spontaneous production data that has been

collected periodically, as the first use may not be the actual first use (especially with a

construction like particles that have a low rate of use in general speech), but could be

representative of the earliest production of the construction. While this evidence is not

conclusive, it does indicate that a difference between adverbs and particles exists.

However, there is clearly a need for additional supportive evidence before this

distinction can be conclusively accepted.

210

4 Additional evidence for differences in the prepositional domain

There are two further pieces of evidence that can be drawn from this child data

that support the view that the prepositional domain can be decomposed into four

component parts. First, there are differences in errors, and second, there are differences

in the rates of frequency that remain constant across speakers. Neither of these areas

provide evidence that is as strong as the comparison of correct to incorrect uses in

obligatory contexts, but both contribute a better view of what is occurring.

4.1 Errors

The vast majority of errors were omissions (83%), followed by substitutions2

(13%) and other miscellaneous errors (4%); Tables (7.2) and (7.9) provide the specific

patterns for each child. Additionally, no patterns emerge with regard to prepositional

element: omissions occurred more frequently in every prepositional type than any other

kind of error, except in Nina’s adverbs, where she had 24 miscellaneous errors, in

contrast with only 7 errors of omission and 5 substitutions, and Adam’s particles, where

3 substitution errors contrast with one omission.

2 Substitutions were further examined to see whether any patterns could be established in the types ofelements that were used as the substituting elements; no clear patterns were found.

211

Element type Adverb Particle Semi-lexical Functional TotalAdam

Omissions 1.9% 0.1% 66.4% 20.6% 89.1%Substitutions 0.7% 0.4% 7.0% 0.6% 8.7%

Other 0.9% 0.0% 0.7% 0.6% 2.2%Total 3.6% 0.6% 74.1% 21.7% 100.0%

EveOmissions 0.8% 0.8% 75.7% 12.7% 90.0%

Substitutions 0.8% 0.0% 6.8% 2.0% 9.6%Other 0.0% 0.0% 0.0% 0.4% 0.4%Total 1.6% 0.8% 82.5% 15.1% 100.0%

NaomiOmissions 3.5% 2.4% 60.0% 16.5% 82.4%

Substitutions 1.2% 0.0% 7.1% 0.0% 8.2%Other 1.2% 2.4% 4.7% 1.2% 9.4%Total 5.9% 4.7% 71.8% 17.6% 100.0%

NinaOmissions 1.9% 0.3% 58.5% 6.5% 67.1%

Substitutions 1.3% 0.5% 18.9% 2.4% 23.2%Other 6.5% 0.0% 2.4% 0.8% 9.7%Total 9.7% 0.8% 79.8% 9.7% 100.0%

SarahOmissions 2.8% 0.0% 55.5% 27.1% 85.3%

Substitutions 1.4% 0.0% 11.5% 0.5% 13.3%Other 0.0% 0.0% 1.4% 0.0% 1.4%Total 4.1% 0.0% 68.3% 27.5% 100.0%

Table (7.9): Error types as a percent of each child’s total errors

Additionally, all children made more errors with semi-lexical prepositions than

any other prepositional element: the total of 1214 semi-lexical errors represents 76% of

all errors. As the overall rate of use of semi-lexical prepositions far outweighs any other

prepositional element (they account for 61% of all correct contexts and 57% of all

obligatory prepositional contexts), this is to be expected. Similarly, particles, which are

the least used element across all five children (accounting for 4% of all correct contexts

and 3% of all total contexts), also account for the lowest proportion of errors (a total of

13 particle errors were made, representing 0.8% of all errors). Interestingly, adverbs,

212

which represent 29% of all prepositional contexts (7085 contexts total), have far fewer

errors than functional prepositions, which represent only 6% of the data (1505

contexts). With 78 errors, adverbs represent 5% of all errors made, while the 295 errors

made with functional prepositions accounts for 18% of all errors. This would seem to

indicate that for some reason functional prepositions are disproportionately

problematic, and supports the view that purely functional elements would be the most

problematic for children to acquire. The high semi-lexical rate of error and corresponding

low particle one cannot be taken as indicators of their status in the prepositional domain,

as these error rates may simply be projections of their overall rates of usage. With

functional prepositions, this is clearly not the case, as they are used much less often

than adverbs, but errors occur in these functional contexts far more frequently than in

the adverb contexts.

It must be noted that adverbs and particles may have such generally low error

rates because obligatory contexts where the adverb or particle is omitted can’t always be

identified. If, for instance, a child produces the utterance ‘I wanna go’ and means to say

‘I wanna go out’, the researcher has no way of knowing that a prepositional context is

occurring. This is very problematic with adverbs, which are often purely

optional—unless something in the context can provide clear cues to meaning, the context

must be taken as a non-prepositional one. A similar scenario occurs with particles;

while not optional, if they are missing from a context, the verb usually makes sense and

creates a grammatical phrase without the particle (consider eat up the jello/eat the jello);

again, without clear, contextual cues, the missing particle cannot be detected. Semi-

213

lexical and functional prepositions, because of their Case-assigning properties act to

relate elements in a sentence; omissions of these elements are usually clearly identifiable

on the basis of syntax alone. Context, however, usually plays a role when substitutions

are being identified. That is not to say that errors in adverb and particle contexts aren’t

coded, but it is to recognize the limitations of research based on spontaneous

production. Other methods, such as elicitation tasks would provide clearer evidence of

errors, because the target utterances can be more closely (although not completely)

controlled for.

Lastly, and importantly, the rates of errors as a percent of total prepositional

contexts support the view that there are acquisitional differences between the four

prepositional elements. Functional prepositions are the most problematic elements for

children, with a rate of 19.6% (295 total errors in 1505 contexts containing functional

prepositions). Semi-lexical prepositions are the second most problematic elements, with

a rate of 8.3% (1214 errors in 14653 semi-lexical contexts). Particles have a rate of 1.5%

(13 errors in 853 particle contexts), and adverbs have an even lower rate of 1.1% (79

errors in 7085 adverb contexts). These figures show that the prepositional elements can

be ordered according to how problematic they are for children to use: adverbs are the

least problematic, followed by particles, semi-lexical prepositions, and finally functional

prepositions.

4.2 Frequency of use

One interesting characteristic of the adult data was that all the adults converged

on a single pattern of distribution for all four prepositional elements; not only were the

214

levels of use ordered similarly in all adult data, but all adults showed a nearly identical

rate of distribution. As seen in Table (7.10), semi-lexical prepositions were used the

most, followed by adverbs, functional prepositions, and particles. And all of the adults

use very similar frequencies for each of the four categories.

Adam Eve Naomi Nina Sarah Mean St. Dev.Adverb 22% 22% 26% 17% 24% 22% 3%Particle 3% 2% 7% 2% 6% 4% 2%

Semi-Lexical Prep 63% 66% 58% 66% 59% 62% 4%Functional Prep 12% 10% 10% 14% 11% 11% 2%

Total 100% 100% 100% 100% 100%

Table (7.10): Frequency of use, adult data

The children also show a similar pattern, but not until the later MLU Groups

(Table (7.11)). In all children the proportion of functional prepositions to all

prepositional elements increases over time; the children begin their first available MLU

Groups with no uses, or with a very small percentage. As this increases proportionally

over time, the adverbs decrease. Semi-lexical prepositions also increase over time, but

particles show a lot of variation, and no specific pattern can be identified. These

patterns can be seen graphically in Figure (7.2).

215

Group 1 2 3 4 5 6 7MLU 1.0-1.49 1.5-1.99 2.0-2.49 2.5-2.99 3.0-3.49 3.5-3.99 4.0+ Mean St. Dev.

AdamAdverb * * 57% 40% 35% 27% 32% 40% 12%Particle * * 3% 5% 6% 4% 6% 4% 1%

Semi-Lexical Prep * * 39% 52% 53% 62% 55% 52% 8%Functional Prep * * 0% 3% 6% 7% 7% 4% 3%

Total 100% 100% 100% 100% 100% 100%

EveAdverb * 67% 48% 39% 24% * * 45% 18%Particle * 1% 2% 2% 5% * * 2% 2%

Semi-Lexical Prep * 31% 42% 55% 64% * * 48% 14%Functional Prep * 1% 8% 4% 7% * * 5% 3%

Total 100% 100% 100% 100% 100%

NaomiAdverb 98% 72% 55% 53% 35% 28% 28% 57% 26%Particle 2% 5% 6% 4% 5% 5% 4% 5% 1%

Semi-Lexical Prep 0% 22% 40% 43% 55% 58% 60% 36% 22%Functional Prep 0% 0% 0% 0% 5% 9% 7% 2% 4%

Total 100% 100% 100% 100% 100% 100% 100% 100%

NinaAdverb * 33% 40% 26% 17% 17% 23% 27% 9%Particle * 2% 0% 1% 2% 4% 2% 2% 1%

Semi-Lexical Prep * 62% 57% 68% 73% 71% 68% 66% 6%Functional Prep * 3% 2% 6% 8% 8% 8% 5% 3%

Total 100% 100% 100% 100% 100% 100% 100%

SarahAdverb * 37% 44% 23% 32% 25% * 32% 9%Particle * 7% 4% 3% 5% 4% * 4% 2%

Semi-Lexical Prep * 56% 51% 61% 57% 64% * 58% 5%Functional Prep * 0% 2% 13% 6% 8% * 6% 5%

Total 100% 100% 100% 100% 100% 100%* No data available

Table (7.11): Proportion of prepositional elements in each MLU Group, child data

216

100%90%80%70%60%50%40%30%20%10%0%

Adverbs Particles Semi-lex preps Funct preps

1

2

3

4

5

6

7

Figure (7.2): Average proportion of use across MLU Group, child data

Overall, it is clear that the relative distribution of adverbs, particles, semi-lexical

and functional prepositions is the same for children and adults, in that they all use more

semi-lexical prepositions than any other prepositional element. This is followed by

adverbs, and then functional prepositions, with particles accounting for the smallest

rates of use. While the children do not show the same rates of use as the adults in all

MLU Groups (their initial stages, of course, contain fewer instances that the adults),

their rates of usage mirror the adult patterns in their last available MLU Group.

5 Conclusion

In examining the data of Adam, Eve, Naomi, Nina, and Sarah, the predictions

made in Chapter 5 have been supported. It was hypothesized that if the lexical and

functional qualities of different prepositional types were accurately captured, then

differences in the acquisition of these elements should be identified. More explicitly, it

was predicted that the purely lexical elements, those lacking functional content but

having lexical content (the [+Lexical, –Functional] category) should be acquired first, and

217

the purely functional ones, those having functional abilities, but lacking lexical content

(the [–Lexical, +Functional] category) should be acquired last. This prediction was

borne out, as adverbs were acquired first, and functional prepositions were acquired last.

Additional evidence for differences between prepositional elements was drawn from

error patterns, where more errors were made with functional prepositions than adverbs.

Differences in the rate of frequency of each prepositional category were also identified.

It was found that functional features outranked lexical features in establishing the

order of acquisition of all four prepositional categories. That adverbs would be acquired

before functional prepositions was predicted; the ordering of particles and semi-lexical

prepositions was left open. It was found that particles were acquired before semi-lexical

prepositions, leading to the overall acquisitional ordering found in (4).

(4) Observed order of acquisition:–Functional +Lexical Adverb–Functional –Lexical Particle+Functional +Lexical Semi-lexical preposition+Functional –Lexical Functional preposition

It can be concluded that the application of the functional features rank higher than the

lexical ones in establishing the acquisitional ordering of prepositional elements.

Importantly, the children’s patterns of acquisition cannot be attributed to input

frequency or simplification. Because adults produced prepositional elements with the

same level of frequency of use over time, it cannot be claimed that the children’s lack of

functional prepositions early on is related to lack of exposure. Additionally, it cannot be

concluded that the elements that the children hear most frequently (the semi-lexical

prepositions) will be acquired faster than the other elements (semi-lexical prepositions

218

are acquired after adverbs and particles. The conclusion must be that there is something

about the linguistic structures themselves that must be responsible for the patterns

found in the data.

Lastly, these data seem to support a constructionist approach, rather than a

maturational approach to language acquisition. Under the Maturation Hypothesis

(Chomsky 1988), linguistic principles are seen a biologically constrained, and thus the

certain linguistic structures become available at certain, preset periods. This predicts

that the child should show a “swift and error-free onset” of the targeted structure

(Braunwald 1985: 105). In contrast, the Continuity Hypothesis emphasizes the

progression of one stage to the next, and the each linguistic structure would be built

gradually over time. In the prepositional domain, the children seem to be gradual

building toward use of prepositional elements: their rates of correct usage grow gradually

over time, until the child has fully mastered the element. This is especially true with the

semi-lexical and functional prepositions, which both demonstrate periods of numerous

errors before correct usage is attained. Additionally, as the timing of acquisition varies

for each of the children (the elements are not acquired at the same chronological age nor

at the same linguistic stage, as indicated by MLU), a preset period of maturation seems

unlikely.