International Journal of Language Studies (IJLS), Vol. 6(4), 2012 (pp. 25-40) 25 Ellen Thompson, Florida International University, USA
Javier Collado-Isasi, Florida International University, USA Maria Omana, Florida International University, USA Amanda Yousuf, Florida International University, USA
Bjorkman (2010) claims that symmetric 'and' involves coordination of CPs, while asymmetric ‘and’ involves TPs. If the processor is guided by structural considerations, we predict a possible two-way split in the processing costs of these structures. We examine the processing time involved in sentences interpreted as: (i) temporal, (ii) causal, (iii) logical, versus the distinctions of (i) asymmetric (TP structure), (ii) symmetric (CP structure). Symmetric 'and' structures involve longer processing times than those of asymmetric, causal 'and', and although the processing times of structures with logical 'and' are longer than with temporal 'and', this distinction does not approach statistical significance.
Keywords: Coordination; Conjunction; Production Times; Clausal 'and'; RSVP
1. Introduction
A topic of current research under lively discussion in the psycholinguistics literature is the relative role that structure and meaning play in the on-line processing of sentences. In this study, we contribute to the this debate by examining the processing of syntactically and semantically distinct interpretations of clausal ‘and’.
In particular, we investigate the predictions for processing of a syntactically articulated theory of the distinction among different interpretations of clausal 'and'. Bjorkman (2010) claims that the logical interpretations of sentential
‘and’ are structurally distinct from temporal and causal interpretations;
according to her analysis, symmetric interpretations involve coordination of CPs, while asymmetric interpretations involve conjunction of a smaller structure: TPs. Previous analyses of these constructions have assumed that, although semantically distinct, these interpretations are structurally identical.
According to Bjorkman’s analysis, we expect to find a possible two-way split in the processing costs of these structures, given that the processor is sensitive to abstract structural distinctions. No such distinction should be evident if the traditional approach to these constructions is correct.
Therefore, this research examines: (1) the processing complexity of sentences interpreted as semantically distinct: (i) temporal, (ii) causal, and (iii) logical, versus (2) the processing complexity of sentences interpreted as syntactically distinct: (i) asymmetric (TP structure), and (ii) symmetric (CP structure). We investigate this issue by conducting an experiment involving Response Serial Visual Presentation (RSVP) (Rayner & Sereno 1994; Garrod 2006). In this methodology, readers see sequences of words in the center of a computer screen presented at a fixed rate. The subject then carries out an additional task, such as identifying a word in the sequence, or answering a yes/no question about the experimental sentence. In the current experiment, the subject is asked to read aloud the sentence that they have read.
We find partial confirmation for the claim that abstract syntactic properties guide on-line processing of sentences. Sentences with symmetric 'and' involve longer processing times than those of asymmetric, causal 'and', and although the processing times of structures with logical 'and' are longer than those with temporal 'and', this distinction does not approach statistical significance.
We thus conclude that the predictions of a structurally-guided parser are confirmed with respect to symmetric versus asymmetric causal interpretations of ‘and’, and that further investigation is required into the processing of symmetric versus asymmetric temporal interpretations of ‘and’.
2. Background
2.1. Distinct Interpretations of Clausal Conjunction
In the literature, the semantic representation of conjunction has been argued to correspond to set union—for example, (1a) is represented with the semantics of (1b).
(1) a. Martha and Frances
b. ‘the set containing Martha and Frances’
Although this analysis seems intuitive for conjunction of nominals, it does not straightforwardly carry over to the semantics of sentential conjunction; (2a) does not refer to (2b).
(1) a. Paula colored her hair and Ruth looked through the newspaper.
b. #‘the set containing Paula colored her hair and Ruth looked through the newspaper’
Rather, in the semantics literature, sentential conjunction is standardly represented by the logical connective ∧, an operation on the logical values of two propositions that produces a value of true if both of its operands are true.
The truth table for sentential conjunction is thus as follows:
Table 1.
Truth Table for Logical Connective ⋀
p q p ∧ q
T T T
T F F
F T F
F F F
There are multiple interpretations of structures with clausal ‘and’, and there are competing semantic analyses of these interpretations. For example, Culicover (1970) and Culicover and Jackendoff (1997) discuss examples of
“left-subordinating and”, in which a conditional interpretation holds between the two clauses, as in (3a), which has the interpretation in (3b):
(3) a. You lose your travel documents and your visa will be voided.
b. ‘If you lose your travel documents, then your visa will be voided’
We do not examine the conditional interpretation in this project (although see Bjorkman 2010 for related discussion).
This paper focuses on the logical interpretation of 'and' discussed above and illustrated in (4a), the temporal interpretation shown in (4b), and the causal meaning, as in (4c):
(4) a. Water freezes at 0◦C and ethanol freezes at -114◦C.
b. The lights came on and the singer stepped onto the stage.
c. The lights were off and I couldn’t see.
It has been noted that logical 'and' is symmetric; it allows a reversal of the two conjuncts with a maintenance of meaning; (5a) is equivalent to (4a):
(5) a. Ethanol freezes at -114◦C and water freezes at 0◦C.
In contrast, temporal and causal 'and' do not permit reversal of the two conjuncts with the same meaning - compare (4b) and (6a), and (6c) versus (6b)) - these uses are therefore characterized as asymmetric.
(6) a. The singer stepped onto the stage and the lights came on.
b. I couldn’t see and the lights were off.
Literature on the semantics of conjunction has argued that clausal ‘and’
consistently has a logical semantics, and it is pragmatic and discourse factors which derive the temporal and causal interpretations (Grice, 1975; Posner, 1980; Schmerling, 1975).
In favor of this claim is the observation that without overt ‘and’, two sentences in sequence are interpreted as they are with ‘and’; the sentence in (7a) is interpreted as ordered events (temporal interpretation), while (7b)
results in a causal interpretation between the event of the first sentence and the event of the second sentence.
(7) a. The lights came on; the singer stepped onto the stage.
b. The lights were off; I couldn’t see.
Given that pragmatic inference derives the semantic relation between the events in (7a) and (7b), this pragmatic inference can also derive the semantic relation in cases with overt ‘and’. However, Bjorkman notes that sequences with overt ‘and’ are more restricted semantically than those with no overt
‘and’. The order of independent sentences may be reversed without a change in meaning, as in (8a-8b); however, as noted above, the temporal and causal intepretations of causal ‘and’ resist reversal (Bar-Lev & Palacas 1980).
(8) a. The singer stepped onto the stage; the lights came on.
b. I couldn’t see; the lights were off.
Bjorkman (2010), in contrast, argues that the difference between symmetric and asymmetric 'and' is semantic. Furthermore, she claims that this semantic distinction is reflected in the syntax of the conjunction structures: symmetric coordination involves conjunction of CP structures, whereas asymmetric coordination involves conjunction of TP structures.
2.2. Syntax of Coordination
There are several distinct analyses of the syntax of coordination. Gazdar et al 1985 propose the Ternary Branching structure in (9). However, as noted in contemporary research on coordination, this structure is incompatible with the widely-adopted claim that all syntactic structures are binary-branching.
(9) Ternary Branching (Gazdar et al., 1985)
The claim that conjunction involves a unique syntactic phrase headed by the Boolean operator is argued for in Johannessen, 1993, 1998; Kayne, 1994.
According to these analyses, the first conjunct is located in the Specifier of &P, and the second conjunct in the Complement of &.
(10) Boolean Phrase (Johannessen, 1993, 1998; Kayne, 1994) XP
XPw|p and XP
Yet another proposal is that coordination involves adjunction onto a Boolean Phrase, as in (11).
(11) Adjunction Structure (Munn, 1993)
Alternative Analyses include the claim that conjunction involves a phrase determined categorically by its complement, as in (12) (see Zhang 2009 for discussion):
(12) Complement-as-Head Phrase (Zhang, 2009)
What is common to these syntactic analyses of conjunction is that the syntactic category of the first conjunct is identical to that of the second conjunct. Recall that Bjorkman claims that the syntax of symmetric conjunction involves a CP structure, while that of asymmetric conjunction involves a TP structure.
XP
YP3 X’3
X XP
XP
XPWO BP
Bei YP
&P
XP3 &’3
& YP
2.3. Evidence for Structural Distinction
Bjorkman argues that the syntactic distribution of symmetric and asymmetric coordination provides evidence in favor of her structural analysis. She notes that there is a semantic contrast between the complementation structures in (13a) and (13b).
(13) a. The newspaper reported that [a new mayor was elected] and [there was a riot]
b. The newspaper reported [that a new mayor was elected] and [that there was a riot]
Given that there is no over complementizer in the second conjunct of (13a), this involves coordination of TPs, and favors a temporal or causal interpretation of the sentence – first, a new mayor was elected and then there was a riot, or a new major was elected and because of this there was a riot.
The sentence in (13b), with an overt complementizer in the second conjunct, requires a CP coordination analysis, favors, on the other hand, a logical interpretation.
3. METHOD 3.1. Processing Consequences
Given Bjorkman’s claim that the logical interpretation of causal ‘and’ requires CP coordination, whereas the temporal and causal interpretations involve TP coordination, it is interesting to examine the predictions this theory makes for processing cost associated with these structures.
If we make the uncontroversial assumption that empty syntactic structure has processing cost (Frazier & Clifton 2002), then it is predicted that logical interpretations of clausal ‘and’ require more processing cost than temporal and causal intepretations. We set out to test this claim by conducting a Rapid Serial Visual Presentation experiment.
3.2. Experimental Methodology
The methodology employed in this experiment is Response Serial Visual Presentation (RSVP), presented using a PowerPoint Presentation and Wave Pad Sound Editor recording software. (Rayner & Sereno 1994; Garrod 2006).
In the RSVP methodology, readers see sequences of words in the center of a computer screen presented at a fixed rate. The subject then carries out an additional task, such as identifying a word in the sequence, or answering a yes/no question about the experimental sentence. In the current experiment, the subject is asked to read aloud the sentence that they have read.
The experiment runs as follows: Subjects are seated in front of the computer.
A cross appears in the center of the screen and the subject presses the space bar to start to see words. Sentences are displayed a single word at a time, and each word is onscreen for 800 ms.
Participants silently repeat each word to themselves, without labial movement. At the end of each sentence, participants are prompted by a marker “X” and a tone. This indicates the completion of the sentence and prompts the subject to reproduce the target sentence.
The production of each structure is recorded and processed using WavePad Sound Editor by measuring the distance between the beginning of the sounded prompt and the completion of the participants’ utterance.
The subject is initially seated at the computer and is given a pre-trial training of 10 sentences. The sixty experimental sentences are randomly distributed into three separate PowerPoint layouts and randomly selected for each experiment. (The complete list of experimental sentences is provided in Appendix A.)
3.3. Experimental Items
Examples of the experimental items used are provided in (14):
(14) a. Logical and Gabriel ordered the pasta and Lily had some chicken.
Sarah studies in the library and Connie works from home.
b. Temporal and The man fell and the woman laughed.
She won the lottery and they bought a yacht.
c. Causal and A hurricane hit and the schools closed.
The room was dark and I tripped.
Subjects are presented with sixty sentences in total, thirty experimental and thirty fillers. Of the thirty experimental sentences, ten are logical, ten are temporal, and ten are causal.
Examples of filler sentences are provided in (15):
(15) a. We went to the zoo yesterday with my nephew.
b. Sarah walked down a dark alleyway to get to school.
3.4. Subjects
There were eight experimental subjects, four females and four males, all of whom are monolingual, native speakers of American English. Ages range from nineteen to fifty-three, with a mean age of 33.87. Level of education ranges from High School to Associate’s Degree.
4. Results
The mean production times by subject of each structure is presented in
(Table 2). Recall that this is the length of time determined by WavePad Sound Editor by measuring the distance between the beginning of the sounded prompt and the completion of the participants’ utterance (see Appendix B for a complete list of subjects’ production times.)
Table 2
Mean Production Times
The research question introduced at the beginning of this paper was whether symmetric structures (logical interpretations) would involve significantly longer processing times than asymmetric structures (temporal and causal interpretations.) Examine Figure 1.
Figure 1. Boxplot of means of length of production of causal, temporal, and logical clausal structures.
Statistical Analysis was completed using Standard Procurement System (SPSS) Version 15. A one-way repeated measures ANOVA was carried out on the production times for the three coordinate structures. Post hoc tests were carried out using Fisher’s LSD pair wise comparison at the 5% significance level.
In order to determine that the length of sentences of each type do not play a role in determining production speed, we calculate the mean of characters for each sentence type, and from this the mean production rate per second for each sentence type.
Table 3
Mean Number of Characters Per Sentence
Logical Temporal Causal
37.1 36.5 33.7
Table 4 shows the mean characters per second produced for each construction type:
Table 4
Mean Characters Per Second
Logical Temporal Causal
8.770
characters/second
9.455
characters/second
9.683
characters/second Thus the analysis of speed of production supports the ranking determined in the Boxplot in Figure 1 above.
5. Discussion
The first result of note is that Symmetric Structures (M = 4.23) are produced significantly longer than Asymmetric Structures (M = 3.67). This is predicted by Bjorkman’s approach, because the Asymmetric Structure involves greater syntactic complexity, and hence greater processing cost.
In contrast, it is not the case that the three-way comparison of processing times bears out straightforwardly the predictions of Bjorkman’s analysis. The Logical Structure requires significantly longer time to process (M=4.23) than the Causal Structure (M = 3.48). Note, however that although the Logical Structure does take longer to process (M = 4.23) than the Temporal Interpretation (M = 3.86), as predicted on Bjorkman’s analysis, this difference does not achieve statistical significance. In addition, it is not predicted by Bjorkman’s analysis that the difference between processing time of the two types of Asymmetric Structures, Temporal and Clausal, was significant.
6. Conclusions
To conclude, this study examined the production times of three semantically contrasting structures involving sentential conjunction: Logical, Causal, and Temporal. It is claimed that it is predicted by Bjorkman’s analysis of this semantic contrast according to which the Logical interpretation involves greater semantic and syntactic complexity than the Temporal and Causal interpretations, that Symmetric structures involve significantly longer processing times than Asymmetric structures.
The data argue against a pragmatic analysis of the contrast in interpretations according to which Logical Conjunction is semantically primitive and Causal and Temporal interpretations are derived from this structure by pragmatic inference. This approach would predict that the least costly interpretation processing-wise would be the Logical interpretation, and that the Temporal and Causal interpretations would be more costly. There is no aspect of this analysis which is supported by the data collected here.
Although the differential analysis of Symmetric and Asymmetric Structures is supported by the data, it is not predicted by this approach that: (1) length of processing time of Logical interpretations is not significantly longer than length of time of Temporal interpretations, and (2) length of processing time of Temporal interpretations is significantly longer than that of Causal interpretations.
This work has shown that investigation of the production times of semantically distinct interpretations of clausal ‘and’ provide evidence in favor of the claim that this distinction is semantic and syntactic, and not only pragmatic.
Acknowledgements
We would like to extend our thanks to the English Department at Florida International University and to the Graduate School and the College of Arts and Sciences for their financial support of this research.
The Authors
Ellen Thompson is an Associate Professor of Linguistics in the English Department at Florida International University in Miami, Florida, U.S.A. She received her B.A. in English from the University of Massachusetts in 1990 and her Ph.D. in Linguistics from the University of Maryland in 1996. Among her publications are “Time in Natural Language: Syntactic Interfaces with Semantics and Discourse”, 2005, Mouton de Gruyter; “The Position of the
Subject in Spoken Saudi Arabic: A Processing Perspective”, Arizona Linguistics Working Papers, 2011, University of Arizona, Tuscon, Arizona, with Sawsan Werfelli; “Full and Reduced Relative Clauses in Verb Particle Constructions”, Proceedings of the Psycholinguistics in Flanders (PIF) Conference, 2011, with David Lopez, Eunice Louis, Jonathan T. Miles, Kezia Walker; "The Structure of Bounded Events", Linguistic Inquiry, Spring 2006, Vol. 37:2, 211-228.
Javier Collado-Isasi is a graduate student in the English Department at Florida International University.
Maria Omana holds a B.A. in English from Florida International University.
Amanda Yousuf holds an M.A. in Linguistics from Florida International University.
References
Bar-Lev, Z., & Palacas, A. (1980). Semantic command over pragmatic priority.
Lingua, 51,137 –146.
Bjokman, B. (2010). A syntactic correlate of a semantic asymmetry, paper presented at NELS 41.
Blakemore, D., & Carston, R. (1999). The Pragmatics of and-conjunctions: The Non-narrative cases. UCL Working Papers in Linguistics, 11,1–20.
Blakemore, D., & Carston, R. (2005). The pragmatics of sentential coordination with and. Lingua, 115, 569–589.
Culicover, P. W. (1970). One more can of beer. Linguistic Inquiry 1, 366–369.
Culicover, P. W., & Jackendoff, R. (1997). Semantic subordination despite syntactic coordination. Linguistic Inquiry 28, 195–217.
Ferreira, V. S., & Humphreys, K.R. (2001). Syntactic influences on lexical and morphological processing in language production. Journal of Memory and Language 44, 52–80.
Frazier, L., & Rayner, K. (1987). Resolution of syntactic category ambiguities:
eye movements in parsing lexically ambiguous sentences. Journal of Memory and Language 26, 505–526.
Frazier, L., & Clifton, C., Jr. (2002). Parsing coordinates and ellipsis: copy α, Syntax, 4:1, 1–22.
Garrod, S. (2006). Psycholinguistic research methods. In M. Traxler & M.A.
Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 455-503). San Diego: Academic Press.
Gazdar, G., Klein, E., Pullum, G., & Sag, I. (1985). Generalized phrase structure grammar. Cambridge: Harvard University Press.
Goldsmith, J. A. (1985). A principled exception to the coordinate structure constraint. Papers from the Chicago Linguistic Society Meeting, 21, 133–
143.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J.L. Morgan (Eds.) Syntax and semantics 3: Speech acts, (pp. 41–58). Amsterdam: Elsevier.
Kayne, R. (1994). The antisymmetry of syntax. (Linguistic Inquiry monograph 25.) Cambridge, MA: MIT Press.
Lakoff, G. (1986). Frame semantic control of the coordinate structure constraint. Papers from the Chicago Linguistic Society Meeting, 22, 152–
167.
Munn, A. (1993). Topics in the syntax and semantics of coordinate structures.
Doctoral dissertation. College Park, MD: University of Maryland.
Posner, R. (1980). Semantics and pragmatics of sentence connectives in natural language. Speech Act Theory and Pragmatics, 168–203.
Postal, P.M. (1998). Three investigations of extraction. Cambridge, MA: MIT Press.
Potter, M.C. (1984). Rapid serial visual presentation (RSVP): A method for studying language processing. D.E. Kieras & M.A. Just (eds.), New methods in reading comprehension research. Hillsdale, NJ: Erlbaum.
Potter, M.C. & Lombardi, L. (1990). Regeneration in the short-term recall of sentences. Journal of Memory and Language 29, 633–654.
Rayner K. & Sereno S.C. (1994). ‘Eye movements in reading: Psycholinguistic studies.’ In Gernsbacher, M.A. (Ed.), Handbook of Psycholinguistics. San Diego: Academic Press. 57–82.
Ross, J.R. (1967). Constraints on variables in syntax. Doctoral dissertation.
Cambridge, MA: MIT.
Zhang, N. (2009). Coordination in syntax. Cambridge studies in linguistics series 123, Cambridge: Cambridge University Press.
Appendix A – Experimental Sentences Logical and
1. An apple is sweet and a lime tastes sour.
2. Wolves hunt in packs and lions run in prides.
3. Pencils use lead and pens write with ink.
4. The girls like running and the boys enjoy swimming.
5. Jenny wrote an essay and Harold typed a poem.
6. John baked a cake and Mary made a pie.
7. Sarah studies in the library and Connie works from home.
8. I bought a red blouse and she purchased gold shoes.
9. Gabriel ordered the pasta and Lily had some chicken.
10. Bill played golf and Pat went shopping.
Temporal and
1. John cooked dinner and Mary cleaned the dishes.
2. The player scored and the team won the game.
3. The man fell and the woman laughed.
4. The girl mowed the lawn and the boy timed the hedge.
5. She won the lottery and they bought a yacht.
6. The plane landed and the passengers boarded.
7. We spent the day in town and I went to Harrods.
8. The lights came on and the singer stepped onto the stage.
9. The alarm went off and I jumped out of bed.
10. I advertised a sale and someone called me.
Causal and
1. The bomb was lit and it exploded.
2. He swung the hammer and it hit his thumb.
3. John threw the ball and the window broke.
4. The man fell from a cliff and he broke his leg.
5. Mary cheated on her exam and she was grounded.
6. She studied all night and she managed an A on the test.
7. Betty dropped her purse and her wallet fell out.
8. The dam broke and the valley flooded.
9. The sniper shot him and he died 10. A hurricane hit and the schools closed.
Filler Sentences
1. Mary said that Bill will go to the park.
2. I want to paint my nails blue.
3. He watches TV all day every day.
4. He loves school.
5. Some people like to eat fish.
6. I hope that one day I will be a millionaire.
7. The girl kicked the boy in the leg.
8. The picture on the wall is going to fall.
9. The car was broken into.
10. The tornado tore the roof off of the house.
11. We went to the zoo yesterday with my nephew.
12. The boy was expelled because he cheated on a test.
13. The couple got a divorce after 10 years of marriage.
14. I was laid off from work two years ago.
15. The building was twenty stories tall.
16. George Clooney put on an amazing show.
17. John told Bill that Mary ran away to Europe.
18. When I was young, I wanted to be an astronaut.
19. The Discovery space shuttle will no longer go to space.
20. Sarah walked down a dark alleyway to get to school.
21. John laughed so hard that he cried.
22. The man’s beard was so long that it touched the floor.
23. Smoking is bad for your health.
24. Two little mice fell into a bucket of cream.
25. Sam was given a standing ovation.
26. Toasting bread brings out its flavor.
27. The shop closed on Saturday.
28. Hurricanes are difficult to predict.
29. Sofas are great places to take a nap,
30. Tropical fruits are locally available in most supermarkets.
Appendix B
Mean Processing Times for All Subjects of Each Sentence
